Symfony Cookbook 2.7

The Cookbook
Version: 2.7
generated on July 28, 2016
The Cookbook (2.7)

This work is licensed under the Attribution-Share Alike 3.0 Unported license (http://creativecommons.org/
licenses/by-sa/3.0/).
You are free to share (to copy, distribute and transmit the work), and to remix (to adapt the work) under the
following conditions:
Attribution: You must attribute the work in the manner specified by the author or licensor (but not in
any way that suggests that they endorse you or your use of the work).
Share Alike: If you alter, transform, or build upon this work, you may distribute the resulting work only
under the same, similar or a compatible license. For any reuse or distribution, you must make clear to
others the license terms of this work.
The information in this book is distributed on an as is basis, without warranty. Although every precaution
has been taken in the preparation of this work, neither the author(s) nor SensioLabs shall have any liability to
any person or entity with respect to any loss or damage caused or alleged to be caused directly or indirectly by
the information contained in this work.
If you find typos or errors, feel free to report them by creating a ticket on the Symfony ticketing system
(http://github.com/symfony/symfony-docs/issues). Based on tickets and users feedback, this book is
continuously updated.
Contents at a Glance
How to Use Assetic for Asset Management ..........................................................................................7
Combining, Compiling and Minimizing Web Assets with PHP Libraries.............................................13
How to Minify CSS/JS Files (Using UglifyJS and UglifyCSS) ...............................................................16
How to Minify JavaScripts and Stylesheets with YUI Compressor.......................................................20
How to Use Assetic for Image Optimization with Twig Functions ......................................................22
How to Apply an Assetic Filter to a specific File Extension .................................................................24
How to Install 3rd Party Bundles .......................................................................................................26
Best Practices for Reusable Bundles ...................................................................................................29
How to Use Bundle Inheritance to Override Parts of a Bundle ............................................................36
How to Override any Part of a Bundle ...............................................................................................38
How to Remove the AcmeDemoBundle .............................................................................................41
How to Load Service Configuration inside a Bundle ...........................................................................44
How to Create Friendly Configuration for a Bundle ...........................................................................47
How to Simplify Configuration of multiple Bundles ...........................................................................53
How to Use Varnish to Speed up my Website ....................................................................................55
Caching Pages that Contain CSRF Protected Forms ...........................................................................59
Installing Composer ..........................................................................................................................60
How to Master and Create new Environments ...................................................................................61
How to Override Symfony's default Directory Structure .....................................................................66
Using Parameters within a Dependency Injection Class ......................................................................69
Understanding how the Front Controller, Kernel and Environments Work together............................71
How to Set external Parameters in the Service Container ....................................................................74
How to Use the Apache Router .........................................................................................................76
Configuring a Web Server .................................................................................................................79
How to Organize Configuration Files ................................................................................................85
How to Create a Console Command .................................................................................................89
How to Use the Console....................................................................................................................93
How to Style a Console Command ....................................................................................................95
How to Call a Command from a Controller ..................................................................................... 102
How to Generate URLs from the Console ........................................................................................ 104
How to Enable Logging in Console Commands ............................................................................... 106
How to Define Commands as Services ............................................................................................. 110
How to Customize Error Pages ........................................................................................................ 112
How to Define Controllers as Services ............................................................................................. 117
How to Upload Files ....................................................................................................................... 122
How to Optimize your Development Environment for Debugging.................................................... 128
PDF brought to you by
Contents at a Glance | iii
How to Deploy a Symfony Application ............................................................................................ 130

Deploying to Microsoft Azure Website Cloud .................................................................................. 134
Deploying to Heroku Cloud ............................................................................................................ 147
Deploying to Platform.sh................................................................................................................. 152
Deploying to fortrabbit ................................................................................................................... 156
How to use Doctrine Extensions: Timestampable, Sluggable, Translatable, etc. ................................ 160
How to Register Event Listeners and Subscribers ............................................................................. 161
How to Use Doctrine DBAL ............................................................................................................ 164
How to Generate Entities from an Existing Database........................................................................ 166
How to Work with multiple Entity Managers and Connections ........................................................ 169
How to Register custom DQL Functions.......................................................................................... 172
How to Define Relationships with Abstract Classes and Interfaces.................................................... 173
How to Provide Model Classes for several Doctrine Implementations ............................................... 176
How to Implement a Simple Registration Form................................................................................ 179
How to Use PdoSessionHandler to Store Sessions in the Database .................................................... 185
How to Use MongoDbSessionHandler to Store Sessions in a MongoDB Database............................. 189
Console Commands........................................................................................................................ 191
How to Send an Email..................................................................................................................... 192
How to Use Gmail to Send Emails ................................................................................................... 195
How to Use the Cloud to Send Emails ............................................................................................. 197
How to Work with Emails during Development............................................................................... 199
How to Spool Emails....................................................................................................................... 202
How to Test that an Email is Sent in a Functional Test ..................................................................... 204
How to Create Event Listeners and Subscribers................................................................................ 206
How to Set Up Before and After Filters ............................................................................................ 210
How to Extend a Class without Using Inheritance............................................................................ 214
How to Customize a Method Behavior without Using Inheritance .................................................... 217
How to use Expressions in Security, Routing, Services, and Validation ............................................. 219
How to Customize Form Rendering ................................................................................................ 222
How to Use Data Transformers ....................................................................................................... 235
How to Dynamically Modify Forms Using Form Events ................................................................... 243
How to Embed a Collection of Forms .............................................................................................. 254
How to Create a Custom Form Field Type....................................................................................... 266
How to Create a Form Type Extension ............................................................................................ 271
How to Reduce Code Duplication with "inherit_data" ..................................................................... 276
How to Unit Test your Forms.......................................................................................................... 279
How to Configure empty Data for a Form Class............................................................................... 284
How to Use the submit() Function to Handle Form Submissions...................................................... 286
How to Use the virtual Form Field Option....................................................................................... 289
Using Bower with Symfony ............................................................................................................. 290
How to Install or Upgrade to the Latest, Unreleased Symfony Version .............................................. 293
How to Use Monolog to Write Logs ................................................................................................ 295
How to Configure Monolog to Email Errors .................................................................................... 300
How to Configure Monolog to Display Console Messages................................................................ 302
How to Configure Monolog to Exclude 404 Errors from the Log ...................................................... 304
How to Log Messages to different Files ............................................................................................ 305
How to Create a custom Data Collector........................................................................................... 307
iv | Contents at a Glance
Contents at a Glance | 4
How to Use Matchers to Enable the Profiler Conditionally ............................................................... 311

Switching the Profiler Storage .......................................................................................................... 313
How to Access Profiling Data Programmatically............................................................................... 314
The PSR-7 Bridge ............................................................................................................................ 316
How to Configure Symfony to Work behind a Load Balancer or a Reverse Proxy .............................. 318
How to Register a new Request Format and Mime Type................................................................... 321
How to Force Routes to always Use HTTPS or HTTP ...................................................................... 322
How to Allow a "/" Character in a Route Parameter ......................................................................... 323
How to Configure a Redirect without a custom Controller ............................................................... 324
How to Use HTTP Methods beyond GET and POST in Routes ........................................................ 326
How to Use Service Container Parameters in your Routes ................................................................ 328
How to Create a custom Route Loader ............................................................................................ 330
Redirect URLs with a Trailing Slash................................................................................................. 334
How to Pass Extra Information from a Route to a Controller............................................................ 336
Looking up Routes from a Database: Symfony CMF DynamicRouter ............................................... 337
How to Build a Traditional Login Form ........................................................................................... 338
How to Load Security Users from the Database (the Entity Provider)................................................ 343
How to Add "Remember Me" Login Functionality ........................................................................... 351
How to Impersonate a User ............................................................................................................. 355
How to Customize your Form Login................................................................................................ 358
How to Create a custom User Provider ............................................................................................ 361
How to Create a Custom Form Password Authenticator................................................................... 366
How to Authenticate Users with API Keys ....................................................................................... 369
How to Create a custom Authentication Provider............................................................................. 377
Using pre Authenticated Security Firewalls ...................................................................................... 386
How to Change the default Target Path Behavior ............................................................................. 388
Using CSRF Protection in the Login Form........................................................................................ 390
How to Choose the Password Encoder Algorithm Dynamically ........................................................ 392
How to Use multiple User Providers ................................................................................................ 394
How to Restrict Firewalls to a Specific Request ................................................................................ 396
How to Restrict Firewalls to a Specific Host ..................................................................................... 398
How to Use Voters to Check User Permissions................................................................................. 399
How to Use Access Control Lists (ACLs) ......................................................................................... 403
How to Use advanced ACL Concepts .............................................................................................. 407
How to Force HTTPS or HTTP for different URLs........................................................................... 411
How to Secure any Service or Method in your Application ............................................................... 412
How Does the Security access_control Work?.................................................................................. 415
How to Use the Serializer ................................................................................................................ 419
How to Work with Scopes .............................................................................................................. 422
How to Work with Compiler Passes in Bundles ............................................................................... 425
Session Proxy Examples .................................................................................................................. 426
Making the Locale "Sticky" during a User's Session .......................................................................... 428
Configuring the Directory where Session Files are Saved .................................................................. 431
Bridge a legacy Application with Symfony Sessions .......................................................................... 433
Limit Session Metadata Writes ........................................................................................................ 434
Avoid Starting Sessions for Anonymous Users.................................................................................. 435
How Symfony2 Differs from Symfony1 ............................................................................................ 436

Contents at a Glance | v
How to Inject Variables into all Templates (i.e. global Variables) ...................................................... 441
How to Use and Register Namespaced Twig Paths ........................................................................... 443
How to Use PHP instead of Twig for Templates............................................................................... 445
How to Write a custom Twig Extension .......................................................................................... 450
How to Render a Template without a custom Controller.................................................................. 452
How to Simulate HTTP Authentication in a Functional Test ............................................................ 454
How to Simulate Authentication with a Token in a Functional Test.................................................. 455
How to Test the Interaction of several Clients .................................................................................. 457
How to Use the Profiler in a Functional Test.................................................................................... 458
How to Test Code that Interacts with the Database.......................................................................... 460
How to Test Doctrine Repositories .................................................................................................. 463
How to Customize the Bootstrap Process before Running Tests........................................................ 465
Upgrading a Patch Version (e.g. 2.6.0 to 2.6.1) ................................................................................ 467
Upgrading a Minor Version (e.g. 2.5.3 to 2.6.1) ............................................................................... 468
Upgrading a Major Version (e.g. 2.7.0 to 3.0.0)................................................................................ 470
Upgrading a Third-Party Bundle for a Major Symfony Version ......................................................... 474
How to Create a custom Validation Constraint ................................................................................ 478
How to Handle Different Error Levels.............................................................................................. 482
How to Dynamically Configure Validation Groups .......................................................................... 484
How to Use PHP's built-in Web Server ............................................................................................ 486
How to Create a SOAP Web Service in a Symfony Controller ........................................................... 489
How to Create and Store a Symfony Project in Git ........................................................................... 492
How to Create and Store a Symfony Project in Subversion................................................................ 495
Using Symfony with Homestead/Vagrant......................................................................................... 499
vi | Contents at a Glance
Contents at a Glance | 6
Chapter 1
How to Use Assetic for Asset Management

Assetic combines two major ideas: assets and filters. The assets are files such as CSS, JavaScript and image
files. The filters are things that can be applied to these files before they are served to the browser. This
allows a separation between the asset files stored in the application and the files actually presented to the
user.
Without Assetic, you just serve the files that are stored in the application directly:
Listing 1-1
<script src="{{ asset('js/script.js') }}"></script>
But with Assetic, you can manipulate these assets however you want (or load them from anywhere) before
serving them. This means you can:
Minify and combine all of your CSS and JS files
Run all (or just some) of your CSS or JS files through some sort of compiler, such as LESS, SASS or
CoffeeScript
Run image optimizations on your images
Assets
Using Assetic provides many advantages over directly serving the files. The files do not need to be stored
where they are served from and can be drawn from various sources such as from within a bundle.
You can use Assetic to process CSS stylesheets, JavaScript files and images. The philosophy behind
adding either is basically the same, but with a slightly different syntax.
Including JavaScript Files

To include JavaScript files, use the javascripts tag in any template:
Listing 1-2
1
2
3
{% javascripts '@AppBundle/Resources/public/js/*' %}
<script src="{{ asset_url }}"></script>
{% endjavascripts %}

Chapter 1: How to Use Assetic for Asset Management | 7
If your application templates use the default block names from the Symfony Standard Edition, the
javascripts tag will most commonly live in the javascripts block:
Listing 1-3
1
2
3
4
5
6
7
{# ... #}
{% block javascripts %}
{% javascripts '@AppBundle/Resources/public/js/*' %}
{% endblock %}
{# ... #}
You can also include CSS stylesheets: see Including CSS Stylesheets.
In this example, all files in the Resources/public/js/ directory of the AppBundle will be loaded and
served from a different location. The actual rendered tag might simply look like:
Listing 1-4
<script src="/app_dev.php/js/abcd123.js"></script>
This is a key point: once you let Assetic handle your assets, the files are served from a different location.
This will cause problems with CSS files that reference images by their relative path. See Fixing CSS Paths
with the cssrewrite Filter.
Including CSS Stylesheets

To bring in CSS stylesheets, you can use the same technique explained above, except with the
stylesheets tag:
Listing 1-5
1
2
3
{% stylesheets 'bundles/app/css/*' filter='cssrewrite' %}

<link rel="stylesheet" href="{{ asset_url }}" />
{% endstylesheets %}
If your application templates use the default block names from the Symfony Standard Edition, the
stylesheets tag will most commonly live in the stylesheets block:
Listing 1-6
1
2
3
4
5
6
7
{# ... #}
{% block stylesheets %}
{% stylesheets 'bundles/app/css/*' filter='cssrewrite' %}
{% endblock %}
{# ... #}
But because Assetic changes the paths to your assets, this will break any background images (or other
paths) that uses relative paths, unless you use the cssrewrite filter.
Notice that in the original example that included JavaScript files, you referred to the files using
a path like @AppBundle/Resources/public/file.js, but that in this example, you referred
to the CSS files using their actual, publicly-accessible path: bundles/app/css. You can use
either, except that there is a known issue that causes the cssrewrite filter to fail when using the
@AppBundle syntax for CSS stylesheets.

Including Images
To include an image you can use the image tag.
Listing 1-7
1
2
3
{% image '@AppBundle/Resources/public/images/example.jpg' %}
<img src="{{ asset_url }}" alt="Example" />
{% endimage %}
You can also use Assetic for image optimization. More information in How to Use Assetic for Image
Optimization with Twig Functions.
Instead of using Assetic to include images, you may consider using the LiipImagineBundle1
community bundle, which allows to compress and manipulate images (rotate, resize, watermark,
etc.) before serving them.
Fixing CSS Paths with the cssrewrite Filter

Since Assetic generates new URLs for your assets, any relative paths inside your CSS files will break. To
fix this, make sure to use the cssrewrite filter with your stylesheets tag. This parses your CSS files
and corrects the paths internally to reflect the new location.
You can see an example in the previous section.
When using the cssrewrite filter, don't refer to your CSS files using the @AppBundle syntax. See
the note in the above section for details.
Combining Assets
One feature of Assetic is that it will combine many files into one. This helps to reduce the number of
HTTP requests, which is great for front-end performance. It also allows you to maintain the files more
easily by splitting them into manageable parts. This can help with re-usability as you can easily split
project-specific files from those which can be used in other applications, but still serve them as a single
file:
Listing 1-8
1
2
3
4
5
6
{% javascripts
'@AppBundle/Resources/public/js/*'
'@AcmeBarBundle/Resources/public/js/form.js'
'@AcmeBarBundle/Resources/public/js/calendar.js' %}
In the dev environment, each file is still served individually, so that you can debug problems more easily.
However, in the prod environment (or more specifically, when the debug flag is false), this will be
rendered as a single script tag, which contains the contents of all of the JavaScript files.
If you're new to Assetic and try to use your application in the prod environment (by using the
app.php controller), you'll likely see that all of your CSS and JS breaks. Don't worry! This is on
purpose. For details on using Assetic in the prod environment, see Dumping Asset Files.
And combining files doesn't only apply to your files. You can also use Assetic to combine third party
assets, such as jQuery, with your own into a single file:
1. https://github.com/liip/LiipImagineBundle

Listing 1-9
1
2
3
4
5
{% javascripts
'@AppBundle/Resources/public/js/thirdparty/jquery.js'
'@AppBundle/Resources/public/js/*' %}
Using Named Assets

AsseticBundle configuration directives allow you to define named asset sets. You can do so by defining
the input files, filters and output files in your configuration under the assetic section. Read more in
the assetic config reference.
Listing 1-10
1
2
3
4
5
6
7
# app/config/config.yml
assetic:
assets:
jquery_and_ui:
inputs:
- '@AppBundle/Resources/public/js/thirdparty/jquery.js'
- '@AppBundle/Resources/public/js/thirdparty/jquery.ui.js'
After you have defined the named assets, you can reference them in your templates with the
@named_asset notation:
Listing 1-11
1
2
3
4
5
{% javascripts
'@jquery_and_ui'
'@AppBundle/Resources/public/js/*' %}
Filters
Once they're managed by Assetic, you can apply filters to your assets before they are served. This includes
filters that compress the output of your assets for smaller file sizes (and better frontend optimization).
Other filters can compile CoffeeScript files to JavaScript and process SASS into CSS. In fact, Assetic has a
long list of available filters.
Many of the filters do not do the work directly, but use existing third-party libraries to do the heavylifting. This means that you'll often need to install a third-party library to use a filter. The great advantage
of using Assetic to invoke these libraries (as opposed to using them directly) is that instead of having to
run them manually after you work on the files, Assetic will take care of this for you and remove this step
altogether from your development and deployment processes.
To use a filter, you first need to specify it in the Assetic configuration. Adding a filter here doesn't mean
it's being used - it just means that it's available to use (you'll use the filter below).
For example to use the UglifyJS JavaScript minifier the following configuration should be defined:
Listing 1-12
1
2
3
4
5
assetic:
filters:
uglifyjs2:
bin: /usr/local/bin/uglifyjs
Now, to actually use the filter on a group of JavaScript files, add it into your template:
Listing 1-13
1
2
3
{% javascripts '@AppBundle/Resources/public/js/*' filter='uglifyjs2' %}


A more detailed guide about configuring and using Assetic filters as well as details of Assetic's debug
mode can be found in How to Minify CSS/JS Files (Using UglifyJS and UglifyCSS).
Controlling the URL Used

If you wish to, you can control the URLs that Assetic produces. This is done from the template and is
relative to the public document root:
Listing 1-14
1
2
3
{% javascripts '@AppBundle/Resources/public/js/*' output='js/compiled/main.js' %}

Symfony also contains a method for cache busting, where the final URL generated by Assetic
contains a query parameter that can be incremented via configuration on each deployment. For more
information, see the version configuration option.
Dumping Asset Files

In the dev environment, Assetic generates paths to CSS and JavaScript files that don't physically exist on
your computer. But they render nonetheless because an internal Symfony controller opens the files and
serves back the content (after running any filters).
This kind of dynamic serving of processed assets is great because it means that you can immediately see
the new state of any asset files you change. It's also bad, because it can be quite slow. If you're using a lot
of filters, it might be downright frustrating.
Fortunately, Assetic provides a way to dump your assets to real files, instead of being generated
dynamically.
Dumping Asset Files in the prod Environment

In the prod environment, your JS and CSS files are represented by a single tag each. In other words,
instead of seeing each JavaScript file you're including in your source, you'll likely just see something like
this:
Listing 1-15
<script src="/js/abcd123.js"></script>
Moreover, that file does not actually exist, nor is it dynamically rendered by Symfony (as the asset files
are in the dev environment). This is on purpose - letting Symfony generate these files dynamically in a
production environment is just too slow.
Instead, each time you use your application in the prod environment (and therefore, each time you
deploy), you should run the following command:
Listing 1-16
$ php app/console assetic:dump --env=prod --no-debug
This will physically generate and write each file that you need (e.g. /js/abcd123.js). If you update
any of your assets, you'll need to run this again to regenerate the file.

Dumping Asset Files in the dev Environment

By default, each asset path generated in the dev environment is handled dynamically by Symfony. This
has no disadvantage (you can see your changes immediately), except that assets can load noticeably slow.
If you feel like your assets are loading too slowly, follow this guide.
First, tell Symfony to stop trying to process these files dynamically. Make the following change in your
config_dev.yml file:
Listing 1-17
1
2
3
# app/config/config_dev.yml
assetic:
use_controller: false
Next, since Symfony is no longer generating these assets for you, you'll need to dump them manually. To
do so, run the following command:
Listing 1-18
$ php app/console assetic:dump
This physically writes all of the asset files you need for your dev environment. The big disadvantage is
that you need to run this each time you update an asset. Fortunately, by using the assetic:watch
command, assets will be regenerated automatically as they change:
Listing 1-19
$ php app/console assetic:watch
The assetic:watch command was introduced in AsseticBundle 2.4. In prior versions, you had to use
the --watch option of the assetic:dump command for the same behavior.
Since running this command in the dev environment may generate a bunch of files, it's usually a good
idea to point your generated asset files to some isolated directory (e.g. /js/compiled), to keep things
organized:
Listing 1-20
1
2
3
{% javascripts '@AppBundle/Resources/public/js/*' output='js/compiled/main.js' %}


Chapter 2
Combining, Compiling and Minimizing Web

Assets with PHP Libraries
The official Symfony Best Practices recommend to use Assetic to manage web assets, unless you are
comfortable with JavaScript-based front-end tools.
Even if those JavaScript-based solutions are the most suitable ones from a technical point of view, using
pure PHP alternative libraries can be useful in some scenarios:
If you can't install or use npm and the other JavaScript solutions;
If you prefer to limit the amount of different technologies used in your applications;
If you want to simplify application deployment.
In this article, you'll learn how to combine and minimize CSS and JavaScript files and how to compile
Sass files using PHP-only libraries with Assetic.
Installing the Third-Party Compression Libraries

Assetic includes a lot of ready-to-use filters, but it doesn't include their associated libraries. Therefore,
before enabling the filters used in this article, you must install two libraries. Open a command console,
browse to your project directory and execute the following commands:
Listing 2-1
1
2
$ composer require leafo/scssphp

$ composer require patchwork/jsqueeze
Organizing your Web Asset Files

This example will include a setup using the Bootstrap CSS framework, jQuery, FontAwesome and
some regular CSS and JavaScript application files (called main.css and main.js). The recommended
directory structure for this set-up looks like this:
Listing 2-2

Chapter 2: Combining, Compiling and Minimizing Web Assets with PHP Libraries | 13
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
web/assets/
css
main.css
code-highlight.css
js
bootstrap.js
jquery.js
main.js
scss
bootstrap
_alerts.scss
...
_variables.scss
_wells.scss
mixins
_alerts.scss
...
_vendor-prefixes.scss
bootstrap.scss
font-awesome
_animated.scss
...
_variables.scss
font-awesome.scss
Combining and Minimizing CSS Files and Compiling SCSS Files

First, configure a new scssphp Assetic filter:
Listing 2-3
1
2
3
4
5
6
assetic:
filters:
scssphp:
formatter: 'Leafo\ScssPhp\Formatter\Compressed'
# ...
The value of the formatter option is the fully qualified class name of the formatter used by the filter to
produce the compiled CSS file. Using the compressed formatter will minimize the resulting file, regardless
of whether the original files are regular CSS files or SCSS files.
Next, update your Twig template to add the {% stylesheets %} tag defined by Assetic:
Listing 2-4
1
2
3
4
5
6
7
8
9
10
11
12
13
{# app/Resources/views/base.html.twig #}
<!DOCTYPE html>
<html>
<head>

{% stylesheets filter="scssphp" output="css/app.css"
"assets/scss/bootstrap.scss"
"assets/scss/font-awesome.scss"
"assets/css/*.css"
%}
This simple configuration compiles, combines and minifies the SCSS files into a regular CSS file that's put
in web/css/app.css. This is the only CSS file which will be served to your visitors.

Combining and Minimizing JavaScript Files

First, configure a new jsqueeze Assetic filter as follows:
Listing 2-5
1
2
3
4
5
assetic:
filters:
jsqueeze: ~
# ...
Next, update the code of your Twig template to add the {% javascripts %} tag defined by Assetic:
Listing 2-6
1
2
3
4
5
6
7
8
9
10
11
12


{% javascripts filter="?jsqueeze" output="js/app.js"
"assets/js/jquery.js"
"assets/js/bootstrap.js"
"assets/js/main.js"
%}
</body>
</html>
This simple configuration combines all the JavaScript files, minimizes the contents and saves the output
in the web/js/app.js file, which is the one that is served to your visitors.
The leading ? character in the jsqueeze filter name tells Assetic to only apply the filter when not in
debug mode. In practice, this means that you'll see unminified files while developing and minimized files
in the prod environment.

Chapter 3
How to Minify CSS/JS Files (Using UglifyJS and

UglifyCSS)
UglifyJS1 is a JavaScript parser/compressor/beautifier toolkit. It can be used to combine and minify
JavaScript assets so that they require less HTTP requests and make your site load faster. UglifyCSS2 is a
CSS compressor/beautifier that is very similar to UglifyJS.
In this cookbook, the installation, configuration and usage of UglifyJS is shown in detail. UglifyCSS
works pretty much the same way and is only talked about briefly.
Install UglifyJS
UglifyJS is available as a Node.js3 module. First, you need to install Node.js4 and then, decide the
installation method: global or local.
Global Installation
The global installation method makes all your projects use the very same UglifyJS version, which
simplifies its maintenance. Open your command console and execute the following command (you may
need to run it as a root user):
Listing 3-1
$ npm install -g uglify-js
Now you can execute the global uglifyjs command anywhere on your system:
Listing 3-2
$ uglifyjs --help
1. https://github.com/mishoo/UglifyJS
2. https://github.com/fmarcia/UglifyCSS
3. https://nodejs.org/
4. https://nodejs.org/

Chapter 3: How to Minify CSS/JS Files (Using UglifyJS and UglifyCSS) | 16
Local Installation
It's also possible to install UglifyJS inside your project only, which is useful when your project requires
a specific UglifyJS version. To do this, install it without the -g option and specify the path where to put
the module:
Listing 3-3
1
2
$ cd /path/to/your/symfony/project
$ npm install uglify-js --prefix app/Resources
It is recommended that you install UglifyJS in your app/Resources folder and add the node_modules
folder to version control. Alternatively, you can create an npm package.json5 file and specify your
dependencies there.
Now you can execute the uglifyjs command that lives in the node_modules directory:
Listing 3-4
$ "./app/Resources/node_modules/.bin/uglifyjs" --help
Configure the uglifyjs2 Filter

Now we need to configure Symfony to use the uglifyjs2 filter when processing your JavaScripts:
Listing 3-5
1
2
3
4
5
6
assetic:
filters:
uglifyjs2:
# the path to the uglifyjs executable
The path where UglifyJS is installed may vary depending on your system. To find out where npm
stores the bin folder, execute the following command:
Listing 3-6
$ npm bin -g
It should output a folder on your system, inside which you should find the UglifyJS executable.
If you installed UglifyJS locally, you can find the bin folder inside the node_modules folder. It's
called .bin in this case.
You now have access to the uglifyjs2 filter in your application.
Configure the node Binary

Assetic tries to find the node binary automatically. If it cannot be found, you can configure its location
using the node key:
Listing 3-7
1
2
3
4
5
6
assetic:
# the path to the node executable
node: /usr/bin/nodejs
filters:
uglifyjs2:
5. http://browsenpm.org/package.json

7
8
# the path to the uglifyjs executable

Minify your Assets

In order to apply UglifyJS on your assets, add the filter option in the asset tags of your templates to
tell Assetic to use the uglifyjs2 filter:
Listing 3-8
1
2
3
{% javascripts '@AppBundle/Resources/public/js/*' filter='uglifyjs2' %}

The above example assumes that you have a bundle called AppBundle and your JavaScript files
are in the Resources/public/js directory under your bundle. However you can include your
JavaScript files no matter where they are.
With the addition of the uglifyjs2 filter to the asset tags above, you should now see minified
JavaScripts coming over the wire much faster.
Disable Minification in Debug Mode

Minified JavaScripts are very difficult to read, let alone debug. Because of this, Assetic lets you disable a
certain filter when your application is in debug (e.g. app_dev.php) mode. You can do this by prefixing
the filter name in your template with a question mark: ?. This tells Assetic to only apply this filter when
debug mode is off (e.g. app.php):
Listing 3-9
1
2
3
{% javascripts '@AppBundle/Resources/public/js/*' filter='?uglifyjs2' %}

To try this out, switch to your prod environment (app.php). But before you do, don't forget to clear
your cache and dump your assetic assets.
Instead of adding the filters to the asset tags, you can also configure which filters to apply for each
file in your application configuration file. See Filtering Based on a File Extension for more details.
Install, Configure and Use UglifyCSS

The usage of UglifyCSS works the same way as UglifyJS. First, make sure the node package is installed:
Listing 3-10
1
2
3
4
5
6
# global installation
$ npm install -g uglifycss
# local installation
$ cd /path/to/your/symfony/project
$ npm install uglifycss --prefix app/Resources
Next, add the configuration for this filter:

Listing 3-11

1
2
3
4
5
assetic:
filters:
uglifycss:
bin: /usr/local/bin/uglifycss
To use the filter for your CSS files, add the filter to the Assetic stylesheets helper:
Listing 3-12
1
2
3
{% stylesheets 'bundles/App/css/*' filter='uglifycss' filter='cssrewrite' %}

Just like with the uglifyjs2 filter, if you prefix the filter name with ? (i.e. ?uglifycss), the
minification will only happen when you're not in debug mode.

Chapter 4
How to Minify JavaScripts and Stylesheets with

YUI Compressor
The YUI Compressor is no longer maintained by Yahoo1. That's why you are strongly advised to
avoid using YUI utilities unless strictly necessary. Read How to Minify CSS/JS Files (Using UglifyJS
and UglifyCSS) for a modern and up-to-date alternative.
Yahoo! provides an excellent utility for minifying JavaScripts and stylesheets so they travel over the wire
faster, the YUI Compressor2. Thanks to Assetic, you can take advantage of this tool very easily.
Download the YUI Compressor JAR

The YUI Compressor is written in Java and distributed as a JAR. Download the JAR3 from the Yahoo!
website and save it to app/Resources/java/yuicompressor.jar.
Configure the YUI Filters

Now you need to configure two Assetic filters in your application, one for minifying JavaScripts with the
YUI Compressor and one for minifying stylesheets:
Listing 4-1
1
2
3
4
5
6
7
8
assetic:
# java: '/usr/bin/java'
filters:
yui_css:
jar: '%kernel.root_dir%/Resources/java/yuicompressor.jar'
yui_js:
jar: '%kernel.root_dir%/Resources/java/yuicompressor.jar'
1. http://yuiblog.com/blog/2013/01/24/yui-compressor-has-a-new-owner/
2. http://yui.github.io/yuicompressor/
3. https://github.com/yui/yuicompressor/releases

Chapter 4: How to Minify JavaScripts and Stylesheets with YUI Compressor | 20
Windows users need to remember to update config to proper Java location. In Windows7 x64 bit by
default it's C:\Program Files (x86)\Java\jre6\bin\java.exe.
You now have access to two new Assetic filters in your application: yui_css and yui_js. These will
use the YUI Compressor to minify stylesheets and JavaScripts, respectively.
Minify your Assets

You have YUI Compressor configured now, but nothing is going to happen until you apply one of these
filters to an asset. Since your assets are a part of the view layer, this work is done in your templates:
Listing 4-2
1
2
3
{% javascripts '@AppBundle/Resources/public/js/*' filter='yui_js' %}

The above example assumes that you have a bundle called AppBundle and your JavaScript files are
in the Resources/public/js directory under your bundle. This isn't important however - you
can include your JavaScript files no matter where they are.
With the addition of the yui_js filter to the asset tags above, you should now see minified JavaScripts
coming over the wire much faster. The same process can be repeated to minify your stylesheets.
Listing 4-3
1
2
3
{% stylesheets '@AppBundle/Resources/public/css/*' filter='yui_css' %}

<link rel="stylesheet" type="text/css" media="screen" href="{{ asset_url }}" />
Disable Minification in Debug Mode

Minified JavaScripts and stylesheets are very difficult to read, let alone debug. Because of this, Assetic lets
you disable a certain filter when your application is in debug mode. You can do this by prefixing the filter
name in your template with a question mark: ?. This tells Assetic to only apply this filter when debug
mode is off.
Listing 4-4
1
2
3
{% javascripts '@AppBundle/Resources/public/js/*' filter='?yui_js' %}

Instead of adding the filter to the asset tags, you can also globally enable it by adding the apply_to
attribute to the filter configuration, for example in the yui_js filter apply_to: "\.js$". To
only have the filter applied in production, add this to the config_prod file rather than the common
config file. For details on applying filters by file extension, see Filtering Based on a File Extension.

Chapter 4: How to Minify JavaScripts and Stylesheets with YUI Compressor | 21
Chapter 5
How to Use Assetic for Image Optimization

with Twig Functions
Among its many filters, Assetic has four filters which can be used for on-the-fly image optimization. This
allows you to get the benefits of smaller file sizes without having to use an image editor to process each
image. The results are cached and can be dumped for production so there is no performance hit for your
end users.
Using Jpegoptim
Jpegoptim1 is a utility for optimizing JPEG files. To use it with Assetic, make sure to have it already
installed on your system and then, configure its location using the bin option of the jpegoptim filter:
Listing 5-1
1
2
3
4
5
assetic:
filters:
jpegoptim:
bin: path/to/jpegoptim
It can now be used from a template:

Listing 5-2
1
2
3
4
{% image '@AppBundle/Resources/public/images/example.jpg'
filter='jpegoptim' output='/images/example.jpg' %}
<img src="{{ asset_url }}" alt="Example"/>
{% endimage %}
Removing all EXIF Data

By default, the jpegoptim filter removes some meta information stored in the image. To remove all
EXIF data and comments, set the strip_all option to true:
Listing 5-3
1. http://www.kokkonen.net/tjko/projects.html

Chapter 5: How to Use Assetic for Image Optimization with Twig Functions | 22
1
2
3
4
5
6
assetic:
filters:
jpegoptim:
strip_all: true
Lowering Maximum Quality

By default, the jpegoptim filter doesn't alter the quality level of the JPEG image. Use the max option to
configure the maximum quality setting (in a scale of 0 to 100). The reduction in the image file size will
of course be at the expense of its quality:
Listing 5-4
1
2
3
4
5
6
assetic:
filters:
jpegoptim:
max: 70
Shorter Syntax: Twig Function

If you're using Twig, it's possible to achieve all of this with a shorter syntax by enabling and using a
special Twig function. Start by adding the following configuration:
Listing 5-5
1
2
3
4
5
6
7
8
assetic:
filters:
jpegoptim:
twig:
functions:
jpegoptim: ~
The Twig template can now be changed to the following:

Listing 5-6
<img src="{{ jpegoptim('@AppBundle/Resources/public/images/example.jpg') }}" alt="Example"/>
You can also specify the output directory for images in the Assetic configuration file:
Listing 5-7
1
2
3
4
5
6
7
8
assetic:
filters:
jpegoptim:
twig:
functions:
jpegoptim: { output: images/*.jpg }
For uploaded images, you can compress and manipulate them using the LiipImagineBundle2
community bundle.
2. http://knpbundles.com/liip/LiipImagineBundle

Chapter 5: How to Use Assetic for Image Optimization with Twig Functions | 23
Chapter 6
How to Apply an Assetic Filter to a specific File

Extension
Assetic filters can be applied to individual files, groups of files or even, as you'll see here, files that have
a specific extension. To show you how to handle each option, suppose that you want to use Assetic's
CoffeeScript filter, which compiles CoffeeScript files into JavaScript.
The main configuration is just the paths to coffee, node and node_modules. An example
configuration might look like this:
Listing 6-1
1
2
3
4
5
6
7
assetic:
filters:
coffee:
bin:
/usr/bin/coffee
node:
/usr/bin/node
node_paths: [/usr/lib/node_modules/]
Filter a single File

You can now serve up a single CoffeeScript file as JavaScript from within your templates:
Listing 6-2
1
2
3
{% javascripts '@AppBundle/Resources/public/js/example.coffee' filter='coffee' %}

This is all that's needed to compile this CoffeeScript file and serve it as the compiled JavaScript.
Filter multiple Files

You can also combine multiple CoffeeScript files into a single output file:
Listing 6-3

Chapter 6: How to Apply an Assetic Filter to a specific File Extension | 24
1
2
3
4
5
{% javascripts '@AppBundle/Resources/public/js/example.coffee'
'@AppBundle/Resources/public/js/another.coffee'
filter='coffee' %}
Both files will now be served up as a single file compiled into regular JavaScript.
Filtering Based on a File Extension

One of the great advantages of using Assetic is reducing the number of asset files to lower HTTP requests.
In order to make full use of this, it would be good to combine all your JavaScript and CoffeeScript files
together since they will ultimately all be served as JavaScript. Unfortunately just adding the JavaScript
files to the files to be combined as above will not work as the regular JavaScript files will not survive the
CoffeeScript compilation.
This problem can be avoided by using the apply_to option, which allows you to specify which filter
should always be applied to particular file extensions. In this case you can specify that the coffee filter
is applied to all .coffee files:
Listing 6-4
1
2
3
4
5
6
7
8
assetic:
filters:
coffee:
bin:
node:
node_paths:
apply_to:
/usr/bin/coffee
/usr/bin/node
[/usr/lib/node_modules/]
'\.coffee$'
With this option, you no longer need to specify the coffee filter in the template. You can also list
regular JavaScript files, all of which will be combined and rendered as a single JavaScript file (with only
the .coffee files being run through the CoffeeScript filter):
Listing 6-5
1
2
3
4
5
{% javascripts '@AppBundle/Resources/public/js/example.coffee'
'@AppBundle/Resources/public/js/another.coffee'
'@AppBundle/Resources/public/js/regular.js' %}

Chapter 6: How to Apply an Assetic Filter to a specific File Extension | 25
Chapter 7
How to Install 3rd Party Bundles

Most bundles provide their own installation instructions. However, the basic steps for installing a bundle
are the same:
A) Add Composer Dependencies
B) Enable the Bundle
C) Configure the Bundle
A) Add Composer Dependencies

Dependencies are managed with Composer, so if Composer is new to you, learn some basics in their
documentation1. This involves two steps:
1) Find out the Name of the Bundle on Packagist

The README for a bundle (e.g. FOSUserBundle2) usually tells you its name (e.g. friendsofsymfony/
user-bundle). If it doesn't, you can search for the bundle on the Packagist.org3 site.
Looking for bundles? Try searching at KnpBundles.com4: the unofficial archive of Symfony Bundles.
2) Install the Bundle via Composer

Now that you know the package name, you can install it via Composer:
Listing 7-1
$ composer require friendsofsymfony/user-bundle
1. https://getcomposer.org/doc/00-intro.md
2. https://github.com/FriendsOfSymfony/FOSUserBundle
3. https://packagist.org
4. http://knpbundles.com/

Chapter 7: How to Install 3rd Party Bundles | 26
This will choose the best version for your project, add it to composer.json and download its code into
the vendor/ directory. If you need a specific version, include it as the second argument of the composer
require5 command:
Listing 7-2
$ composer require friendsofsymfony/user-bundle "~2.0"
B) Enable the Bundle

At this point, the bundle is installed in your Symfony project (in vendor/friendsofsymfony/) and
the autoloader recognizes its classes. The only thing you need to do now is register the bundle in
AppKernel:
Listing 7-3
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
// app/AppKernel.php
// ...
class AppKernel extends Kernel
{
// ...
public function registerBundles()
{
$bundles = array(
// ...
new FOS\UserBundle\FOSUserBundle(),
);
// ...
}
}
In a few rare cases, you may want a bundle to be only enabled in the development environment. For
example, the DoctrineFixturesBundle helps to load dummy data - something you probably only want to
do while developing. To only load this bundle in the dev and test environments, register the bundle in
this way:
Listing 7-4
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
// ...
{
// ...
{
$bundles = array(
// ...
);
if (in_array($this->getEnvironment(), array('dev', 'test'))) {
$bundles[] = new Doctrine\Bundle\FixturesBundle\DoctrineFixturesBundle();
}
// ...
}
}
5. https://getcomposer.org/doc/03-cli.md#require

C) Configure the Bundle

It's pretty common for a bundle to need some additional setup or configuration in app/config/
config.yml. The bundle's documentation will tell you about the configuration, but you can also get a
reference of the bundle's configuration via the config:dump-reference command:
Listing 7-5
$ app/console config:dump-reference AsseticBundle
Instead of the full bundle name, you can also pass the short name used as the root of the bundle's
configuration:
Listing 7-6
$ app/console config:dump-reference assetic
The output will look like this:

Listing 7-7
1
2
3
4
5
6
7
8
9
10
11
assetic:
debug:
use_controller:
enabled:
profiler:
read_from:
write_to:
java:
node:
node_paths:
# ...
'%kernel.debug%'
'%kernel.debug%'
false
'%kernel.root_dir%/../web'
'%assetic.read_from%'
/usr/bin/java
/usr/local/bin/node
[]
Other Setup
At this point, check the README file of your brand new bundle to see what to do next. Have fun!

Chapter 8
Best Practices for Reusable Bundles

There are two types of bundles:
Application-specific bundles: only used to build your application;
Reusable bundles: meant to be shared across many projects.
This article is all about how to structure your reusable bundles so that they're easy to configure and
extend. Many of these recommendations do not apply to application bundles because you'll want to keep
those as simple as possible. For application bundles, just follow the practices shown throughout the book
and cookbook.
The best practices for application-specific bundles are discussed in The Symfony Framework Best Practices.
Bundle Name
A bundle is also a PHP namespace. The namespace must follow the PSR-01 or PSR-42 interoperability
standards for PHP namespaces and class names: it starts with a vendor segment, followed by zero or more
category segments, and it ends with the namespace short name, which must end with a Bundle suffix.
A namespace becomes a bundle as soon as you add a bundle class to it. The bundle class name must
follow these simple rules:
Use only alphanumeric characters and underscores;

Use a CamelCased name;
Use a descriptive and short name (no more than two words);
Prefix the name with the concatenation of the vendor (and optionally the category namespaces);
Suffix the name with Bundle.
Here are some valid bundle namespaces and class names:
1. http://www.php-fig.org/psr/psr-0/

Chapter 8: Best Practices for Reusable Bundles | 29
Namespace
Bundle Class Name
Acme\Bundle\BlogBundle
AcmeBlogBundle
Acme\BlogBundle
AcmeBlogBundle
By convention, the getName() method of the bundle class should return the class name.
If you share your bundle publicly, you must use the bundle class name as the name of the repository
(AcmeBlogBundle and not BlogBundle for instance).
Symfony core Bundles do not prefix the Bundle class with Symfony and always add a Bundle subnamespace; for example: FrameworkBundle3.
Each bundle has an alias, which is the lower-cased short version of the bundle name using underscores
(acme_blog for AcmeBlogBundle). This alias is used to enforce uniqueness within a project and for
defining bundle's configuration options (see below for some usage examples).
Directory Structure
The basic directory structure of an AcmeBlogBundle must read as follows:
Listing 8-1
1
2
3
4
5
6
7
8
9
10
11
12
13
<your-bundle>/
AcmeBlogBundle.php
Controller/
README.md
LICENSE
Resources/
config/
doc/
index.rst
translations/
views/
public/
Tests/
The following files are mandatory, because they ensure a structure convention that automated tools
can rely on:
AcmeBlogBundle.php:
This is the class that transforms a plain directory into a Symfony bundle (change
this to your bundle's name);
README.md: This file contains the basic description of the bundle and it usually shows some basic
examples and links to its full documentation (it can use any of the markup formats supported by
GitHub, such as README.rst);
LICENSE: The full contents of the license used by the code. Most third-party bundles are published
under the MIT license, but you can choose any license4;
Resources/doc/index.rst: The root file for the Bundle documentation.
The depth of sub-directories should be kept to the minimum for most used classes and files (two levels
maximum).
3. http://api.symfony.com/2.7/Symfony/Bundle/FrameworkBundle/FrameworkBundle.html
4. http://choosealicense.com/

The bundle directory is read-only. If you need to write temporary files, store them under the cache/
or log/ directory of the host application. Tools can generate files in the bundle directory structure, but
only if the generated files are going to be part of the repository.
The following classes and files have specific emplacements (some are mandatory and others are just
conventions followed by most developers):
Type
Directory
Mandatory?
Commands
Command/
Yes
Controllers
Controller/
No
Service Container Extensions
DependencyInjection/
Yes
Event Listeners
EventListener/
No
Model classes [1]
Model/
No
Configuration
Resources/config/
No
Web Resources (CSS, JS, images)
Resources/public/
Yes
Translation files
Resources/translations/
Yes
Templates
Resources/views/
Yes
Unit and Functional Tests
Tests/
No
[1] See How to Provide Model Classes for several Doctrine Implementations for how to handle the mapping
with a compiler pass.
Classes
The
bundle
directory
structure
is
used
as
the
namespace
ContentController
controller
is
stored
in
ContentController.php
and
the
fully
Acme\BlogBundle\Controller\ContentController.
hierarchy.
For
instance,
Acme/BlogBundle/Controller/
qualified
class
name
is
All classes and files must follow the Symfony coding standards.
Some classes should be seen as facades and should be as short as possible, like Commands, Helpers,
Listeners and Controllers.
Classes that connect to the event dispatcher should be suffixed with Listener.
Exception classes should be stored in an Exception sub-namespace.
Vendors
A bundle must not embed third-party PHP libraries. It should rely on the standard Symfony autoloading
instead.
A bundle should not embed third-party libraries written in JavaScript, CSS or any other language.
Tests
A bundle should come with a test suite written with PHPUnit and stored under the Tests/ directory.
Tests should follow the following principles:
The test suite must be executable with a simple phpunit command run from a sample application;
The functional tests should only be used to test the response output and some profiling information
if you have some;
The tests should cover at least 95% of the code base.
A test suite must not contain AllTests.php scripts, but must rely on the existence of a
phpunit.xml.dist file.
Documentation
All classes and functions must come with full PHPDoc.
Extensive documentation should also be provided in the reStructuredText format, under the
Resources/doc/ directory; the Resources/doc/index.rst file is the only mandatory file and
must be the entry point for the documentation.
Installation Instructions
In order to ease the installation of third-party bundles, consider using the following standardized
instructions in your README.md file.
Listing 8-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
Installation
============
Step 1: Download the Bundle
--------------------------Open a command console, enter your project directory and execute the
following command to download the latest stable version of this bundle:
```bash
$ composer require <package-name> "~1"
```
This command requires you to have Composer installed globally, as explained
in the [installation chapter](https://getcomposer.org/doc/00-intro.md)
of the Composer documentation.
Step 2: Enable the Bundle
------------------------Then, enable the bundle by adding it to the list of registered bundles
in the àpp/AppKernel.php` file of your project:
```php
<?php
// ...
{
{
$bundles = array(
// ...
new <vendor>\<bundle-name>\<bundle-long-name>(),
);
// ...
}

42
43
44
// ...
}
```
The example above assumes that you are installing the latest stable version of the bundle, where you don't
have to provide the package version number (e.g. composer require friendsofsymfony/userbundle). If the installation instructions refer to some past bundle version or to some unstable version,
include the version constraint (e.g. composer require friendsofsymfony/user-bundle
"~2.0@dev").
Optionally, you can add more installation steps (Step 3, Step 4, etc.) to explain other required installation
tasks, such as registering routes or dumping assets.
Routing
If the bundle provides routes, they must be prefixed with the bundle alias. For example, if your bundle is
called AcmeBlogBundle, all its routes must be prefixed with acme_blog_.
Templates
If a bundle provides templates, they must use Twig. A bundle must not provide a main layout, except if
it provides a full working application.
Translation Files
If a bundle provides message translations, they must be defined in the XLIFF format; the domain should
be named after the bundle name (acme_blog).
A bundle must not override existing messages from another bundle.
Configuration
To provide more flexibility, a bundle can provide configurable settings by using the Symfony built-in
mechanisms.
For simple configuration settings, rely on the default parameters entry of the Symfony configuration.
Symfony parameters are simple key/value pairs; a value being any valid PHP value. Each parameter name
should start with the bundle alias, though this is just a best-practice suggestion. The rest of the parameter
name will use a period (.) to separate different parts (e.g. acme_blog.author.email).
The end user can provide values in any configuration file:
Listing 8-3
1
2
3
parameters:
acme_blog.author.email: '[email protected]'
Retrieve the configuration parameters in your code from the container:

Listing 8-4
$container->getParameter('acme_blog.author.email');
Even if this mechanism is simple enough, you should consider using the more advanced semantic bundle
configuration.
Versioning
Bundles must be versioned following the Semantic Versioning Standard5.
Services
If the bundle defines services, they must be prefixed with the bundle alias. For example, AcmeBlogBundle
services must be prefixed with acme_blog.
In addition, services not meant to be used by the application directly, should be defined as private.
You can learn much more about service loading in bundles reading this article: How to Load Service
Configuration inside a Bundle.
Composer Metadata
The composer.json file should include at least the following metadata:
name
Consists of the vendor and the short bundle name. If you are releasing the bundle on your own
instead of on behalf of a company, use your personal name (e.g. johnsmith/blog-bundle). The bundle
short name excludes the vendor name and separates each word with an hyphen. For example:
AcmeBlogBundle is transformed into blog-bundle and AcmeSocialConnectBundle is transformed into socialconnect-bundle.
description
A brief explanation of the purpose of the bundle.

type
Use the symfony-bundle value.

license
MIT
is the preferred license for Symfony bundles, but you can use any other license.
autoload
This information is used by Symfony to load the classes of the bundle. The PSR-46 autoload
standard is recommended for modern bundles, but PSR-07 standard is also supported.
In order to make it easier for developers to find your bundle, register it on Packagist8, the official
repository for Composer packages.
Custom Validation Constraints

Starting with Symfony 2.5, a new Validation API was introduced. In fact, there are 3 modes, which the
user can configure in their project:
2.4: the original 2.4 and earlier validation API;
2.5: the new 2.5 and later validation API;
5. http://semver.org/
8. https://packagist.org/

2.5-BC: the new 2.5 API with a backwards-compatible layer so that the 2.4 API still works. This is
only available in PHP 5.3.9+.
Starting with Symfony 2.7, the support for the 2.4 API has been dropped and the minimal PHP
version required for Symfony was increased to 5.3.9. If your bundles requires Symfony >=2.7, you
don't need to take care about the 2.4 API anymore.
As a bundle author, you'll want to support both API's, since some users may still be using the 2.4 API.
Specifically, if your bundle adds a violation directly to the ExecutionContext9 (e.g. like in a custom
validation constraint), you'll need to check for which API is being used. The following code, would work
for all users:
Listing 8-5
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
use Symfony\Component\Validator\ConstraintValidator;
use Symfony\Component\Validator\Constraint;
use Symfony\Component\Validator\Context\ExecutionContextInterface;
// ...
class ContainsAlphanumericValidator extends ConstraintValidator
{
public function validate($value, Constraint $constraint)
{
if ($this->context instanceof ExecutionContextInterface) {
// the 2.5 API
$this->context->buildViolation($constraint->message)
->setParameter('%string%', $value)
->addViolation()
;
} else {
// the 2.4 API
$this->context->addViolation(
$constraint->message,
array('%string%' => $value)
);
}
}
}
Learn more from the Cookbook

How to Load Service Configuration inside a Bundle
9. http://api.symfony.com/2.7/Symfony/Component/Validator/Context/ExecutionContext.html

Chapter 9
How to Use Bundle Inheritance to Override

Parts of a Bundle
When working with third-party bundles, you'll probably come across a situation where you want to
override a file in that third-party bundle with a file in one of your own bundles. Symfony gives you a very
convenient way to override things like controllers, templates, and other files in a bundle's Resources/
directory.
For example, suppose that you're installing the FOSUserBundle1, but you want to override its base
layout.html.twig template, as well as one of its controllers. Suppose also that you have your own
UserBundle where you want the overridden files to live. Start by registering the FOSUserBundle as the
"parent" of your bundle:
Listing 9-1
1
2
3
4
5
6
7
8
9
10
11
12
// src/UserBundle/UserBundle.php
namespace UserBundle;
use Symfony\Component\HttpKernel\Bundle\Bundle;
class UserBundle extends Bundle
{
public function getParent()
{
return 'FOSUserBundle';
}
}
By making this simple change, you can now override several parts of the FOSUserBundle simply by
creating a file with the same name.
Despite the method name, there is no parent/child relationship between the bundles, it is just a way
to extend and override an existing bundle.
1. https://github.com/friendsofsymfony/fosuserbundle

Chapter 9: How to Use Bundle Inheritance to Override Parts of a Bundle | 36
Overriding Controllers
add some functionality to the registerAction of a
RegistrationController that lives inside FOSUserBundle. To do so, just create your own
RegistrationController.php file, override the bundle's original method, and change its
functionality:
Suppose
Listing 9-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
you
want
to
// src/UserBundle/Controller/RegistrationController.php
namespace UserBundle\Controller;
use FOS\UserBundle\Controller\RegistrationController as BaseController;
class RegistrationController extends BaseController
{
public function registerAction()
{
$response = parent::registerAction();
// ... do custom stuff

return $response;
}
}
Depending
on
how
severely
you need to change the behavior, you

parent::registerAction() or completely replace its logic with your own.
might
call
Overriding controllers in this way only works if the bundle refers to the controller using the standard
FOSUserBundle:Registration:register syntax in routes and templates. This is the best
practice.
Overriding Resources: Templates, Routing, etc

Most resources can also be overridden, simply by creating a file in the same location as your parent
bundle.
For example, it's very common to need to override the FOSUserBundle's layout.html.twig template
so that it uses your application's base layout. Since the file lives at Resources/views/
layout.html.twig in the FOSUserBundle, you can create your own file in the same location of
UserBundle. Symfony will ignore the file that lives inside the FOSUserBundle entirely, and use your file
instead.
The same goes for routing files and some other resources.
The overriding of resources only works when you refer to resources with the @FOSUserBundle/
Resources/config/routing/security.xml method. If you refer to resources without using
the @BundleName shortcut, they can't be overridden in this way.
Translation and validation files do not work in the same way as described above. Read "Translations"
if you want to learn how to override translations and see "Validation Metadata" for tricks to override
the validation.

Chapter 9: How to Use Bundle Inheritance to Override Parts of a Bundle | 37
Chapter 10
How to Override any Part of a Bundle

This document is a quick reference for how to override different parts of third-party bundles.
Templates
For information on overriding templates, see
Overriding Bundle Templates.
How to Use Bundle Inheritance to Override Parts of a Bundle
Routing
Routing is never automatically imported in Symfony. If you want to include the routes from any
bundle, then they must be manually imported from somewhere in your application (e.g. app/config/
routing.yml).
The easiest way to "override" a bundle's routing is to never import it at all. Instead of importing a
third-party bundle's routing, simply copy that routing file into your application, modify it, and import it
instead.
Controllers
Assuming the third-party bundle involved uses non-service controllers (which is almost always the case),
you can easily override controllers via bundle inheritance. For more information, see How to Use Bundle
Inheritance to Override Parts of a Bundle. If the controller is a service, see the next section on how to
override it.

Chapter 10: How to Override any Part of a Bundle | 38
Services & Configuration

In order to override/extend a service, there are two options. First, you can set the parameter holding
the service's class name to your own class by setting it in app/config/config.yml. This of course is
only possible if the class name is defined as a parameter in the service config of the bundle containing
the service. For example, to override the class used for Symfony's translator service, you would
override the translator.class parameter. Knowing exactly which parameter to override may take
some research. For the translator, the parameter is defined and used in the Resources/config/
translation.xml file in the core FrameworkBundle:
Listing 10-1
1
2
3
parameters:
translator.class: Acme\HelloBundle\Translation\Translator
Secondly, if the class is not available as a parameter, you want to make sure the class is always overridden
when your bundle is used or if you need to modify something beyond just the class name, you should use
a compiler pass:
Listing 10-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
// src/Acme/DemoBundle/DependencyInjection/Compiler/OverrideServiceCompilerPass.php
namespace Acme\DemoBundle\DependencyInjection\Compiler;
use Symfony\Component\DependencyInjection\Compiler\CompilerPassInterface;
use Symfony\Component\DependencyInjection\ContainerBuilder;
class OverrideServiceCompilerPass implements CompilerPassInterface
{
public function process(ContainerBuilder $container)
{
$definition = $container->getDefinition('original-service-id');
$definition->setClass('Acme\DemoBundle\YourService');
}
}
In this example you fetch the service definition of the original service, and set its class name to your own
class.
See How to Work with Compiler Passes in Bundles for information on how to use compiler passes. If you
want to do something beyond just overriding the class, like adding a method call, you can only use the
compiler pass method.
Entities & Entity Mapping

Due to the way Doctrine works, it is not possible to override entity mapping of a bundle. However,
if a bundle provides a mapped superclass (such as the User entity in the FOSUserBundle) one can
override attributes and associations. Learn more about this feature and its limitations in the Doctrine
documentation1.
Forms
In order to override a form type, it has to be registered as a service (meaning it is tagged as form.type).
You can then override it as you would override any service as explained in Services & Configuration.
This, of course, will only work if the type is referred to by its alias rather than being instantiated, e.g.:
Listing 10-3
$builder->add('name', 'custom_type');
1. http://docs.doctrine-project.org/projects/doctrine-orm/en/latest/reference/inheritance-mapping.html#overrides

rather than:
Listing 10-4
$builder->add('name', new CustomType());
Validation Metadata
Symfony loads all validation configuration files from every bundle and combines them into one validation
metadata tree. This means you are able to add new constraints to a property, but you cannot override
them.
To override this, the 3rd party bundle needs to have configuration for validation groups. For instance,
the FOSUserBundle has this configuration. To create your own validation, add the constraints to a new
validation group:
Listing 10-5
1
2
3
4
5
6
7
8
9
10
# src/Acme/UserBundle/Resources/config/validation.yml
FOS\UserBundle\Model\User:
properties:
plainPassword:
- NotBlank:
groups: [AcmeValidation]
- Length:
min: 6
minMessage: fos_user.password.short
groups: [AcmeValidation]
Now, update the FOSUserBundle configuration, so it uses your validation groups instead of the original
ones.
Translations
Translations are not related to bundles, but to domains. That means that you can override the
translations from any translation file, as long as it is in the correct domain.
The last translation file always wins. That means that you need to make sure that the bundle
containing your translations is loaded after any bundle whose translations you're overriding. This is
done in AppKernel.
Translation files are also not aware of bundle inheritance. If you want to override translations
from the parent bundle, be sure that the parent bundle is loaded before the child bundle in the
AppKernel class.
The file that always wins is the one that is placed in app/Resources/translations, as those
files are always loaded last.

Chapter 11
How to Remove the AcmeDemoBundle

The Symfony Standard Edition comes with a complete demo that lives inside a bundle called
AcmeDemoBundle. It is a great boilerplate to refer to while starting a project, but you'll probably want to
eventually remove it.
This article uses the AcmeDemoBundle as an example, but you can use these steps to remove any
bundle.
1. Unregister the Bundle in the AppKernel

To disconnect the bundle from the framework, you should remove the bundle from the
AppKernel::registerBundles() method. The bundle is normally found in the $bundles array
but the AcmeDemoBundle is only registered in the development environment and you can find it inside
the if statement below:
Listing 11-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
// ...
{
{
$bundles = array(...);
if (in_array($this->getEnvironment(), array('dev', 'test'))) {
// comment or remove this line:
// $bundles[] = new Acme\DemoBundle\AcmeDemoBundle();
// ...
}
}
}

Chapter 11: How to Remove the AcmeDemoBundle | 41
2. Remove Bundle Configuration

Now that Symfony doesn't know about the bundle, you need to remove any configuration and routing
configuration inside the app/config directory that refers to the bundle.
2.1 Remove Bundle Routing

The routing for the AcmeDemoBundle can be found in app/config/routing_dev.yml. Remove the
_acme_demo entry at the bottom of this file.
2.2 Remove Bundle Configuration

Some bundles contain configuration in one of the app/config/config*.yml files. Be sure to remove
the related configuration from these files. You can quickly spot bundle configuration by looking for an
acme_demo (or whatever the name of the bundle is, e.g. fos_user for the FOSUserBundle) string in
the configuration files.
The AcmeDemoBundle doesn't have configuration. However, the bundle is used in the configuration for
the app/config/security.yml file. You can use it as a boilerplate for your own security, but you
can also remove everything: it doesn't matter to Symfony if you remove it or not.
3. Remove the Bundle from the Filesystem

Now you have removed every reference to the bundle in your application, you should remove the bundle
from the filesystem. The bundle is located in the src/Acme/DemoBundle directory. You should remove
this directory and you can remove the Acme directory as well.
If you don't know the location of a bundle, you can use the getPath()1 method to get the path of
the bundle:
Listing 11-2
dump($this->container->get('kernel')->getBundle('AcmeDemoBundle')->getPath());
die();
3.1 Remove Bundle Assets

Remove the assets of the bundle in the web/ directory (e.g. web/bundles/acmedemo for the
AcmeDemoBundle).
4. Remove Integration in other Bundles

This doesn't apply to the AcmeDemoBundle - no other bundles depend on it, so you can skip this
step.
Some bundles rely on other bundles, if you remove one of the two, the other will probably not work. Be
sure that no other bundles, third party or self-made, rely on the bundle you are about to remove.
1. http://api.symfony.com/2.7/Symfony/Component/HttpKernel/Bundle/BundleInterface.html#method_getPath

If one bundle relies on another, in most cases it means that it uses some services from the bundle.
Searching for the bundle alias string may help you spot them (e.g. acme_demo for bundles
depending on AcmeDemoBundle).
If a third party bundle relies on another bundle, you can find that bundle mentioned in the
composer.json file included in the bundle directory.

Chapter 12
How to Load Service Configuration inside a

Bundle
In Symfony, you'll find yourself using many services. These services can be registered in the app/
config/ directory of your application. But when you want to decouple the bundle for use in other
projects, you want to include the service configuration in the bundle itself. This article will teach you how
to do that.
Creating an Extension Class

In order to load service configuration, you have to create a Dependency Injection (DI) Extension for your
bundle. This class has some conventions in order to be detected automatically. But you'll later see how
you can change it to your own preferences. By default, the Extension has to comply with the following
conventions:
It has to live in the DependencyInjection namespace of the bundle;
The name is equal to the bundle name with the Bundle suffix replaced by Extension (e.g. the Extension
class of the AppBundle would be called AppExtension and the one for AcmeHelloBundle would be
called AcmeHelloExtension).
The Extension class should implement the
extend the Extension2 class:
Listing 12-1
1
2
3
4
5
6
7
8
ExtensionInterface1, but usually you would simply
// src/Acme/HelloBundle/DependencyInjection/AcmeHelloExtension.php
namespace Acme\HelloBundle\DependencyInjection;
use Symfony\Component\HttpKernel\DependencyInjection\Extension;
class AcmeHelloExtension extends Extension
{
1. http://api.symfony.com/2.7/Symfony/Component/DependencyInjection/Extension/ExtensionInterface.html
2. http://api.symfony.com/2.7/Symfony/Component/DependencyInjection/Extension/Extension.html

Chapter 12: How to Load Service Configuration inside a Bundle | 44
9
10
11
12
13
public function load(array $configs, ContainerBuilder $container)

{
// ... you'll load the files here later
}
}
Manually Registering an Extension Class

When not following the conventions, you will have to manually register your extension. To do this,
you should override the Bundle::getContainerExtension()3 method to return the instance of the
extension:
Listing 12-2
1
2
3
4
5
6
7
8
9
10
// ...
use Acme\HelloBundle\DependencyInjection\UnconventionalExtensionClass;
class AcmeHelloBundle extends Bundle
{
public function getContainerExtension()
{
return new UnconventionalExtensionClass();
}
}
Since the new Extension class name doesn't follow the naming conventions, you should also override
Extension::getAlias()4 to return the correct DI alias. The DI alias is the name used to refer
to the bundle in the container (e.g. in the app/config/config.yml file). By default, this is done
by removing the Extension suffix and converting the class name to underscores (e.g.
AcmeHelloExtension's DI alias is acme_hello).
Using the load() Method

In the load() method, all services and parameters related to this extension will be loaded. This method
doesn't get the actual container instance, but a copy. This container only has the parameters from the
actual container. After loading the services and parameters, the copy will be merged into the actual
container, to ensure all services and parameters are also added to the actual container.
In the load() method, you can use PHP code to register service definitions, but it is more common if
you put these definitions in a configuration file (using the Yaml, XML or PHP format). Luckily, you can
use the file loaders in the extension!
For instance, assume you have a file called services.xml in the Resources/config directory of
your bundle, your load method looks like:
Listing 12-3
1
2
3
4
5
6
7
8
9
10
11
12
use Symfony\Component\DependencyInjection\Loader\XmlFileLoader;
use Symfony\Component\Config\FileLocator;
// ...
{
$loader = new XmlFileLoader(
$container,
new FileLocator(__DIR__.'/../Resources/config')
);
$loader->load('services.xml');
}
3. http://api.symfony.com/2.7/Symfony/Component/HttpKernel/Bundle/Bundle.html#method_build
4. http://api.symfony.com/2.7/Symfony/Component/DependencyInjection/Extension/Extension.html#method_getAlias

Other available loaders are the YamlFileLoader, PhpFileLoader and IniFileLoader.

The IniFileLoader can only be used to load parameters and it can only load them as strings.
If you removed the default file with service definitions (i.e. app/config/services.yml), make
sure to also remove it from the imports key in app/config/config.yml.
Using Configuration to Change the Services

The Extension is also the class that handles the configuration for that particular bundle (e.g. the
configuration in app/config/config.yml). To read more about it, see the "How to Create Friendly
Configuration for a Bundle" article.
Adding Classes to Compile

Symfony creates a big classes.php file in the cache directory to aggregate the contents of the PHP
classes that are used in every request. This reduces the I/O operations and increases the application
performance.
Your bundles can also add their own classes into this file thanks to the addClassesToCompile()
method. Define the classes to compile as an array of their fully qualified class names:
Listing 12-4
1
2
3
4
5
6
7
8
9
10
11
// ...
{
// ...
$this->addClassesToCompile(array(
'AppBundle\\Manager\\UserManager',
'AppBundle\\Utils\\Slugger',
// ...
));
}
If some class extends from other classes, all its parents are automatically included in the list of classes
to compile.
Beware that this technique can't be used in some cases:

When classes contain annotations, such as controllers with @Route annotations and entities with @ORM
or @Assert annotations, because the file location retrieved from PHP reflection changes;
When classes use the __DIR__ and __FILE__ constants, because their values will change when loading
these classes from the classes.php file.

Chapter 13
How to Create Friendly Configuration for a

Bundle
If you open your application configuration file (usually app/config/config.yml), you'll see a
number of different configuration sections, such as framework, twig and doctrine. Each of these
configures a specific bundle, allowing you to define options at a high level and then let the bundle make
all the low-level, complex changes based on your settings.
For example, the following configuration tells the FrameworkBundle to enable the form integration,
which involves the definition of quite a few services as well as integration of other related components:
Listing 13-1
1
2
framework:
form: true
Using Parameters to Configure your Bundle

If you don't have plans to share your bundle between projects, it doesn't make sense to use this more
advanced way of configuration. Since you use the bundle only in one project, you can just change
the service configuration each time.
If you do want to be able to configure something from within config.yml, you can always create a
parameter there and use that parameter somewhere else.
Using the Bundle Extension

The basic idea is that instead of having the user override individual parameters, you let the user
configure just a few, specifically created, options. As the bundle developer, you then parse through that
configuration and load correct services and parameters inside an "Extension" class.
As an example, imagine you are creating a social bundle, which provides integration with Twitter and
such. To be able to reuse your bundle, you have to make the client_id and client_secret variables
configurable. Your bundle configuration would look like:

Chapter 13: How to Create Friendly Configuration for a Bundle | 47
Listing 13-2
1
2
3
4
5
acme_social:
twitter:
client_id: 123
client_secret: your_secret
Read more about the extension in How to Load Service Configuration inside a Bundle.
If a bundle provides an Extension class, then you should not generally override any service container
parameters from that bundle. The idea is that if an Extension class is present, every setting that
should be configurable should be present in the configuration made available by that class. In
other words, the extension class defines all the public configuration settings for which backward
compatibility will be maintained.
For parameter handling within a dependency injection container see Using Parameters within a Dependency
Injection Class.
Processing the $configs Array

First things first, you have to create an extension class as explained in How to Load Service Configuration
inside a Bundle.
Whenever a user includes the acme_social key (which is the DI alias) in a configuration file, the
configuration under it is added to an array of configurations and passed to the load() method of your
extension (Symfony automatically converts XML and YAML to an array).
For the configuration example in the previous section, the array passed to your load() method will look
like this:
Listing 13-3
1
2
3
4
5
6
7
8
array(
array(
'twitter' => array(
'client_id' => 123,
'client_secret' => 'your_secret',
),
),
)
Notice that this is an array of arrays, not just a single flat array of the configuration values. This is
intentional, as it allows Symfony to parse several configuration resources. For example, if acme_social
appears in another configuration file - say config_dev.yml - with different values beneath it, the
incoming array might look like this:
Listing 13-4
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
array(
// values from config.yml
array(
'twitter' => array(
'client_id' => 123,
'client_secret' => 'your_secret',
),
),
// values from config_dev.yml
array(
'twitter' => array(
'client_id' => 456,
),
),
)

The order of the two arrays depends on which one is set first.
But don't worry! Symfony's Config component will help you merge these values, provide defaults and give
the user validation errors on bad configuration. Here's how it works. Create a Configuration class
in the DependencyInjection directory and build a tree that defines the structure of your bundle's
configuration.
The Configuration class to handle the sample configuration looks like:
Listing 13-5
// src/Acme/SocialBundle/DependencyInjection/Configuration.php
namespace Acme\SocialBundle\DependencyInjection;
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
use Symfony\Component\Config\Definition\Builder\TreeBuilder;
use Symfony\Component\Config\Definition\ConfigurationInterface;
class Configuration implements ConfigurationInterface
{
public function getConfigTreeBuilder()
{
$treeBuilder = new TreeBuilder();
$rootNode = $treeBuilder->root('acme_social');
$rootNode
->children()
->arrayNode('twitter')
->children()
->integerNode('client_id')->end()
->scalarNode('client_secret')->end()
->end()
->end() // twitter
->end()
;
return $treeBuilder;
}
}
The Configuration class can be much more complicated than shown here, supporting "prototype" nodes,
advanced validation, XML-specific normalization and advanced merging. You can read more about this in
the Config component documentation. You can also see it in action by checking out some core Configuration
classes, such as the one from the FrameworkBundle Configuration1 or the TwigBundle Configuration2.
This class can now be used in your load() method to merge configurations and force validation (e.g. if
an additional option was passed, an exception will be thrown):
Listing 13-6
1
2
3
4
5
6
7

{
$configuration = new Configuration();
$config = $this->processConfiguration($configuration, $configs);
// ...
}
The processConfiguration() method uses the configuration tree you've defined in the
Configuration class to validate, normalize and merge all the configuration arrays together.
1. https://github.com/symfony/symfony/blob/master/src/Symfony/Bundle/FrameworkBundle/DependencyInjection/Configuration.php
2. https://github.com/symfony/symfony/blob/master/src/Symfony/Bundle/TwigBundle/DependencyInjection/Configuration.php

Instead of calling processConfiguration() in your extension each time you provide some
configuration options, you might want to use the ConfigurableExtension3 to do this
automatically for you:
Listing 13-7
1
2
3
4
5
6
7
8
9
10
11
12
13
14
use Symfony\Component\HttpKernel\DependencyInjection\ConfigurableExtension;
class AcmeHelloExtension extends ConfigurableExtension
{
// note that this method is called loadInternal and not load
protected function loadInternal(array $mergedConfig, ContainerBuilder $container)
{
// ...
}
}
This class uses the getConfiguration() method to get the Configuration instance. You should
override it if your Configuration class is not called Configuration or if it is not placed in the same
namespace as the extension.
Processing the Configuration yourself

Using the Config component is fully optional. The load() method gets an array of configuration
values. You can simply parse these arrays yourself (e.g. by overriding configurations and using
isset4 to check for the existence of a value). Be aware that it'll be very hard to support XML.
Listing 13-8
1
2
3
4
5
6
7
8
9
10

{
$config = array();
// let resources override the previous set value
foreach ($configs as $subConfig) {
$config = array_merge($config, $subConfig);
}
// ... now use the flat $config array

}
Modifying the Configuration of Another Bundle

If you have multiple bundles that depend on each other, it may be useful to allow one Extension class
to modify the configuration passed to another bundle's Extension class, as if the end-developer has
actually placed that configuration in their app/config/config.yml file. This can be achieved using a
prepend extension. For more details, see How to Simplify Configuration of multiple Bundles.
Dump the Configuration

The config:dump-reference command dumps the default configuration of a bundle in the console
using the Yaml format.
3. http://api.symfony.com/2.7/Symfony/Component/HttpKernel/DependencyInjection/ConfigurableExtension.html
4. http://php.net/manual/en/function.isset.php

As long as your bundle's configuration is located in the standard location

(YourBundle\DependencyInjection\Configuration) and does not have a constructor it will
work automatically. If you have something different, your Extension class must override the
Extension::getConfiguration()5 method and return an instance of your Configuration.
Supporting XML
Symfony allows people to provide the configuration in three different formats: Yaml, XML and PHP.
Both Yaml and PHP use the same syntax and are supported by default when using the Config component.
Supporting XML requires you to do some more things. But when sharing your bundle with others, it is
recommended that you follow these steps.
Make your Config Tree ready for XML

The Config component provides some methods by default to allow it to correctly process XML
configuration. See "Normalization" of the component documentation. However, you can do some
optional things as well, this will improve the experience of using XML configuration:
Choosing an XML Namespace

In XML, the XML namespace6 is used to determine which elements belong to the configuration of a
specific bundle. The namespace is returned from the Extension::getNamespace()7 method. By
convention, the namespace is a URL (it doesn't have to be a valid URL nor does it need to exists).
By default, the namespace for a bundle is http://example.org/dic/schema/DI_ALIAS, where
DI_ALIAS is the DI alias of the extension. You might want to change this to a more professional URL:
Listing 13-9
1
2
3
4
5
6
7
8
9
10
11
12
// ...
{
// ...
public function getNamespace()
{
return 'http://acme_company.com/schema/dic/hello';
}
}
Providing an XML Schema

XML has a very useful feature called XML schema8. This allows you to describe all possible elements and
attributes and their values in an XML Schema Definition (an xsd file). This XSD file is used by IDEs for
auto completion and it is used by the Config component to validate the elements.
In order to use the schema, the XML configuration file must provide an xsi:schemaLocation
attribute pointing to the XSD file for a certain XML namespace. This location always starts with the
XML namespace. This XML namespace is then replaced with the XSD validation base path returned
from Extension::getXsdValidationBasePath()9 method. This namespace is then followed by
the rest of the path from the base path to the file itself.
5. http://api.symfony.com/2.7/Symfony/Component/HttpKernel/DependencyInjection/Extension.html#method_getConfiguration
6. https://en.wikipedia.org/wiki/XML_namespace
7. http://api.symfony.com/2.7/Symfony/Component/DependencyInjection/Extension/Extension.html#method_getNamespace
8. https://en.wikipedia.org/wiki/XML_schema
9. http://api.symfony.com/2.7/Symfony/Component/DependencyInjection/Extension/ExtensionInterface.html#method_getXsdValidationBasePath

By convention, the XSD file lives in the Resources/config/schema, but you can place it anywhere
you like. You should return this path as the base path:
Listing 13-10
1
2
3
4
5
6
7
8
9
10
11
12
// ...
{
// ...
public function getXsdValidationBasePath()
{
return __DIR__.'/../Resources/config/schema';
}
}
the XSD file is called hello-1.0.xsd, the schema

http://acme_company.com/schema/dic/hello/hello-1.0.xsd:
Assuming
Listing 13-11
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
location
will
be


<?xml version="1.0" ?>
<container xmlns="http://symfony.com/schema/dic/services"
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xmlns:acme-hello="http://acme_company.com/schema/dic/hello"
xsi:schemaLocation="http://acme_company.com/schema/dic/hello
http://acme_company.com/schema/dic/hello/hello-1.0.xsd">
<acme-hello:config>

</acme-hello:config>


</container>

Chapter 14
How to Simplify Configuration of multiple

Bundles
When building reusable and extensible applications, developers are often faced with a choice: either
create a single large bundle or multiple smaller bundles. Creating a single bundle has the drawback that
it's impossible for users to choose to remove functionality they are not using. Creating multiple bundles
has the drawback that configuration becomes more tedious and settings often need to be repeated for
various bundles.
Using the below approach, it is possible to remove the disadvantage of the multiple bundle approach by
enabling a single Extension to prepend the settings for any bundle. It can use the settings defined in the
app/config/config.yml to prepend settings just as if they had been written explicitly by the user in
the application configuration.
For example, this could be used to configure the entity manager name to use in multiple bundles. Or it
can be used to enable an optional feature that depends on another bundle being loaded as well.
To give an Extension the power to do this, it needs to implement PrependExtensionInterface1:
Listing 14-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
use Symfony\Component\DependencyInjection\Extension\PrependExtensionInterface;
class AcmeHelloExtension extends Extension implements PrependExtensionInterface
{
// ...
public function prepend(ContainerBuilder $container)
{
// ...
}
}
1. http://api.symfony.com/2.7/Symfony/Component/DependencyInjection/Extension/PrependExtensionInterface.html

Chapter 14: How to Simplify Configuration of multiple Bundles | 53
Inside the prepend()2 method, developers have full access to the ContainerBuilder3 instance just
before the load()4 method is called on each of the registered bundle Extensions. In order to prepend
settings to a bundle extension developers can use the prependExtensionConfig()5 method on the
ContainerBuilder6 instance. As this method only prepends settings, any other settings done explicitly
inside the app/config/config.yml would override these prepended settings.
The following example illustrates how to prepend a configuration setting in multiple bundles as well as
disable a flag in multiple bundles in case a specific other bundle is not registered:
Listing 14-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
public function prepend(ContainerBuilder $container)

{
// get all bundles
$bundles = $container->getParameter('kernel.bundles');
// determine if AcmeGoodbyeBundle is registered
if (!isset($bundles['AcmeGoodbyeBundle'])) {
// disable AcmeGoodbyeBundle in bundles
$config = array('use_acme_goodbye' => false);
foreach ($container->getExtensions() as $name => $extension) {
switch ($name) {
case 'acme_something':
case 'acme_other':
// set use_acme_goodbye to false in the config of
// acme_something and acme_other note that if the user manually
// configured use_acme_goodbye to true in the app/config/config.yml
// then the setting would in the end be true and not false
$container->prependExtensionConfig($name, $config);
break;
}
}
}
// process the configuration of AcmeHelloExtension

$configs = $container->getExtensionConfig($this->getAlias());
// use the Configuration class to generate a config array with
// the settings "acme_hello"
$config = $this->processConfiguration(new Configuration(), $configs);
// check if entity_manager_name is set in the "acme_hello" configuration
if (isset($config['entity_manager_name'])) {
// prepend the acme_something settings with the entity_manager_name
$config = array('entity_manager_name' => $config['entity_manager_name']);
$container->prependExtensionConfig('acme_something', $config);
}
}
The above would be the equivalent of writing the following into the app/config/config.yml in case
AcmeGoodbyeBundle is not registered and the entity_manager_name setting for acme_hello is set
to non_default:
Listing 14-3
1
2
3
4
5
6
7
8
9
acme_something:
# ...
use_acme_goodbye: false
entity_manager_name: non_default
acme_other:
# ...
use_acme_goodbye: false
2. http://api.symfony.com/2.7/Symfony/Component/DependencyInjection/Extension/PrependExtensionInterface.html#method_prepend
3. http://api.symfony.com/2.7/Symfony/Component/DependencyInjection/ContainerBuilder.html
4. http://api.symfony.com/2.7/Symfony/Component/DependencyInjection/Extension/ExtensionInterface.html#method_load
5. http://api.symfony.com/2.7/Symfony/Component/DependencyInjection/ContainerBuilder.html#method_prependExtensionConfig
6. http://api.symfony.com/2.7/Symfony/Component/DependencyInjection/ContainerBuilder.html

Chapter 14: How to Simplify Configuration of multiple Bundles | 54
Chapter 15
How to Use Varnish to Speed up my Website

Because Symfony's cache uses the standard HTTP cache headers, the Symfony Reverse Proxy can easily
be replaced with any other reverse proxy. Varnish1 is a powerful, open-source, HTTP accelerator capable
of serving cached content fast and including support for Edge Side Includes.
Make Symfony Trust the Reverse Proxy

Varnish automatically forwards the IP as X-Forwarded-For and leaves the X-Forwarded-Proto
header in the request. If you do not configure Varnish as trusted proxy, Symfony will see all requests as
coming through insecure HTTP connections from the Varnish host instead of the real client.
Remember to configure framework.trusted_proxies in the Symfony configuration so that Varnish is seen
as a trusted proxy and the X-Forwarded headers are used.
Varnish, in its default configuration, sends the X-Forwarded-For header but does not filter out the
Forwarded header. If you have access to the Varnish configuration file, you can configure Varnish to
remove the Forwarded header:
Listing 15-1
1
2
3
sub vcl_recv {
remove req.http.Forwarded;
}
If you do not have access to your Varnish configuration, you can instead configure Symfony to distrust
the Forwarded header as detailed in the cookbook.
Routing and X-FORWARDED Headers

To ensure that the Symfony Router generates URLs correctly with Varnish, an X-Forwarded-Port
header must be present for Symfony to use the correct port number.
This port number corresponds to the port your setup is using to receive external connections (80 is the
default value for HTTP connections). If the application also accepts HTTPS connections, there could be
1. https://www.varnish-cache.org

Chapter 15: How to Use Varnish to Speed up my Website | 55
another proxy (as Varnish does not do HTTPS itself) on the default HTTPS port 443 that handles the
SSL termination and forwards the requests as HTTP requests to Varnish with an X-Forwarded-Proto
header. In this case, you need to add the following configuration snippet:
Listing 15-2
1
2
3
4
5
6
7
sub vcl_recv {
if (req.http.X-Forwarded-Proto == "https" ) {
set req.http.X-Forwarded-Port = "443";
} else {
set req.http.X-Forwarded-Port = "80";
}
}
Cookies and Caching

By default, a sane caching proxy does not cache anything when a request is sent with cookies or a basic
authentication header. This is because the content of the page is supposed to depend on the cookie value
or authentication header.
If you know for sure that the backend never uses sessions or basic authentication, have Varnish remove
the corresponding header from requests to prevent clients from bypassing the cache. In practice, you will
need sessions at least for some parts of the site, e.g. when using forms with CSRF Protection. In this
situation, make sure to only start a session when actually needed and clear the session when it is no longer
needed. Alternatively, you can look into Caching Pages that Contain CSRF Protected Forms.
Cookies created in JavaScript and used only in the frontend, e.g. when using Google Analytics, are
nonetheless sent to the server. These cookies are not relevant for the backend and should not affect the
caching decision. Configure your Varnish cache to clean the cookies header2. You want to keep the session
cookie, if there is one, and get rid of all other cookies so that pages are cached if there is no active session.
Unless you changed the default configuration of PHP, your session cookie has the name PHPSESSID:
Listing 15-3
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
sub vcl_recv {
// Remove all cookies except the session ID.
if (req.http.Cookie) {
set req.http.Cookie = ";" + req.http.Cookie;
set req.http.Cookie = regsuball(req.http.Cookie,
"; +", ";");

";(PHPSESSID)=", "; \1=");
";[^ ][^;]*", "");
"^[; ]+|[; ]+$", "");
if (req.http.Cookie == "") {
// If there are no more cookies, remove the header to get page cached.
unset req.http.Cookie;
}
}
}
If content is not different for every user, but depends on the roles of a user, a solution is to separate
the cache per group. This pattern is implemented and explained by the FOSHttpCacheBundle3 under
the name User Context4.
2. https://www.varnish-cache.org/trac/wiki/VCLExampleRemovingSomeCookies
3. http://foshttpcachebundle.readthedocs.org/
4. http://foshttpcachebundle.readthedocs.org/en/latest/features/user-context.html

Ensure Consistent Caching Behavior

Varnish uses the cache headers sent by your application to determine how to cache content. However,
versions prior to Varnish 4 did not respect Cache-Control: no-cache, no-store and private.
To ensure consistent behavior, use the following configuration if you are still using Varnish 3:
Listing 15-4
1
2
3
4
5
6
7
8
9
10
11
sub vcl_fetch {
/* By default, Varnish3 ignores Cache-Control: no-cache and private
https://www.varnish-cache.org/docs/3.0/tutorial/increasing_your_hitrate.html#cache-control
*/
if (beresp.http.Cache-Control ~ "private" ||
beresp.http.Cache-Control ~ "no-cache" ||
beresp.http.Cache-Control ~ "no-store"
) {
return (hit_for_pass);
}
}
You can see the default behavior of Varnish in the form of a VCL file: default.vcl5 for Varnish 3,
builtin.vcl6 for Varnish 4.
Enable Edge Side Includes (ESI)

As explained in the Edge Side Includes section, Symfony detects whether it talks to a reverse proxy that
understands ESI or not. When you use the Symfony reverse proxy, you don't need to do anything. But to
make Varnish instead of Symfony resolve the ESI tags, you need some configuration in Varnish. Symfony
uses the Surrogate-Capability header from the Edge Architecture7 described by Akamai.
Varnish only supports the src attribute for ESI tags (onerror and alt attributes are ignored).
First, configure Varnish so that it advertises its ESI support by adding a Surrogate-Capability
header to requests forwarded to the backend application:
Listing 15-5
1
2
3
4
sub vcl_recv {
// Add a Surrogate-Capability header to announce ESI support.
set req.http.Surrogate-Capability = "abc=ESI/1.0";
}
The abc part of the header isn't important unless you have multiple "surrogates" that need to
advertise their capabilities. See Surrogate-Capability Header8 for details.
Then, optimize Varnish so that it only parses the response contents when there is at least one ESI tag by
checking the Surrogate-Control header that Symfony adds automatically:
Listing 15-6
1
2
sub vcl_backend_response {
// Check for ESI acknowledgement and remove Surrogate-Control header
5. https://github.com/varnish/Varnish-Cache/blob/3.0/bin/varnishd/default.vcl
6. https://github.com/varnish/Varnish-Cache/blob/4.1/bin/varnishd/builtin.vcl
7. http://www.w3.org/TR/edge-arch
8. http://www.w3.org/TR/edge-arch

3
4
5
6
7
if (beresp.http.Surrogate-Control ~ "ESI/1.0") {
unset beresp.http.Surrogate-Control;
set beresp.do_esi = true;
}
}
If you followed the advice about ensuring a consistent caching behavior, those VCL functions already
exist. Just append the code to the end of the function, they won't interfere with each other.
Cache Invalidation
If you want to cache content that changes frequently and still serve the most recent version to users, you
need to invalidate that content. While cache invalidation9 allows you to purge content from your proxy
before it has expired, it adds complexity to your caching setup.
The open source FOSHttpCacheBundle10 takes the pain out of cache invalidation by helping you to
organize your caching and invalidation setup.
The documentation of the FOSHttpCacheBundle11 explains how to configure Varnish and other
reverse proxies for cache invalidation.
9. http://tools.ietf.org/html/rfc2616#section-13.10

Chapter 16
Caching Pages that Contain CSRF Protected

Forms
CSRF tokens are meant to be different for every user. This is why you need to be cautious if you try to
cache pages with forms including them.
For more information about how CSRF protection works in Symfony, please check CSRF Protection.
Why Caching Pages with a CSRF token is Problematic

Typically, each user is assigned a unique CSRF token, which is stored in the session for validation. This
means that if you do cache a page with a form containing a CSRF token, you'll cache the CSRF token of
the first user only. When a user submits the form, the token won't match the token stored in the session
and all users (except for the first) will fail CSRF validation when submitting the form.
In fact, many reverse proxies (like Varnish) will refuse to cache a page with a CSRF token. This is because
a cookie is sent in order to preserve the PHP session open and Varnish's default behavior is to not cache
HTTP requests with cookies.
How to Cache Most of the Page and still be able to Use CSRF Protection
To cache a page that contains a CSRF token, you can use more advanced caching techniques like ESI
fragments, where you cache the full page and embedding the form inside an ESI tag with no cache at all.
Another option would be to load the form via an uncached AJAX request, but cache the rest of the HTML
response.
Or you can even load just the CSRF token with an AJAX request and replace the form field value with it.

Chapter 16: Caching Pages that Contain CSRF Protected Forms | 59
Chapter 17
Installing Composer
Composer1 is the package manager used by modern PHP applications. Use Composer to manage
dependencies in your Symfony applications and to install Symfony Components in your PHP projects.
It's recommended to install Composer globally in your system as explained in the following sections.
Install Composer on Linux and Mac OS X

1. Run the installer as described in the official Composer documentation2;
2. Execute the following command to move the composer.phar to a directory that is in your
path:
Listing 17-1
$ sudo mv composer.phar /usr/local/bin/composer
Install Composer on Windows

Download the installer from getcomposer.org3, execute it and follow the instructions.
Learn more
Read the Composer documentation4 to learn more about its usage and features.
1. https://getcomposer.org/
2. https://getcomposer.org/download
3. https://getcomposer.org/Composer-Setup.exe
4. https://getcomposer.org/doc/00-intro.md

Chapter 17: Installing Composer | 60
Chapter 18
How to Master and Create new Environments

Every application is the combination of code and a set of configuration that dictates how that code should
function. The configuration may define the database being used, if something should be cached or how
verbose logging should be.
In Symfony, the idea of "environments" is the idea that the same codebase can be run using multiple
different configurations. For example, the dev environment should use configuration that makes
development easy and friendly, while the prod environment should use a set of configuration optimized
for speed.
Different Environments, different Configuration Files

A typical Symfony application begins with three environments: dev, prod, and test. As mentioned,
each environment simply represents a way to execute the same codebase with different configuration. It
should be no surprise then that each environment loads its own individual configuration file. If you're
using the YAML configuration format, the following files are used:
for the dev environment: app/config/config_dev.yml
for the prod environment: app/config/config_prod.yml
for the test environment: app/config/config_test.yml
This works via a simple standard that's used by default inside the AppKernel class:
Listing 18-1
1
2
3
4
5
6
7
8
9
10
11
12
13
// ...
{
// ...
public function registerContainerConfiguration(LoaderInterface $loader)
{
$loader->load($this->getRootDir().'/config/config_'.$this->getEnvironment().'.yml');
}
}

Chapter 18: How to Master and Create new Environments | 61
As you can see, when Symfony is loaded, it uses the given environment to determine which configuration
file to load. This accomplishes the goal of multiple environments in an elegant, powerful and transparent
way.
Of course, in reality, each environment differs only somewhat from others. Generally, all environments
will share a large base of common configuration. Opening the config_dev.yml configuration file, you
can see how this is accomplished easily and transparently:
Listing 18-2
1
2
3
4
imports:
- { resource: config.yml }
# ...
To share common configuration, each environment's configuration file simply first imports from a
central configuration file (config.yml). The remainder of the file can then deviate from the default
configuration by overriding individual parameters. For example, by default, the web_profiler toolbar
is disabled. However, in the dev environment, the toolbar is activated by modifying the value of the
toolbar option in the config_dev.yml configuration file:
Listing 18-3
1
2
3
4
5
6
7
imports:
web_profiler:
toolbar: true
# ...
Executing an Application in different Environments

To execute the application in each environment, load up the application using either the app.php (for
the prod environment) or the app_dev.php (for the dev environment) front controller:
Listing 18-4
1
2
http://localhost/app.php
http://localhost/app_dev.php
-> *prod* environment

-> *dev* environment
The given URLs assume that your web server is configured to use the web/ directory of the
application as its root. Read more in Installing Symfony.
If you open up one of these files, you'll quickly see that the environment used by each is explicitly set:
Listing 18-5
1
2
3
4
5
6
// web/app.php
// ...
$kernel = new AppKernel('prod', false);
// ...
The prod key specifies that this application will run in the prod environment. A Symfony application
can be executed in any environment by using this code and changing the environment string.
The test environment is used when writing functional tests and is not accessible in the browser
directly via a front controller. In other words, unlike the other environments, there is no
app_test.php front controller file.

Debug Mode
Important, but unrelated to the topic of environments is the false argument as the second argument
to the AppKernel constructor. This specifies if the application should run in "debug mode".
Regardless of the environment, a Symfony application can be run with debug mode set to true or
false. This affects many things in the application, such as if errors should be displayed or if cache
files are dynamically rebuilt on each request. Though not a requirement, debug mode is generally set
to true for the dev and test environments and false for the prod environment.
Internally, the value of the debug mode becomes the kernel.debug parameter used inside the
service container. If you look inside the application configuration file, you'll see the parameter used,
for example, to turn logging on or off when using the Doctrine DBAL:
Listing 18-6
1
2
3
4
doctrine:
dbal:
logging: '%kernel.debug%'
# ...
As of Symfony 2.3, showing errors or not no longer depends on the debug mode. You'll need to
enable that in your front controller by calling enable()1.
Selecting the Environment for Console Commands

By default, Symfony commands are executed in the dev environment and with the debug mode enabled.
Use the --env and --no-debug options to modify this behavior:
Listing 18-7
1
2
3
4
5
6
7
8
# 'dev' environment and debug enabled

$ php app/console command_name
# 'prod' environment (debug is always disabled for 'prod')
$ php app/console command_name --env=prod
# 'test' environment and debug disabled
$ php app/console command_name --env=test --no-debug
In addition to the --env and --debug options, the behavior of Symfony commands can also be
controlled with environment variables. The Symfony console application checks the existence and value
of these environment variables before executing any command:
SYMFONY_ENV
Sets the execution environment of the command to the value of this variable (dev, prod, test, etc.);
SYMFONY_DEBUG
If 0, debug mode is disabled. Otherwise, debug mode is enabled.

These environment variables are very useful for production servers because they allow you to ensure that
commands always run in the prod environment without having to add any command option.
Creating a new Environment

By default, a Symfony application has three environments that handle most cases. Of course, since an
environment is nothing more than a string that corresponds to a set of configuration, creating a new
environment is quite easy.
1. http://api.symfony.com/2.7/Symfony/Component/Debug/Debug.html#method_enable

Suppose, for example, that before deployment, you need to benchmark your application. One way
to benchmark the application is to use near-production settings, but with Symfony's web_profiler
enabled. This allows Symfony to record information about your application while benchmarking.
The best way to accomplish this is via a new environment called, for example, benchmark. Start by
creating a new configuration file:
Listing 18-8
1
2
3
4
5
6
# app/config/config_benchmark.yml
imports:
- { resource: config_prod.yml }
framework:
profiler: { only_exceptions: false }
Due to the way in which parameters are resolved, you cannot use them to build paths in imports
dynamically. This means that something like the following doesn't work:
Listing 18-9
1
2
3
imports:
- { resource: '%kernel.root_dir%/parameters.yml' }
And with this simple addition, the application now supports a new environment called benchmark.
This new configuration file imports the configuration from the prod environment and modifies it. This
guarantees that the new environment is identical to the prod environment, except for any changes
explicitly made here.
Because you'll want this environment to be accessible via a browser, you should also create a front
controller for it. Copy the web/app.php file to web/app_benchmark.php and edit the environment
to be benchmark:
Listing 18-10
1
2
3
4
5
6
7
// web/app_benchmark.php
// ...
// change just this line
$kernel = new AppKernel('benchmark', false);
// ...
The new environment is now accessible via:

Listing 18-11
http://localhost/app_benchmark.php
Some environments, like the dev environment, are never meant to be accessed on any deployed
server by the public. This is because certain environments, for debugging purposes, may give too
much information about the application or underlying infrastructure. To be sure these environments
aren't accessible, the front controller is usually protected from external IP addresses via the following
code at the top of the controller:
Listing 18-12
if (!in_array(@$_SERVER['REMOTE_ADDR'], array('127.0.0.1', '::1'))) {

die('You are not allowed to access this file. Check '.basename(__FILE__).' for more information.');
}
Environments and the Cache Directory

Symfony takes advantage of caching in many ways: the application configuration, routing configuration,
Twig templates and more are cached to PHP objects stored in files on the filesystem.
By default, these cached files are largely stored in the app/cache directory. However, each environment
caches its own set of files:
Listing 18-13
1
2
3
4
5
6
<your-project>/
app/
cache/
dev/
prod/
...
# cache directory for the *dev* environment

# cache directory for the *prod* environment
Sometimes, when debugging, it may be helpful to inspect a cached file to understand how something
is working. When doing so, remember to look in the directory of the environment you're using (most
commonly dev while developing and debugging). While it can vary, the app/cache/dev directory
includes the following:
appDevDebugProjectContainer.php
The cached "service container" that represents the cached application configuration.
appDevUrlGenerator.php
The PHP class generated from the routing configuration and used when generating URLs.
appDevUrlMatcher.php
The PHP class used for route matching - look here to see the compiled regular expression logic used
to match incoming URLs to different routes.
twig/
This directory contains all the cached Twig templates.

You can easily change the directory location and name. For more information read the article How
to Override Symfony's default Directory Structure.
Going further
Read the article on How to Set external Parameters in the Service Container.

Chapter 19
How to Override Symfony's default Directory

Structure
Symfony automatically ships with a default directory structure. You can easily override this directory
structure to create your own. The default directory structure is:
Listing 19-1
1
2
3
4
5
6
7
8
9
10
11
12
13
your-project/
app/
cache/
config/
logs/
...
src/
...
vendor/
...
web/
app.php
...
Override the cache Directory

You can change the default cache directory by overriding the getCacheDir method in the AppKernel
class of your application:
Listing 19-2
1
2
3
4
5
6
7
8
9
10
11
12
// ...
{
// ...
public function getCacheDir()
{
return $this->rootDir.'/'.$this->environment.'/cache';
}
}

Chapter 19: How to Override Symfony's default Directory Structure | 66
$this->rootDir is the absolute path to the app directory and $this->environment is the current
environment (i.e. dev). In this case you have changed the location of the cache directory to app/
{environment}/cache.
You should keep the cache directory different for each environment, otherwise some unexpected
behavior may happen. Each environment generates its own cached configuration files, and so each
needs its own directory to store those cache files.
Override the logs Directory

Overriding the logs directory is the same as overriding the cache directory. The only difference is that
you need to override the getLogDir method:
Listing 19-3
1
2
3
4
5
6
7
8
9
10
11
12
// ...
{
// ...
public function getLogDir()
{
return $this->rootDir.'/'.$this->environment.'/logs';
}
}
Here you have changed the location of the directory to app/{environment}/logs.
Override the web Directory

If you need to rename or move your web directory, the only thing you need to guarantee is that the path
to the app directory is still correct in your app.php and app_dev.php front controllers. If you simply
renamed the directory, you're fine. But if you moved it in some way, you may need to modify these paths
inside those files:
Listing 19-4
require_once __DIR__.'/../Symfony/app/bootstrap.php.cache';
require_once __DIR__.'/../Symfony/app/AppKernel.php';
You also need to change the extra.symfony-web-dir option in the composer.json file:
Listing 19-5
1
2
3
4
5
6
7
{
...
"extra": {
...
"symfony-web-dir": "my_new_web_dir"
}
}
Some shared hosts have a public_html web directory root. Renaming your web directory from
web to public_html is one way to make your Symfony project work on your shared host. Another
way is to deploy your application to a directory outside of your web root, delete your public_html
directory, and then replace it with a symbolic link to the web in your project.

If you use the AsseticBundle, you need to configure the read_from option to point to the correct
web directory:
Listing 19-6
1
2
3
4
5
6
# ...
assetic:
# ...
read_from: '%kernel.root_dir%/../../public_html'
Now you just need to clear the cache and dump the assets again and your application should work:
Listing 19-7
1
2
$ php app/console cache:clear --env=prod

Override the vendor Directory

To override the vendor directory, you need to introduce changes in the app/autoload.php and
composer.json files.
The change in the composer.json will look like this:
Listing 19-8
1
2
3
4
5
6
{
"config": {
"bin-dir": "bin",
"vendor-dir": "/some/dir/vendor"
},
}
Then, update the path to the autoload.php file in app/autoload.php:

Listing 19-9
// app/autoload.php
// ...
$loader = require '/some/dir/vendor/autoload.php';
This modification can be of interest if you are working in a virtual environment and cannot use NFS
- for example, if you're running a Symfony application using Vagrant/VirtualBox in a guest operating
system.

Chapter 20
Using Parameters within a Dependency

Injection Class
You have seen how to use configuration parameters within Symfony service containers. There are special
cases such as when you want, for instance, to use the %kernel.debug% parameter to make the services
in your bundle enter debug mode. For this case there is more work to do in order to make the system
understand the parameter value. By default, your parameter %kernel.debug% will be treated as a
simple string. Consider the following example:
Listing 20-1
1
2
3
4
5
6
7
8
9
10
11
// inside Configuration class

$rootNode
->children()
->booleanNode('logging')->defaultValue('%kernel.debug%')->end()
// ...
->end()
;
// inside the Extension class
$config = $this->processConfiguration($configuration, $configs);
var_dump($config['logging']);
Now, examine the results to see this closely:

Listing 20-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
my_bundle:
logging: true
# true, as expected
my_bundle:
logging: '%kernel.debug%'
# true/false (depends on 2nd parameter of AppKernel),
# as expected, because %kernel.debug% inside configuration
# gets evaluated before being passed to the extension
my_bundle: ~
# passes the string "%kernel.debug%".
# Which is always considered as true.
# The Configurator does not know anything about
# "%kernel.debug%" being a parameter.

Chapter 20: Using Parameters within a Dependency Injection Class | 69
In order to support this use case, the Configuration class has to be injected with this parameter via
the extension as follows:
Listing 20-3
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
namespace AppBundle\DependencyInjection;
use Symfony\Component\Config\Definition\Builder\TreeBuilder;
use Symfony\Component\Config\Definition\ConfigurationInterface;
class Configuration implements ConfigurationInterface
{
private $debug;
public function __construct($debug)
{
$this->debug = (bool) $debug;
}
public function getConfigTreeBuilder()
{
$treeBuilder = new TreeBuilder();
$rootNode = $treeBuilder->root('my_bundle');
$rootNode
->children()
// ...
->booleanNode('logging')->defaultValue($this->debug)->end()
// ...
->end()
;
return $treeBuilder;
}
}
And set it in the constructor of Configuration via the Extension class:

Listing 20-4
1
2
3
4
5
6
7
8
9
10
11
12
13
14
namespace AppBundle\DependencyInjection;
class AppExtension extends Extension
{
// ...
public function getConfiguration(array $config, ContainerBuilder $container)
{
return new Configuration($container->getParameter('kernel.debug'));
}
}
Setting the Default in the Extension

There are some instances of %kernel.debug% usage within a Configurator class in TwigBundle
and AsseticBundle. However this is because the default parameter value is set by the Extension class.
For example in AsseticBundle, you can find:
Listing 20-5
$container->setParameter('assetic.debug', $config['debug']);
The string %kernel.debug% passed here as an argument handles the interpreting job to the
container which in turn does the evaluation. Both ways accomplish similar goals. AsseticBundle will
not use %kernel.debug% but rather the new %assetic.debug% parameter.

Chapter 20: Using Parameters within a Dependency Injection Class | 70
Chapter 21
Understanding how the Front Controller,

Kernel and Environments Work together
The section How to Master and Create new Environments explained the basics on how Symfony uses
environments to run your application with different configuration settings. This section will explain a bit
more in-depth what happens when your application is bootstrapped. To hook into this process, you need
to understand three parts that work together:
The Front Controller
The Kernel Class
The Environments
Usually, you will not need to define your own front controller or AppKernel class as the Symfony
Standard Edition1 provides sensible default implementations.
This documentation section is provided to explain what is going on behind the scenes.
The Front Controller

The front controller2 is a well-known design pattern; it is a section of code that all requests served by an
application run through.
In the Symfony Standard Edition3, this role is taken by the app.php4 and app_dev.php5 files in the web/
directory. These are the very first PHP scripts executed when a request is processed.
The main purpose of the front controller is to create an instance of the AppKernel (more on that in a
second), make it handle the request and return the resulting response to the browser.
1. https://github.com/symfony/symfony-standard
2. https://en.wikipedia.org/wiki/Front_Controller_pattern
4. https://github.com/symfony/symfony-standard/blob/master/web/app.php
5. https://github.com/symfony/symfony-standard/blob/master/web/app_dev.php

Chapter 21: Understanding how the Front Controller, Kernel and Environments Work together | 71
Because every request is routed through it, the front controller can be used to perform global initialization
prior to setting up the kernel or to decorate6 the kernel with additional features. Examples include:
Configuring the autoloader or adding additional autoloading mechanisms;

Adding HTTP level caching by wrapping the kernel with an instance of AppCache;
Enabling (or skipping) the ClassCache;
Enabling the Debug Component.
The front controller can be chosen by requesting URLs like:

Listing 21-1
http://localhost/app_dev.php/some/path/...
As you can see, this URL contains the PHP script to be used as the front controller. You can use
that to easily switch the front controller or use a custom one by placing it in the web/ directory (e.g.
app_cache.php).
When using Apache and the RewriteRule shipped with the Symfony Standard Edition7, you can omit the
filename from the URL and the RewriteRule will use app.php as the default one.
Pretty much every other web server should be able to achieve a behavior similar to that of the
RewriteRule described above. Check your server documentation for details or see Configuring a Web
Server.
Make sure you appropriately secure your front controllers against unauthorized access. For example,
you don't want to make a debugging environment available to arbitrary users in your production
environment.
Technically, the app/console8 script used when running Symfony on the command line is also a front
controller, only that is not used for web, but for command line requests.
The Kernel Class

The Kernel9 is the core of Symfony. It is responsible for setting up all the bundles that make up your
application and providing them with the application's configuration. It then creates the service container
before serving requests in its handle()10 method.
There are two methods declared in the KernelInterface11 that are left unimplemented in Kernel12
and thus serve as template methods13:
registerBundles()14
It must return an array of all bundles needed to run the application.

registerContainerConfiguration()15
It loads the application configuration.
6. https://en.wikipedia.org/wiki/Decorator_pattern
7. https://github.com/symfony/symfony-standard/blob/master/web/.htaccess
8. https://github.com/symfony/symfony-standard/blob/master/app/console
9. http://api.symfony.com/2.7/Symfony/Component/HttpKernel/Kernel.html
10. http://api.symfony.com/2.7/Symfony/Component/HttpKernel/HttpKernelInterface.html#method_handle
11. http://api.symfony.com/2.7/Symfony/Component/HttpKernel/KernelInterface.html
12. http://api.symfony.com/2.7/Symfony/Component/HttpKernel/Kernel.html
13. https://en.wikipedia.org/wiki/Template_method_pattern
14. http://api.symfony.com/2.7/Symfony/Component/HttpKernel/KernelInterface.html#method_registerBundles
15. http://api.symfony.com/2.7/Symfony/Component/HttpKernel/KernelInterface.html#method_registerContainerConfiguration

To fill these (small) blanks, your application needs to subclass the Kernel and implement these methods.
The resulting class is conventionally called the AppKernel.
Again, the Symfony Standard Edition provides an AppKernel16 in the app/ directory. This class uses
the name of the environment - which is passed to the Kernel's constructor17 method and is available
via getEnvironment()18 - to decide which bundles to create. The logic for that is in
registerBundles(), a method meant to be extended by you when you start adding bundles to your
application.
You are, of course, free to create your own, alternative or additional AppKernel variants. All you need
is to adapt your (or add a new) front controller to make use of the new kernel.
The name and location of the AppKernel is not fixed. When putting multiple Kernels into a single
application, it might therefore make sense to add additional sub-directories, for example app/
admin/AdminKernel.php and app/api/ApiKernel.php. All that matters is that your front
controller is able to create an instance of the appropriate kernel.
Having different AppKernels might be useful to enable different front controllers (on potentially
different servers) to run parts of your application independently (for example, the admin UI, the frontend UI and database migrations).
There's a lot more the AppKernel can be used for, for example overriding the default directory
structure. But odds are high that you don't need to change things like this on the fly by having several
AppKernel implementations.
The Environments
AppKernel has to implement another method registerContainerConfiguration()19. This method is responsible for loading the application's
As
just
mentioned,
the
configuration from the right environment.

Environments have been covered extensively in the previous chapter, and you probably remember that
the Symfony Standard Edition comes with three of them - dev, prod and test.
More technically, these names are nothing more than strings passed from the front controller to the
AppKernel's
constructor.
This
name
can
then
be
used
in
the
registerContainerConfiguration()20 method to decide which configuration files to load.
The Symfony Standard Edition's AppKernel21 class implements this method by simply loading the app/
config/config_*environment*.yml file. You are, of course, free to implement this method
differently if you need a more sophisticated way of loading your configuration.
16. https://github.com/symfony/symfony-standard/blob/master/app/AppKernel.php
17. http://api.symfony.com/2.7/Symfony/Component/HttpKernel/Kernel.html#method___construct
18. http://api.symfony.com/2.7/Symfony/Component/HttpKernel/Kernel.html#method_getEnvironment
21. https://github.com/symfony/symfony-standard/blob/master/app/AppKernel.php

Chapter 22
How to Set external Parameters in the Service

Container
In the chapter How to Master and Create new Environments, you learned how to manage your application
configuration. At times, it may benefit your application to store certain credentials outside of your project
code. Database configuration is one such example. The flexibility of the Symfony service container allows
you to easily do this.
Environment Variables
Symfony will grab any environment variable prefixed with SYMFONY__ and set it as a parameter in the
service container. Some transformations are applied to the resulting parameter name:
SYMFONY__ prefix is removed;
Parameter name is lowercased;
Double underscores are replaced with a period, as a period is not a valid character in an
environment variable name.
For example, if you're using Apache, environment variables can be set using the following VirtualHost
configuration:
Listing 22-1
1
2
3
4
5
6
7
8
9
10
11
12
<VirtualHost *:80>
ServerName
DocumentRoot
DirectoryIndex
SetEnv
SetEnv
Symfony
"/path/to/symfony_2_app/web"
index.php index.html
SYMFONY__DATABASE__USER user
SYMFONY__DATABASE__PASSWORD secret
<Directory "/path/to/symfony_2_app/web">
AllowOverride All
Allow from All
</Directory>
</VirtualHost>

Chapter 22: How to Set external Parameters in the Service Container | 74
The example above is for an Apache configuration, using the SetEnv1 directive. However, this will
work for any web server which supports the setting of environment variables.
Also, in order for your console to work (which does not use Apache), you must export these as shell
variables. On a Unix system, you can run the following:
Listing 22-2
1
2
$ export SYMFONY__DATABASE__USER=user
$ export SYMFONY__DATABASE__PASSWORD=secret
Now that you have declared an environment variable, it will be present in the PHP $_SERVER global
variable. Symfony then automatically sets all $_SERVER variables prefixed with SYMFONY__ as
parameters in the service container.
You can now reference these parameters wherever you need them.
Listing 22-3
1
2
3
4
5
6
doctrine:
dbal:
driver
dbname:
user:
password:
pdo_mysql
symfony_project
'%database.user%'
'%database.password%'
Constants
The container also has support for setting PHP constants as parameters. See Constants as Parameters for
more details.
Miscellaneous Configuration
The imports directive can be used to pull in parameters stored elsewhere. Importing a PHP file gives
you the flexibility to add whatever is needed in the container. The following imports a file named
parameters.php.
Listing 22-4
1
2
3
imports:
- { resource: parameters.php }
A resource file can be one of many types. PHP, XML, YAML, INI, and closure resources are all
supported by the imports directive.
In parameters.php, tell the service container the parameters that you wish to set. This is useful when
important configuration is in a non-standard format. The example below includes a Drupal database
configuration in the Symfony service container.
Listing 22-5
1
2
3
// app/config/parameters.php
include_once('/path/to/drupal/sites/default/settings.php');
$container->setParameter('drupal.database.url', $db_url);
1. http://httpd.apache.org/docs/current/env.html

Chapter 22: How to Set external Parameters in the Service Container | 75
Chapter 23
How to Use the Apache Router
Using the Apache Router is no longer considered a good practice. The small increase obtained
in the application routing performance is not worth the hassle of continuously updating the routes
configuration.
The Apache Router will be removed in Symfony 3 and it's highly recommended to not use it in your
applications.
Symfony, while fast out of the box, also provides various ways to increase that speed with a little bit of
tweaking. One of these ways is by letting Apache handle routes directly, rather than using Symfony for
this task.
Apache router was deprecated in Symfony 2.5 and will be removed in Symfony 3.0. Since the PHP
implementation of the Router was improved, performance gains were no longer significant (while it's
very hard to replicate the same behavior).
Change Router Configuration Parameters

To dump Apache routes you must first tweak some configuration parameters to tell Symfony to use the
ApacheUrlMatcher instead of the default one:
Listing 23-1
1
2
3
4
# app/config/config_prod.yml
parameters:
router.options.matcher.cache_class: ~ # disable router cache
router.options.matcher_class: Symfony\Component\Routing\Matcher\ApacheUrlMatcher
Note that ApacheUrlMatcher1 extends UrlMatcher2 so even if you don't regenerate the
mod_rewrite rules, everything will work (because at the end of ApacheUrlMatcher::match() a
call to parent::match() is done).
1. http://api.symfony.com/2.7/Symfony/Component/Routing/Matcher/ApacheUrlMatcher.html
2. http://api.symfony.com/2.7/Symfony/Component/Routing/Matcher/UrlMatcher.html

Chapter 23: How to Use the Apache Router | 76
Generating mod_rewrite Rules

To test that it's working, create a very basic route for the AppBundle:
Listing 23-2
1
2
3
4
# app/config/routing.yml
hello:
path: /hello/{name}
defaults: { _controller: AppBundle:Greet:hello }
Now generate the mod_rewrite rules:

Listing 23-3
$ php app/console router:dump-apache -e=prod --no-debug
Which should roughly output the following:

Listing 23-4
1
2
3
4
5
6
7
# skip "real" requests

RewriteCond %{REQUEST_FILENAME} -f
RewriteRule .* - [QSA,L]
# hello
RewriteCond %{REQUEST_URI} ^/hello/([^/]+?)$
RewriteRule .* app.php
[QSA,L,E=_ROUTING__route:hello,E=_ROUTING_name:%1,E=_ROUTING__controller:AppBundle\:Greet\:hello]
You can now rewrite web/.htaccess to use the new rules, so with this example it should look like this:
Listing 23-5
1
2
3
4
5
6
7
8
9
10
11
<IfModule mod_rewrite.c>
RewriteEngine On
# skip "real" requests

RewriteCond %{REQUEST_FILENAME} -f
RewriteRule .* - [QSA,L]
# hello
RewriteCond %{REQUEST_URI} ^/hello/([^/]+?)$
RewriteRule .* app.php
[QSA,L,E=_ROUTING__route:hello,E=_ROUTING_name:%1,E=_ROUTING__controller:AppBundle\:Greet\:hello]
</IfModule>
The procedure above should be done each time you add/change a route if you want to take full
advantage of this setup.
That's it! You're now all set to use Apache routes.
Additional Tweaks
To save some processing time, change occurrences of Request to ApacheRequest in web/app.php:
Listing 23-6
1
2
3
4
5
6
7
8
9
10
// web/app.php
require_once __DIR__.'/../app/bootstrap.php.cache';
require_once __DIR__.'/../app/AppKernel.php';
// require_once __DIR__.'/../app/AppCache.php';
use Symfony\Component\HttpFoundation\ApacheRequest;
$kernel = new AppKernel('prod', false);
$kernel->loadClassCache();

11
12
// $kernel = new AppCache($kernel);

$kernel->handle(ApacheRequest::createFromGlobals())->send();

Chapter 24
Configuring a Web Server

The preferred way to develop your Symfony application is to use PHP's internal web server. However,
when using an older PHP version or when running the application in the production environment, you'll
need to use a fully-featured web server. This article describes several ways to use Symfony with Apache
or Nginx.
When using Apache, you can configure PHP as an Apache module or with FastCGI using PHP FPM.
FastCGI also is the preferred way to use PHP with Nginx.
The Web Directory

The web directory is the home of all of your application's public and static files, including images,
stylesheets and JavaScript files. It is also where the front controllers (app.php and app_dev.php)
live.
The web directory serves as the document root when configuring your web server. In the examples
below, the web/ directory will be the document root. This directory is /var/www/project/web/.
If your hosting provider requires you to change the web/ directory to another location (e.g.
public_html/) make sure you override the location of the web/ directory.
Apache with mod_php/PHP-CGI

The minimum configuration to get your application running under Apache is:
Listing 24-1
1
2
3
4
5
6
7
8
9
10
11
12
<VirtualHost *:80>
ServerName domain.tld
ServerAlias www.domain.tld
DocumentRoot /var/www/project/web
<Directory /var/www/project/web>
AllowOverride All
Order Allow,Deny
Allow from All
</Directory>
# uncomment the following lines if you install assets as symlinks

Chapter 24: Configuring a Web Server | 79
13
14
15
16
17
18
19
20
# or run into problems when compiling LESS/Sass/CoffeScript assets

# <Directory /var/www/project>
#
Options FollowSymlinks
# </Directory>
ErrorLog /var/log/apache2/project_error.log
CustomLog /var/log/apache2/project_access.log combined
</VirtualHost>
your system supports the APACHE_LOG_DIR variable, you may want to use
${APACHE_LOG_DIR}/ instead of hardcoding /var/log/apache2/.
If
Use the following optimized configuration to disable .htaccess support and increase web server
performance:
Listing 24-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
<VirtualHost *:80>
AllowOverride None
Order Allow,Deny
Allow from All
Options -MultiViews
RewriteEngine On
RewriteCond %{REQUEST_FILENAME} !-f
RewriteRule ^(.*)$ app.php [QSA,L]
</IfModule>
</Directory>
#
#
#
#
#
uncomment the following lines if you install assets as symlinks

or run into problems when compiling LESS/Sass/CoffeScript assets
<Directory /var/www/project>
</Directory>
# optionally disable the RewriteEngine for the asset directories

# which will allow apache to simply reply with a 404 when files are
# not found instead of passing the request into the full symfony stack
<Directory /var/www/project/web/bundles>
RewriteEngine Off
</IfModule>
</Directory>
</VirtualHost>
If you are using php-cgi, Apache does not pass HTTP basic username and password to PHP by
default. To work around this limitation, you should use the following configuration snippet:
Listing 24-3
RewriteRule .* - [E=HTTP_AUTHORIZATION:%{HTTP:Authorization}]
Using mod_php/PHP-CGI with Apache 2.4

In Apache 2.4, Order Allow,Deny has been replaced by Require all granted. Hence, you need
to modify your Directory permission settings as follows:

Listing 24-4
1
2
3
4
Require all granted
# ...
</Directory>
For advanced Apache configuration options, read the official Apache documentation1.
Apache with PHP-FPM

To make use of PHP5-FPM with Apache, you first have to ensure that you have the FastCGI process
manager php-fpm binary and Apache's FastCGI module installed (for example, on a Debian based
system you have to install the libapache2-mod-fastcgi and php5-fpm packages).
PHP-FPM uses so-called pools to handle incoming FastCGI requests. You can configure an arbitrary
number of pools in the FPM configuration. In a pool you configure either a TCP socket (IP and port) or
a Unix domain socket to listen on. Each pool can also be run under a different UID and GID:
Listing 24-5
1
2
3
4
5
6
7
8
9
10
; a pool called www

[www]
user = www-data
group = www-data
; use a unix domain socket
listen = /var/run/php5-fpm.sock
; or listen on a TCP socket
listen = 127.0.0.1:9000
Using mod_proxy_fcgi with Apache 2.4

If you are running Apache 2.4, you can easily use mod_proxy_fcgi to pass incoming requests to
PHP-FPM. Configure PHP-FPM to listen on a TCP socket (mod_proxy currently does not support
Unix sockets2), enable mod_proxy and mod_proxy_fcgi in your Apache configuration and use the
SetHandler directive to pass requests for PHP files to PHP FPM:
Listing 24-6
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
<VirtualHost *:80>
# Uncomment the following line to force Apache to pass the Authorization

# header to PHP: required for "basic_auth" under PHP-FPM and FastCGI
#
# SetEnvIfNoCase Âuthorization$ "(.+)" HTTP_AUTHORIZATION=$1
# For Apache 2.4.9 or higher
# Using SetHandler avoids issues with using ProxyPassMatch in combination
# with mod_rewrite or mod_autoindex
<FilesMatch \.php$>
SetHandler proxy:fcgi://127.0.0.1:9000
</FilesMatch>
# If you use Apache version below 2.4.9 you must consider update or use this instead
# ProxyPassMatch ^/(.*\.php(/.*)?)$ fcgi://127.0.0.1:9000/var/www/project/web/$1
# If you run your Symfony application on a subpath of your document root, the
# regular expression must be changed accordingly:
# ProxyPassMatch ^/path-to-app/(.*\.php(/.*)?)$ fcgi://127.0.0.1:9000/var/www/project/web/$1
1. http://httpd.apache.org/docs/
2. https://bz.apache.org/bugzilla/show_bug.cgi?id=54101

23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
# enable the .htaccess rewrites
AllowOverride All
Require all granted
</Directory>
#
#
#
#
#

</Directory>
</VirtualHost>
PHP-FPM with Apache 2.2

On Apache 2.2 or lower, you cannot use mod_proxy_fcgi. You have to use the FastCgiExternalServer3
directive instead. Therefore, your Apache configuration should look something like this:
Listing 24-7
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
<VirtualHost *:80>
AddHandler php5-fcgi .php
Action php5-fcgi /php5-fcgi
Alias /php5-fcgi /usr/lib/cgi-bin/php5-fcgi
FastCgiExternalServer /usr/lib/cgi-bin/php5-fcgi -host 127.0.0.1:9000 -pass-header Authorization
# enable the .htaccess rewrites
AllowOverride All
Order Allow,Deny
Allow from all
</Directory>
#
#
#
#
#

</Directory>
</VirtualHost>
If you prefer to use a Unix socket, you have to use the -socket option instead:
Listing 24-8
FastCgiExternalServer /usr/lib/cgi-bin/php5-fcgi -socket /var/run/php5-fpm.sock -pass-header Authorization
Nginx
The minimum configuration to get your application running under Nginx is:
Listing 24-9
1
2
server {
server_name domain.tld www.domain.tld;
3. http://www.fastcgi.com/mod_fastcgi/docs/mod_fastcgi.html#FastCgiExternalServer

3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
root /var/www/project/web;
location / {
# try to serve file directly, fallback to app.php
try_files $uri /app.php$is_args$args;
}
# DEV
# This rule should only be placed on your development environment
# In production, don't include this and don't deploy app_dev.php or config.php
location ~ ^/(app_dev|config)\.php(/|$) {
fastcgi_pass unix:/var/run/php5-fpm.sock;
fastcgi_split_path_info ^(.+\.php)(/.*)$;
include fastcgi_params;
# When you are using symlinks to link the document root to the
# current version of your application, you should pass the real
# application path instead of the path to the symlink to PHP
# FPM.
# Otherwise, PHP's OPcache may not properly detect changes to
# your PHP files (see https://github.com/zendtech/ZendOptimizerPlus/issues/126
# for more information).
fastcgi_param SCRIPT_FILENAME $realpath_root$fastcgi_script_name;
fastcgi_param DOCUMENT_ROOT $realpath_root;
}
# PROD
location ~ ^/app\.php(/|$) {
fastcgi_pass unix:/var/run/php5-fpm.sock;
fastcgi_split_path_info ^(.+\.php)(/.*)$;
include fastcgi_params;
# When you are using symlinks to link the document root to the
# current version of your application, you should pass the real
# application path instead of the path to the symlink to PHP
# FPM.
# Otherwise, PHP's OPcache may not properly detect changes to
# your PHP files (see https://github.com/zendtech/ZendOptimizerPlus/issues/126
# for more information).
fastcgi_param SCRIPT_FILENAME $realpath_root$fastcgi_script_name;
fastcgi_param DOCUMENT_ROOT $realpath_root;
# Prevents URIs that include the front controller. This will 404:
# http://domain.tld/app.php/some-path
# Remove the internal directive to allow URIs like this
internal;
}
# return 404 for all other php files not matching the front controller
# this prevents access to other php files you don't want to be accessible.
location ~ \.php$ {
return 404;
}
error_log /var/log/nginx/project_error.log;
access_log /var/log/nginx/project_access.log;
}
Depending on your PHP-FPM config, the fastcgi_pass can also be fastcgi_pass

127.0.0.1:9000.
This executes only app.php, app_dev.php and config.php in the web directory. All other files
ending in ".php" will be denied.
If you have other PHP files in your web directory that need to be executed, be sure to include them
in the location block above.

After you deploy to production, make sure that you cannot access the app_dev.php or
config.php scripts (i.e. http://example.com/app_dev.php and http://example.com/
config.php). If you can access these, be sure to remove the DEV section from the above
configuration.
For advanced Nginx configuration options, read the official Nginx documentation4.
4. http://wiki.nginx.org/Symfony

Chapter 25
How to Organize Configuration Files

The default Symfony Standard Edition defines three execution environments called dev, prod and test.
An environment simply represents a way to execute the same codebase with different configurations.
In order to select the configuration file to load for each environment, Symfony executes the
registerContainerConfiguration() method of the AppKernel class:
Listing 25-1
1
2
3
4
5
6
7
8
9
10
11
12
13
use Symfony\Component\HttpKernel\Kernel;
use Symfony\Component\Config\Loader\LoaderInterface;
{
// ...
{
$loader->load($this->getRootDir().'/config/config_'.$this->getEnvironment().'.yml');
}
}
This method loads the app/config/config_dev.yml file for the dev environment and so on. In
turn, this file loads the common configuration file located at app/config/config.yml. Therefore,
the configuration files of the default Symfony Standard Edition follow this structure:
Listing 25-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
<your-project>/
app/
config/
config.yml
config_dev.yml
config_prod.yml
config_test.yml
parameters.yml
parameters.yml.dist
routing.yml
routing_dev.yml
security.yml
src/
vendor/
web/

Chapter 25: How to Organize Configuration Files | 85
This default structure was chosen for its simplicity one file per environment. But as any other Symfony
feature, you can customize it to better suit your needs. The following sections explain different ways to
organize your configuration files. In order to simplify the examples, only the dev and prod environments
are taken into account.
Different Directories per Environment

Instead of suffixing the files with _dev and _prod, this technique groups all the related configuration
files under a directory with the same name as the environment:
Listing 25-3
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
<your-project>/
app/
config/
common/
config.yml
parameters.yml
routing.yml
security.yml
dev/
config.yml
parameters.yml
routing.yml
security.yml
prod/
config.yml
parameters.yml
routing.yml
security.yml
src/
vendor/
web/
To make this work, change the code of the registerContainerConfiguration()1 method:

Listing 25-4
1
2
3
4
5
6
7
8
9
10
11
12
13
{
// ...
{
$loader->load($this->getRootDir().'/config/'.$this->getEnvironment().'/config.yml');
}
}
Then, make sure that each config.yml file loads the rest of the configuration files, including the
common files. For instance, this would be the imports needed for the app/config/dev/config.yml
file:
Listing 25-5
1
2
3
4
5
6
7
# app/config/dev/config.yml
imports:
- { resource: '../common/config.yml' }
- { resource: 'parameters.yml' }
- { resource: 'security.yml' }
# ...

Due to the way in which parameters are resolved, you cannot use them to build paths in imports
dynamically. This means that something like the following doesn't work:
Listing 25-6
1
2
3
imports:
- { resource: '%kernel.root_dir%/parameters.yml' }
Semantic Configuration Files

A different organization strategy may be needed for complex applications with large configuration files.
For instance, you could create one file per bundle and several files to define all application services:
Listing 25-7
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
<your-project>/
app/
config/
bundles/
bundle1.yml
bundle2.yml
...
bundleN.yml
environments/
common.yml
dev.yml
prod.yml
routing/
common.yml
dev.yml
prod.yml
services/
frontend.yml
backend.yml
...
security.yml
src/
vendor/
web/
Again, change the code of the registerContainerConfiguration() method to make Symfony

aware of the new file organization:
Listing 25-8
1
2
3
4
5
6
7
8
9
10
11
12
13
{
// ...
{
$loader->load($this->getRootDir().'/config/environments/'.$this->getEnvironment().'.yml');
}
}
Following the same technique explained in the previous section, make sure to import the appropriate
configuration files from each main file (common.yml, dev.yml and prod.yml).
Advanced Techniques
Symfony loads configuration files using the Config component, which provides some advanced features.
Mix and Match Configuration Formats

Configuration files can import files defined with any other built-in configuration format (.yml, .xml,
.php, .ini):
Listing 25-9
1
2
3
4
5
6
7
8
imports:
- { resource: 'services.xml' }
- { resource: 'security.yml' }
- { resource: 'legacy.php' }
# ...
The IniFileLoader parses the file contents using the parse_ini_file2 function. Therefore,
you can only set parameters to string values. Use one of the other loaders if you want to use other
data types (e.g. boolean, integer, etc.).
If you use any other configuration format, you have to define your own loader class extending it from
FileLoader3. When the configuration values are dynamic, you can use the PHP configuration file
to execute your own logic. In addition, you can define your own services to load configurations from
databases or web services.
Global Configuration Files

Some system administrators may prefer to store sensitive parameters in files outside the project directory.
Imagine that the database credentials for your website are stored in the /etc/sites/mysite.com/
parameters.yml file. Loading this file is as simple as indicating the full file path when importing it
from any other configuration file:
Listing 25-10
1
2
3
4
5
6
imports:
- { resource: '/etc/sites/mysite.com/parameters.yml' }
# ...
Most of the time, local developers won't have the same files that exist on the production servers. For that
reason, the Config component provides the ignore_errors option to silently discard errors when the
loaded file doesn't exist:
Listing 25-11
1
2
3
4
5
6
imports:
- { resource: '/etc/sites/mysite.com/parameters.yml', ignore_errors: true }
# ...
As you've seen, there are lots of ways to organize your configuration files. You can choose one of these or
even create your own custom way of organizing the files. Don't feel limited by the Standard Edition that
comes with Symfony. For even more customization, see "How to Override Symfony's default Directory
Structure".
2. http://php.net/manual/en/function.parse-ini-file.php
3. http://api.symfony.com/2.7/Symfony/Component/DependencyInjection/Loader/FileLoader.html

Chapter 26
How to Create a Console Command

The Console page of the Components section (The Console Component) covers how to create a console
command. This cookbook article covers the differences when creating console commands within the
Symfony Framework.
Automatically Registering Commands

To make the console commands available automatically with Symfony, create a Command directory
inside your bundle and create a PHP file suffixed with Command.php for each command that you want
to provide. For example, if you want to extend the AppBundle to greet you from the command line,
create GreetCommand.php and add the following to it:
Listing 26-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
// src/AppBundle/Command/GreetCommand.php
namespace AppBundle\Command;
use
use
use
use
use
Symfony\Bundle\FrameworkBundle\Command\ContainerAwareCommand;
Symfony\Component\Console\Input\InputArgument;
Symfony\Component\Console\Input\InputInterface;
Symfony\Component\Console\Input\InputOption;
Symfony\Component\Console\Output\OutputInterface;
class GreetCommand extends ContainerAwareCommand

{
protected function configure()
{
$this
->setName('demo:greet')
->setDescription('Greet someone')
->addArgument(
'name',
InputArgument::OPTIONAL,
'Who do you want to greet?'
)
->addOption(
'yell',
null,
InputOption::VALUE_NONE,
'If set, the task will yell in uppercase letters'
)

Chapter 26: How to Create a Console Command | 89
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
;
}
protected function execute(InputInterface $input, OutputInterface $output)
{
$name = $input->getArgument('name');
if ($name) {
$text = 'Hello '.$name;
} else {
$text = 'Hello';
}
if ($input->getOption('yell')) {
$text = strtoupper($text);
}
$output->writeln($text);
}
}
This command will now automatically be available to run:

Listing 26-2
$ php app/console demo:greet Fabien
Register Commands in the Service Container

Just like controllers, commands can be declared as services. See the dedicated cookbook entry for details.
Getting Services from the Service Container

ContainerAwareCommand1 as the base class for the command (instead of the more basic
Command2), you have access to the service container. In other words, you have access to any configured
By using
service:
Listing 26-3
1
2
3
4
5
6
7
8

{
$logger = $this->getContainer()->get('logger');
$logger->info('Executing command for '.$name);
// ...
}
Invoking other Commands

See Calling an Existing Command if you need to implement a command that runs other dependent
commands.
1. http://api.symfony.com/2.7/Symfony/Bundle/FrameworkBundle/Command/ContainerAwareCommand.html
2. http://api.symfony.com/2.7/Symfony/Component/Console/Command/Command.html

Testing Commands
When
testing
commands
used
as
part
of
the
full-stack
framework,
Symfony\Bundle\FrameworkBundle\Console\Application3 should be used instead of

Symfony\Component\Console\Application4:
Listing 26-4
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
use Symfony\Component\Console\Tester\CommandTester;
use Symfony\Bundle\FrameworkBundle\Console\Application;
use AppBundle\Command\GreetCommand;
class ListCommandTest extends \PHPUnit_Framework_TestCase
{
public function testExecute()
{
// mock the Kernel or create one depending on your needs
$application = new Application($kernel);
$application->add(new GreetCommand());
$command = $application->find('demo:greet');
$commandTester = new CommandTester($command);
$commandTester->execute(
array(
'name'
=> 'Fabien',
'--yell' => true,
)
);
$this->assertRegExp('/.../', $commandTester->getDisplay());
// ...
}
}
In the specific case above, the name parameter and the --yell option are not mandatory for the
command to work, but are shown so you can see how to customize them when calling the command.
To be able to use the fully set up service container for your console tests you can extend your test from
KernelTestCase5:
Listing 26-5
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
use
use
use
use
Symfony\Component\Console\Tester\CommandTester;
Symfony\Bundle\FrameworkBundle\Console\Application;
Symfony\Bundle\FrameworkBundle\Test\KernelTestCase;
AppBundle\Command\GreetCommand;
class ListCommandTest extends KernelTestCase

{
public function testExecute()
{
$kernel = $this->createKernel();
$kernel->boot();
$application->add(new GreetCommand());
$command = $application->find('demo:greet');
$commandTester = new CommandTester($command);
$commandTester->execute(
array(
'name'
=> 'Fabien',
'--yell' => true,
3. http://api.symfony.com/2.7/Symfony/Bundle/FrameworkBundle/Console/Application.html
4. http://api.symfony.com/2.7/Symfony/Component/Console/Application.html
5. http://api.symfony.com/2.7/Symfony/Bundle/FrameworkBundle/Test/KernelTestCase.html

22
23
24
25
26
27
28
29
)
);
$this->assertRegExp('/.../', $commandTester->getDisplay());
// ...
}
}

Chapter 27
How to Use the Console

The Using Console Commands, Shortcuts and Built-in Commands page of the components documentation
looks at the global console options. When you use the console as part of the full-stack framework, some
additional global options are available as well.
By default, console commands run in the dev environment and you may want to change this for
some commands. For example, you may want to run some commands in the prod environment for
performance reasons. Also, the result of some commands will be different depending on the environment.
For example, the cache:clear command will clear and warm the cache for the specified environment
only. To clear and warm the prod cache you need to run:
Listing 27-1
$ php app/console cache:clear --env=prod
or the equivalent:
Listing 27-2
$ php app/console cache:clear -e prod
In addition to changing the environment, you can also choose to disable debug mode. This can be useful
where you want to run commands in the dev environment but avoid the performance hit of collecting
debug data:
Listing 27-3
$ php app/console list --no-debug
There is an interactive shell which allows you to enter commands without having to specify php app/
console each time, which is useful if you need to run several commands. To enter the shell run:
Listing 27-4
1
2
$ php app/console --shell

$ php app/console -s
You can now just run commands with the command name:
Listing 27-5
Symfony > list
When using the shell you can choose to run each command in a separate process:
Listing 27-6

Chapter 27: How to Use the Console | 93
1
2
$ php app/console --shell --process-isolation

$ php app/console -s --process-isolation
When you do this, the output will not be colorized and interactivity is not supported so you will need to
pass all command parameters explicitly.
Unless you are using isolated processes, clearing the cache in the shell will not have an effect on
subsequent commands you run. This is because the original cached files are still being used.

Chapter 27: How to Use the Console | 94
Chapter 28
How to Style a Console Command

New in version 2.7: Symfony Styles for console commands were introduced in Symfony 2.7.
One of the most boring tasks when creating console commands is to deal with the styling of the
command's input and output. Displaying titles and tables or asking questions to the user involves a lot of
repetitive code.
Consider for example the code used to display the title of the following command:
Listing 28-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
use Symfony\Bundle\FrameworkBundle\Command\ContainerAwareCommand;
use Symfony\Component\Console\Input\InputInterface;
use Symfony\Component\Console\Output\OutputInterface;
{
// ...
{
$output->writeln(array(
'<info>Lorem Ipsum Dolor Sit Amet</>',
'<info>==========================</>',
'',
));
// ...
}
}
Displaying a simple title requires three lines of code, to change the font color, underline the contents and
leave an additional blank line after the title. Dealing with styles is required for well-designed commands,
but it complicates their code unnecessarily.
In order to reduce that boilerplate code, Symfony commands can optionally use the Symfony Style
Guide. These styles are implemented as a set of helper methods which allow to create semantic
commands and forget about their styling.

Chapter 28: How to Style a Console Command | 95
Basic Usage
In your command, instantiate the SymfonyStyle1 class and pass the $input and $output variables
as its arguments. Then, you can start using any of its helpers, such as title(), which displays the title
of the command:
Listing 28-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
use
use
use
use
Symfony\Component\Console\Style\SymfonyStyle;

{
// ...
{
$io = new SymfonyStyle($input, $output);
$io->title('Lorem Ipsum Dolor Sit Amet');
// ...
}
}
Helper Methods
The SymfonyStyle2 class defines some helper methods that cover the most common interactions
performed by console commands.
Titling Methods
title()3
It displays the given string as the command title. This method is meant to be used only once in a
given command, but nothing prevents you to use it repeatedly:
Listing 28-3
$io->title('Lorem ipsum dolor sit amet');
section()4
It displays the given string as the title of some command section. This is only needed in complex
commands which want to better separate their contents:
Listing 28-4
1
2
3
4
5
6
7
$io->section('Adding a User');
// ...
$io->section('Generating the Password');
// ...
1. http://api.symfony.com/2.7/Symfony/Component/Console/Style/SymfonyStyle.html
2. http://api.symfony.com/2.7/Symfony/Component/Console/Style/SymfonyStyle.html
3. http://api.symfony.com/2.7/Symfony/Component/Console/Style/SymfonyStyle.html#method_title
4. http://api.symfony.com/2.7/Symfony/Component/Console/Style/SymfonyStyle.html#method_section

Content Methods
text()5
It displays the given string or array of strings as regular text. This is useful to render help messages
and instructions for the user running the command:
Listing 28-5
// use simple strings for short messages

$io->text('Lorem ipsum dolor sit amet');
1
2
3
4
5
6
7
8
9
10
11
// ...
// consider using arrays when displaying long messages
$io->text(array(
'Lorem ipsum dolor sit amet',
'Consectetur adipiscing elit',
'Aenean sit amet arcu vitae sem faucibus porta',
));
listing()6
It displays an unordered list of elements passed as an array:

Listing 28-6
1
2
3
4
5
$io->listing(array(
'Element #1 Lorem ipsum dolor sit amet',
));
table()7
It displays the given array of headers and rows as a compact table:

Listing 28-7
1
2
3
4
5
6
7
8
$io->table(
array('Header 1', 'Header 2'),
array(
array('Cell 1-1', 'Cell 1-2'),
)
);
newLine()8
It displays a blank line in the command output. Although it may seem useful, most of the times you
won't need it at all. The reason is that every helper already adds their own blank lines, so you don't
have to care about the vertical spacing:
Listing 28-8
5.
6.
7.
8.
1
2
3
4
5
// outputs a single blank line

$io->newLine();
// outputs three consecutive blank lines
$io->newLine(3);
http://api.symfony.com/2.7/Symfony/Component/Console/Style/SymfonyStyle.html#method_text
http://api.symfony.com/2.7/Symfony/Component/Console/Style/SymfonyStyle.html#method_listing
http://api.symfony.com/2.7/Symfony/Component/Console/Style/SymfonyStyle.html#method_table
http://api.symfony.com/2.7/Symfony/Component/Console/Style/SymfonyStyle.html#method_newLine

Admonition Methods
note()9
It displays the given string or array of strings as a highlighted admonition. Use this helper sparingly
to avoid cluttering command's output:
Listing 28-9
1
2
3
4
5
6
7
8
9
10
11
// use simple strings for short notes

$io->note('Lorem ipsum dolor sit amet');
// ...
// consider using arrays when displaying long notes
$io->note(array(
));
caution()10
Similar to the note() helper, but the contents are more prominently highlighted. The resulting
contents resemble an error message, so you should avoid using this helper unless strictly necessary:
Listing 28-10
1
2
3
4
5
6
7
8
9
10
11
// use simple strings for short caution message

$io->caution('Lorem ipsum dolor sit amet');
// ...
// consider using arrays when displaying long caution messages
$io->caution(array(
));
Progress Bar Methods

progressStart()11
It displays a progress bar with a number of steps equal to the argument passed to the method (don't
pass any value if the length of the progress bar is unknown):
Listing 28-11
1
2
3
4
5
// displays a progress bar of unknown length

$io->progressStart();
// displays a 100-step length progress bar
$io->progressStart(100);
progressAdvance()12
It makes the progress bar advance the given number of steps (or 1 step if no argument is passed):
Listing 28-12
9.
10.
11.
12.
1
2
3
4
5
// advances the progress bar 1 step

$io->progressAdvance();
// advances the progress bar 10 steps
$io->progressAdvance(10);
http://api.symfony.com/2.7/Symfony/Component/Console/Style/SymfonyStyle.html#method_note
http://api.symfony.com/2.7/Symfony/Component/Console/Style/SymfonyStyle.html#method_caution
http://api.symfony.com/2.7/Symfony/Component/Console/Style/SymfonyStyle.html#method_progressStart
http://api.symfony.com/2.7/Symfony/Component/Console/Style/SymfonyStyle.html#method_progressAdvance

progressFinish()13
It finishes the progress bar (filling up all the remaining steps when its length is known):
Listing 28-13
$io->progressFinish();
User Input Methods

ask()14
It asks the user to provide some value:

Listing 28-14
$io->ask('What is your name?');
You can pass the default value as the second argument so the user can simply hit the <Enter> key
to select that value:
Listing 28-15
$io->ask('Where are you from?', 'United States');
In case you need to validate the given value, pass a callback validator as the third argument:
Listing 28-16
1
2
3
4
5
6
7
$io->ask('Number of workers to start', 1, function ($number) {

if (!is_integer($number)) {
throw new \RuntimeException('You must type an integer.');
}
return $number;
});
askHidden()15
It's very similar to the ask() method but the user's input will be hidden and it cannot define a
default value. Use it when asking for sensitive information:
Listing 28-17
1
2
3
4
5
6
7
8
9
10
$io->askHidden('What is your password?');
// validates the given answer

$io->askHidden('What is your password?', function ($password) {
if (empty($password)) {
throw new \RuntimeException('Password cannot be empty.');
}
return $password;
});
confirm()16
It asks a Yes/No question to the user and it only returns true or false:
Listing 28-18
$io->confirm('Restart the web server?');
You can pass the default value as the second argument so the user can simply hit the <Enter> key
to select that value:
Listing 28-19
$io->confirm('Restart the web server?', true);
choice()17
It asks a question whose answer is constrained to the given list of valid answers:
Listing 28-20
13.
14.
15.
16.
17.
$io->choice('Select the queue to analyze', array('queue1', 'queue2', 'queue3'));
http://api.symfony.com/2.7/Symfony/Component/Console/Style/SymfonyStyle.html#method_progressFinish
http://api.symfony.com/2.7/Symfony/Component/Console/Style/SymfonyStyle.html#method_ask
http://api.symfony.com/2.7/Symfony/Component/Console/Style/SymfonyStyle.html#method_askHidden
http://api.symfony.com/2.7/Symfony/Component/Console/Style/SymfonyStyle.html#method_confirm
http://api.symfony.com/2.7/Symfony/Component/Console/Style/SymfonyStyle.html#method_choice

You can pass the default value as the third argument so the user can simply hit the <Enter> key to
select that value:
Listing 28-21
$io->choice('Select the queue to analyze', array('queue1', 'queue2', 'queue3'), 'queue1');
Result Methods
success()18
It displays the given string or array of strings highlighted as a successful message (with a green
background and the [OK] label). It's meant to be used once to display the final result of executing
the given command, but you can use it repeatedly during the execution of the command:
Listing 28-22
1
2
3
4
5
6
7
8
9
10
// use simple strings for short success messages

$io->success('Lorem ipsum dolor sit amet');
// ...
// consider using arrays when displaying long success messages
$io->success(array(
));
warning()19
It displays the given string or array of strings highlighted as a warning message (with a red
background and the [WARNING] label). It's meant to be used once to display the final result of
executing the given command, but you can use it repeatedly during the execution of the command:
Listing 28-23
1
2
3
4
5
6
7
8
9
10
// use simple strings for short warning messages

$io->warning('Lorem ipsum dolor sit amet');
// ...
// consider using arrays when displaying long warning messages
$io->warning(array(
));
error()20
It displays the given string or array of strings highlighted as an error message (with a red background
and the [ERROR] label). It's meant to be used once to display the final result of executing the given
command, but you can use it repeatedly during the execution of the command:
Listing 28-24
1
2
3
4
5
6
7
8
9
10
// use simple strings for short error messages

$io->error('Lorem ipsum dolor sit amet');
// ...
// consider using arrays when displaying long error messages
$io->error(array(
));
18. http://api.symfony.com/2.7/Symfony/Component/Console/Style/SymfonyStyle.html#method_success
19. http://api.symfony.com/2.7/Symfony/Component/Console/Style/SymfonyStyle.html#method_warning
20. http://api.symfony.com/2.7/Symfony/Component/Console/Style/SymfonyStyle.html#method_error

Defining your Own Styles

If you don't like the design of the commands that use the Symfony Style, you can define your own set of
console styles. Just create a class that implements the StyleInterface21:
Listing 28-25
1
2
3
4
5
6
7
8
namespace AppBundle\Console;
use Symfony\Component\Console\Style\StyleInterface;
class CustomStyle implements StyleInterface
{
// ...implement the methods of the interface
}
Then, instantiate this custom class instead of the default SymfonyStyle in your commands. Thanks
to the StyleInterface you won't need to change the code of your commands to change their
appearance:
Listing 28-26
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
namespace AppBundle\Console;
use
use
use
use
AppBundle\Console\CustomStyle;
Symfony\Component\Console\Style\SymfonyStyle;

{
// ...
{
// Before
// $io = new SymfonyStyle($input, $output);
// After
$io = new CustomStyle($input, $output);
// ...
}
}
21. http://api.symfony.com/2.7/Symfony/Component/Console/Style/StyleInterface.html

Chapter 29
How to Call a Command from a Controller

The Console component documentation covers how to create a console command. This cookbook article
covers how to use a console command directly from your controller.
You may have the need to execute some function that is only available in a console command. Usually,
you should refactor the command and move some logic into a service that can be reused in the controller.
However, when the command is part of a third-party library, you wouldn't want to modify or duplicate
their code. Instead, you can execute the command directly.
In comparison with a direct call from the console, calling a command from a controller has a slight
performance impact because of the request stack overhead.
Imagine you want to send spooled Swift Mailer messages by using the swiftmailer:spool:send command.
Run this command from inside your controller via:
Listing 29-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
// src/AppBundle/Controller/SpoolController.php
namespace AppBundle\Controller;
use
use
use
use
use
Symfony\Bundle\FrameworkBundle\Console\Application;
Symfony\Bundle\FrameworkBundle\Controller\Controller;
Symfony\Component\Console\Input\ArrayInput;
Symfony\Component\Console\Output\BufferedOutput;
Symfony\Component\HttpFoundation\Response;
class SpoolController extends Controller

{
public function sendSpoolAction($messages = 10)
{
$kernel = $this->get('kernel');
$application->setAutoExit(false);
$input = new ArrayInput(array(
'command' => 'swiftmailer:spool:send',
'--message-limit' => $messages,
));
// You can use NullOutput() if you don't need the output
$output = new BufferedOutput();
$application->run($input, $output);

Chapter 29: How to Call a Command from a Controller | 102
26
27
28
29
30
31
32
// return the output, don't use if you used NullOutput()

$content = $output->fetch();
// return new Response(""), if you used NullOutput()
return new Response($content);
}
}
Showing Colorized Command Output

By telling the BufferedOutput it is decorated via the second parameter, it will return the Ansi colorcoded content. The SensioLabs AnsiToHtml converter1 can be used to convert this to colorful HTML.
First, require the package:
Listing 29-2
$ composer require sensiolabs/ansi-to-html
Now, use it in your controller:

Listing 29-3
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
// src/AppBundle/Controller/SpoolController.php
use SensioLabs\AnsiConverter\AnsiToHtmlConverter;
use Symfony\Component\Console\Output\BufferedOutput;
use Symfony\Component\HttpFoundation\Response;
// ...
class SpoolController extends Controller
{
public function sendSpoolAction($messages = 10)
{
// ...
$output = new BufferedOutput(
OutputInterface::VERBOSITY_NORMAL,
true // true for decorated
);
// ...
// return the output

$converter = new AnsiToHtmlConverter();
$content = $output->fetch();
return new Response($converter->convert($content));
}
}
The AnsiToHtmlConverter can also be registered as a Twig Extension2, and supports optional
themes.
1. https://github.com/sensiolabs/ansi-to-html
2. https://github.com/sensiolabs/ansi-to-html#twig-integration

Chapter 29: How to Call a Command from a Controller | 103
Chapter 30
How to Generate URLs from the Console

Unfortunately, the command line context does not know about your VirtualHost or domain name.
This means that if you generate absolute URLs within a console command you'll probably end up with
something like http://localhost/foo/bar which is not very useful.
To fix this, you need to configure the "request context", which is a fancy way of saying that you need to
configure your environment so that it knows what URL it should use when generating URLs.
There are two ways of configuring the request context: at the application level and per Command.
Configuring the Request Context Globally

To configure the Request Context - which is used by the URL Generator - you can redefine the
parameters it uses as default values to change the default host (localhost) and scheme (http). You can
also configure the base path if Symfony is not running in the root directory.
Note that this does not impact URLs generated via normal web requests, since those will override the
defaults.
Listing 30-1
# app/config/parameters.yml
parameters:
router.request_context.host: example.org
router.request_context.scheme: https
router.request_context.base_url: my/path
1
2
3
4
5
Configuring the Request Context per Command

To change it only in one command you can simply fetch the Request Context from the router service
and override its settings:
Listing 30-2
1
2
3
4
5
// src/AppBundle/Command/DemoCommand.php
// ...
class DemoCommand extends ContainerAwareCommand
{

Chapter 30: How to Generate URLs from the Console | 104
6
7
8
9
10
11
12
13
14
15

{
$context = $this->getContainer()->get('router')->getContext();
$context->setHost('example.com');
$context->setScheme('https');
$context->setBaseUrl('my/path');
// ... your code here

}
}

Chapter 30: How to Generate URLs from the Console | 105
Chapter 31
How to Enable Logging in Console Commands

The Console component doesn't provide any logging capabilities out of the box. Normally, you run
console commands manually and observe the output, which is why logging is not provided. However,
there are cases when you might need logging. For example, if you are running console commands
unattended, such as from cron jobs or deployment scripts, it may be easier to use Symfony's logging
capabilities instead of configuring other tools to gather console output and process it. This can be
especially handful if you already have some existing setup for aggregating and analyzing Symfony logs.
There are basically two logging cases you would need:
Manually logging some information from your command;
Logging uncaught exceptions.
Manually Logging from a Console Command

This one is really simple. When you create a console command within the full-stack framework as
described in "How to Create a Console Command", your command extends
ContainerAwareCommand1. This means that you can simply access the standard logger service
through the container and use it to do the logging:
Listing 31-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
use
use
use
use
use
use
Symfony\Component\Console\Input\InputArgument;
Psr\Log\LoggerInterface;

{
// ...
{

Chapter 31: How to Enable Logging in Console Commands | 106
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
/** @var $logger LoggerInterface */

if ($name) {
$text = 'Hello '.$name;
} else {
$text = 'Hello';
}
if ($input->getOption('yell')) {
$text = strtoupper($text);
$logger->warning('Yelled: '.$text);
} else {
$logger->info('Greeted: '.$text);
}
$output->writeln($text);
}
}
Depending on the environment in which you run your command (and your logging setup), you should
see the logged entries in app/logs/dev.log or app/logs/prod.log.
Enabling automatic Exceptions Logging

To get your console application to automatically log uncaught exceptions for all of your commands, you
can use console events.
New in version 2.3: Console events were introduced in Symfony 2.3.
First configure a listener for console exception events in the service container:
Listing 31-2
1
2
3
4
5
6
7
# app/config/services.yml
services:
app.listener.command_exception:
class: AppBundle\EventListener\ConsoleExceptionListener
arguments: ['@logger']
tags:
- { name: kernel.event_listener, event: console.exception }
Then implement the actual listener:

Listing 31-3
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
// src/AppBundle/EventListener/ConsoleExceptionListener.php
namespace AppBundle\EventListener;
use Symfony\Component\Console\Event\ConsoleExceptionEvent;
use Psr\Log\LoggerInterface;
class ConsoleExceptionListener
{
private $logger;
public function __construct(LoggerInterface $logger)
{
$this->logger = $logger;
}
public function onConsoleException(ConsoleExceptionEvent $event)
{
$command = $event->getCommand();
$exception = $event->getException();
$message = sprintf(
'%s: %s (uncaught exception) at %s line %s while running console command `%s`',
get_class($exception),

24
25
26
27
28
29
30
31
32
$exception->getMessage(),
$exception->getFile(),
$exception->getLine(),
$command->getName()
);
$this->logger->error($message, array('exception' => $exception));
}
}
In the code above, when any command throws an exception, the listener will receive an event. You
can simply log it by passing the logger service via the service configuration. Your method receives a
ConsoleExceptionEvent2 object, which has methods to get information about the event and the
exception.
Logging non-0 Exit Statuses

The logging capabilities of the console can be further extended by logging non-0 exit statuses. This way
you will know if a command had any errors, even if no exceptions were thrown.
First configure a listener for console terminate events in the service container:
Listing 31-4
1
2
3
4
5
6
7
services:
app.listener.command_error:
class: AppBundle\EventListener\ErrorLoggerListener
arguments: ['@logger']
tags:
- { name: kernel.event_listener, event: console.terminate }
Then implement the actual listener:

Listing 31-5
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
// src/AppBundle/EventListener/ErrorLoggerListener.php
use Symfony\Component\Console\Event\ConsoleTerminateEvent;
use Psr\Log\LoggerInterface;
class ErrorLoggerListener
{
private $logger;
public function __construct(LoggerInterface $logger)
{
$this->logger = $logger;
}
public function onConsoleTerminate(ConsoleTerminateEvent $event)
{
$statusCode = $event->getExitCode();
$command = $event->getCommand();
if ($statusCode === 0) {
return;
}
if ($statusCode > 255) {
$statusCode = 255;
$event->setExitCode($statusCode);
}
$this->logger->warning(sprintf(
2. http://api.symfony.com/2.7/Symfony/Component/Console/Event/ConsoleExceptionEvent.html

31
32
33
34
35
36
'Command `%s` exited with status code %d',

$command->getName(),
$statusCode
));
}
}

Chapter 32
How to Define Commands as Services

By default, Symfony will take a look in the Command directory of each bundle and automatically register
your commands. If a command extends the ContainerAwareCommand1, Symfony will even inject the
container. While making life easier, this has some limitations:
Your command must live in the Command directory;
There's no way to conditionally register your service based on the environment or availability of
some dependencies;
You can't access the container in the configure() method (because setContainer hasn't been called yet);
You can't use the same class to create many commands (i.e. each with different configuration).
To solve these problems, you can register your command as a service and tag it with
console.command:
Listing 32-1
1
2
3
4
5
6
services:
app.command.my_command:
class: AppBundle\Command\MyCommand
tags:
- { name: console.command }
Using Dependencies and Parameters to Set Default Values for Options

Imagine you want to provide a default value for the name option. You could pass one of the following as
the 5th argument of addOption():
a hardcoded string;
a container parameter (e.g. something from parameters.yml);
a value computed by a service (e.g. a repository).
By extending ContainerAwareCommand, only the first is possible, because you can't access the
container inside the configure() method. Instead, inject any parameter or service you need into the

Chapter 32: How to Define Commands as Services | 110
constructor. For example, suppose you store the default value in some %command.default_name%
parameter:
Listing 32-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
use
use
use
use
Symfony\Component\Console\Command\Command;
class GreetCommand extends Command

{
protected $defaultName;
public function __construct($defaultName)
{
$this->defaultName = $defaultName;
parent::__construct();
}
protected function configure()
{
// try to avoid work here (e.g. database query)
// this method is *always* called - see warning below
$defaultName = $this->defaultName;
$this
->setName('demo:greet')
->setDescription('Greet someone')
->addOption(
'name',
'-n',
InputOption::VALUE_REQUIRED,
'Who do you want to greet?',
$defaultName
)
;
}
{
$name = $input->getOption('name');
$output->writeln($name);
}
}
Now, just update the arguments of your service configuration like normal to inject the
command.default_name parameter:
Listing 32-3
1
2
3
4
5
6
7
8
9
10
parameters:
command.default_name: Javier
services:
app.command.my_command:
class: AppBundle\Command\MyCommand
arguments: ["%command.default_name%"]
tags:
- { name: console.command }
Great, you now have a dynamic default value!

Be careful not to actually do any work in configure (e.g. make database queries), as your code will
be run, even if you're using the console to execute a different command.

Chapter 32: How to Define Commands as Services | 111
Chapter 33
How to Customize Error Pages

In Symfony applications, all errors are treated as exceptions, no matter if they are just a 404 Not Found
error or a fatal error triggered by throwing some exception in your code.
In the development environment, Symfony catches all the exceptions and displays a special exception
page with lots of debug information to help you quickly discover the root problem:
Since these pages contain a lot of sensitive internal information, Symfony won't display them in the
production environment. Instead, it'll show a simple and generic error page:

Chapter 33: How to Customize Error Pages | 112
Error pages for the production environment can be customized in different ways depending on your
needs:
1. If you just want to change the contents and styles of the error pages to match the rest of your
application, override the default error templates;
2. If you also want to tweak the logic used by Symfony to generate error pages, override the default
exception controller;
3. If you need total control of exception handling to execute your own logic use the
kernel.exception event.
Overriding the Default Error Templates

When the error page loads, an internal ExceptionController1 is used to render a Twig template to
show the user.
This controller uses the HTTP status code, the request format and the following logic to determine the
template filename:
1. Look for a template for the given format and status code (like error404.json.twig or
error500.html.twig);
2. If the previous template doesn't exist, discard the status code and look for a generic template for
the given format (like error.json.twig or error.xml.twig);
3. If none of the previous template exist, fall back to the generic HTML template (error.html.twig).
To override these templates, simply rely on the standard Symfony method for overriding templates that
live inside a bundle: put them in the app/Resources/TwigBundle/views/Exception/ directory.
A typical project that returns HTML and JSON pages, might look like this:
Listing 33-1
1
2
3
4
5
6
7
8
9
10
11
app/
Resources/
TwigBundle/
views/
Exception/
error404.html.twig
error403.html.twig
error.html.twig
error404.json.twig
error403.json.twig
error.json.twig
# All other HTML errors (including 500)
# All other JSON errors (including 500)
1. http://api.symfony.com/2.7/Symfony/Bundle/TwigBundle/Controller/ExceptionController.html

Example 404 Error Template

To override the 404 error template for HTML pages, create a new error404.html.twig template
located at app/Resources/TwigBundle/views/Exception/:
Listing 33-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
{# app/Resources/TwigBundle/views/Exception/error404.html.twig #}
{% extends 'base.html.twig' %}
{% block body %}
<h1>Page not found</h1>
{# example security usage, see below #}

{% if app.user and is_granted('IS_AUTHENTICATED_FULLY') %}
{# ... #}
{% endif %}
<p>
The requested page couldn't be located. Checkout for any URL
misspelling or <a href="{{ path('homepage') }}">return to the homepage</a>.
</p>
{% endblock %}
In case you need them, the ExceptionController passes some information to the error template
via the status_code and status_text variables that store the HTTP status code and message
respectively.
You can customize the status code by implementing HttpExceptionInterface2 and its required
getStatusCode() method. Otherwise, the status_code will default to 500.
The exception pages shown in the development environment can be customized in the same way
as error pages. Create a new exception.html.twig template for the standard HTML exception
page or exception.json.twig for the JSON exception page.
Avoiding Exceptions when Using Security Functions in Error Templates

One of the common pitfalls when designing custom error pages is to use the is_granted() function in
the error template (or in any parent template inherited by the error template). If you do that, you'll see an
exception thrown by Symfony.
The cause of this problem is that routing is done before security. If a 404 error occurs, the security layer
isn't loaded and thus, the is_granted() function is undefined. The solution is to add the following
check before using this function:
Listing 33-3
1
2
3
{% if app.user and is_granted('...') %}

{# ... #}
{% endif %}
Testing Error Pages during Development

While you're in the development environment, Symfony shows the big exception page instead of your
shiny new customized error page. So, how can you see what it looks like and debug it?
Fortunately, the default ExceptionController allows you to preview your error pages during
development.
2. http://api.symfony.com/2.7/Symfony/Component/HttpKernel/Exception/HttpExceptionInterface.html

New in version 2.6: This feature was introduced in Symfony 2.6. Before, the third-party
WebfactoryExceptionsBundle3 could be used for the same purpose.
To use this feature, you need to have a definition in your routing_dev.yml file like so:
Listing 33-4
1
2
3
4
# app/config/routing_dev.yml
_errors:
resource: "@TwigBundle/Resources/config/routing/errors.xml"
prefix:
/_error
If you're coming from an older version of Symfony, you might need to add this to your
routing_dev.yml file. If you're starting from scratch, the Symfony Standard Edition4 already contains
it for you.
With this route added, you can use URLs like
Listing 33-5
1
2
http://localhost/app_dev.php/_error/{statusCode}
http://localhost/app_dev.php/_error/{statusCode}.{format}
to preview the error page for a given status code as HTML or for a given status code and format.
Overriding the Default ExceptionController

If you need a little more flexibility beyond just overriding the template, then you can change the
controller that renders the error page. For example, you might need to pass some additional variables
into your template.
To do this, simply create a new controller anywhere in your application and set the
twig.exception_controller configuration option to point to it:
Listing 33-6
1
2
3
twig:
exception_controller: AppBundle:Exception:showException
The ExceptionListener5 class used by the TwigBundle as a listener of the kernel.exception

event creates the request that will be dispatched to your controller. In addition, your controller will be
passed two parameters:
exception
A FlattenException6 instance created from the exception being handled.

logger
A DebugLoggerInterface7 instance which may be null in some circumstances.

Instead of creating a new exception controller from scratch you can, of course, also extend the default
ExceptionController8. In that case, you might want to override one or both of the showAction()
and findTemplate() methods. The latter one locates the template to be used.
3. https://github.com/webfactory/exceptions-bundle
4. https://github.com/symfony/symfony-standard/
5. http://api.symfony.com/2.7/Symfony/Component/HttpKernel/EventListener/ExceptionListener.html
6. http://api.symfony.com/2.7//Symfony/Component/Debug/Exception/FlattenException.html
7. http://api.symfony.com/2.7//Symfony/Component/HttpKernel/Log/DebugLoggerInterface.html

In case of extending the ExceptionController9 you may configure a service to pass the Twig
environment and the debug flag to the constructor.
Listing 33-7
1
2
3
4
5
services:
app.exception_controller:
class: AppBundle\Controller\CustomExceptionController
arguments: ['@twig', '%kernel.debug%']
And then configure twig.exception_controller using the controller as services syntax (e.g.
app.exception_controller:showAction).
The error page preview also works for your own controllers set up this way.
Working with the kernel.exception Event

When an exception is thrown, the HttpKernel10 class catches it and dispatches a kernel.exception
event. This gives you the power to convert the exception into a Response in a few different ways.
Working with this event is actually much more powerful than what has been explained before, but also
requires a thorough understanding of Symfony internals. Suppose that your code throws specialized
exceptions with a particular meaning to your application domain.
Writing your own event listener for the kernel.exception event allows you to have a closer look at the
exception and take different actions depending on it. Those actions might include logging the exception,
redirecting the user to another page or rendering specialized error pages.
If your listener calls setResponse() on the GetResponseForExceptionEvent11, event,
propagation will be stopped and the response will be sent to the client.
This approach allows you to create centralized and layered error handling: instead of catching (and
handling) the same exceptions in various controllers time and again, you can have just one (or several)
listeners deal with them.
See ExceptionListener12 class code for a real example of an advanced listener of this type.
This listener handles various security-related exceptions that are thrown in your application (like
AccessDeniedException13) and takes measures like redirecting the user to the login page,
logging them out and other things.
10. http://api.symfony.com/2.7/Symfony/Component/HttpKernel/HttpKernel.html
11. http://api.symfony.com/2.7/Symfony/Component/HttpKernel/Event/GetResponseForExceptionEvent.html
12. http://api.symfony.com/2.7/Symfony/Component/Security/Http/Firewall/ExceptionListener.html
13. http://api.symfony.com/2.7/Symfony/Component/Security/Core/Exception/AccessDeniedException.html

Chapter 34
How to Define Controllers as Services
Defining controllers as services is not officially recommended by Symfony. They are used by
some developers for very specific use cases, such as DDD (domain-driven design) and Hexagonal
Architecture applications.
In the book, you've learned how easily a controller can be used when it extends the base Controller1
class. While this works fine, controllers can also be specified as services. Even if you don't specify your
controllers as services, you might see them being used in some open-source Symfony bundles, so it may
be useful to understand both approaches.
These are the main advantages of defining controllers as services:
The entire controller and any service passed to it can be modified via the service container
configuration. This is useful when developing reusable bundles;
Your controllers are more "sandboxed". By looking at the constructor arguments, it's easy to see
what types of things this controller may or may not do;
Since dependencies must be injected manually, it's more obvious when your controller is becoming
too big (i.e. if you have many constructor arguments).
These are the main drawbacks of defining controllers as services:
It takes more work to create the controllers because they don't have automatic access to the services
or to the base controller shortcuts;
The constructor of the controllers can rapidly become too complex because you must inject every
single dependency needed by them;
The code of the controllers is more verbose because you can't use the shortcuts of the base controller
and you must replace them with some lines of code.
The recommendation from the best practices is also valid for controllers defined as services: avoid putting
your business logic into the controllers. Instead, inject services that do the bulk of the work.
1. http://api.symfony.com/2.7/Symfony/Bundle/FrameworkBundle/Controller/Controller.html

Chapter 34: How to Define Controllers as Services | 117
Defining the Controller as a Service

A controller can be defined as a service in the same way as any other class. For example, if you have the
following simple controller:
Listing 34-1
1
2
3
4
5
6
7
8
9
10
11
12
// src/AppBundle/Controller/HelloController.php
class HelloController
{
public function indexAction($name)
{
return new Response('<html><body>Hello '.$name.'!</body></html>');
}
}
Then you can define it as a service as follows:

Listing 34-2
1
2
3
4
services:
app.hello_controller:
class: AppBundle\Controller\HelloController
Referring to the Service

To refer to a controller that's defined as a service, use the single colon (:) notation. For example,
to forward to the indexAction() method of the service defined above with the id
Listing 34-3
$this->forward('app.hello_controller:indexAction', array('name' => $name));
You cannot drop the Action part of the method name when using this syntax.
You can also route to the service by using the same notation when defining the route _controller
value:
Listing 34-4
1
2
3
4
hello:
path:
/hello
defaults: { _controller: app.hello_controller:indexAction }
You can also use annotations to configure routing using a controller defined as a service. Make sure
you specify the service ID in the @Route annotation. See the FrameworkExtraBundle documentation2
for details.
If your controller implements the __invoke() method, you can simply refer to the service id
(app.hello_controller).
New in version 2.6: Support for __invoke() was introduced in Symfony 2.6.
2. https://symfony.com/doc/current/bundles/SensioFrameworkExtraBundle/annotations/routing.html#controller-as-service

Alternatives to base Controller Methods

When using a controller defined as a service, it will most likely not extend the base Controller
class. Instead of relying on its shortcut methods, you'll interact directly with the services that you need.
Fortunately, this is usually pretty easy and the base Controller class source code3 is a great source on how
to perform many common tasks.
For example, if you want to render a template instead of creating the Response object directly, then
your code would look like this if you were extending Symfony's base controller:
Listing 34-5
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
use Symfony\Bundle\FrameworkBundle\Controller\Controller;
class HelloController extends Controller
{
{
return $this->render(
'AppBundle:Hello:index.html.twig',
array('name' => $name)
);
}
}
If you look at the source code for the render function in Symfony's base Controller class4, you'll see that
this method actually uses the templating service:
Listing 34-6
public function render($view, array $parameters = array(), Response $response = null)

{
return $this->container->get('templating')->renderResponse($view, $parameters, $response);
}
In a controller that's defined as a service, you can instead inject the templating service and use it
directly:
Listing 34-7
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
use Symfony\Bundle\FrameworkBundle\Templating\EngineInterface;
class HelloController
{
private $templating;
public function __construct(EngineInterface $templating)
{
$this->templating = $templating;
}
{
return $this->templating->renderResponse(
'AppBundle:Hello:index.html.twig',
);
}
}
The service definition also needs modifying to specify the constructor argument:
3. https://github.com/symfony/symfony/blob/master/src/Symfony/Bundle/FrameworkBundle/Controller/Controller.php
4. https://github.com/symfony/symfony/blob/master/src/Symfony/Bundle/FrameworkBundle/Controller/Controller.php

Listing 34-8
1
2
3
4
5
services:
class:
AppBundle\Controller\HelloController
arguments: ['@templating']
Rather than fetching the templating service from the container, you can inject only the exact service(s)
that you need directly into the controller.
This does not mean that you cannot extend these controllers from your own base controller. The
move away from the standard base controller is because its helper methods rely on having the
container available which is not the case for controllers that are defined as services. It may be a
good idea to extract common code into a service that's injected rather than place that code into a
base controller that you extend. Both approaches are valid, exactly how you want to organize your
reusable code is up to you.
Base Controller Methods and Their Service Replacements

This list explains how to replace the convenience methods of the base controller:
createForm()5 (service: form.factory)
Listing 34-9
$formFactory->create($type, $data, $options);
createFormBuilder()6 (service: form.factory)

Listing 34-10
$formFactory->createBuilder('form', $data, $options);
createNotFoundException()7
Listing 34-11
new NotFoundHttpException($message, $previous);
forward()8 (service: http_kernel)

Listing 34-12
1
2
3
4
5
6
7
use Symfony\Component\HttpKernel\HttpKernelInterface;
// ...
$request = ...;
$attributes = array_merge($path, array('_controller' => $controller));
$subRequest = $request->duplicate($query, null, $attributes);
$httpKernel->handle($subRequest, HttpKernelInterface::SUB_REQUEST);
generateUrl()9 (service: router)

Listing 34-13
$router->generate($route, $params, $referenceType);
The $referenceType argument must be one of the constants defined in the

UrlGeneratorInterface10.
5.
6.
7.
8.
9.
http://api.symfony.com/2.7/Symfony/Bundle/FrameworkBundle/Controller/Controller.html#method_createForm
http://api.symfony.com/2.7/Symfony/Bundle/FrameworkBundle/Controller/Controller.html#method_createFormBuilder
http://api.symfony.com/2.7/Symfony/Bundle/FrameworkBundle/Controller/Controller.html#method_createNotFoundException
http://api.symfony.com/2.7/Symfony/Bundle/FrameworkBundle/Controller/Controller.html#method_forward
http://api.symfony.com/2.7/Symfony/Bundle/FrameworkBundle/Controller/Controller.html#method_generateUrl
10. http://api.symfony.com/2.7/Symfony/Component/Routing/Generator/UrlGeneratorInterface.html

getDoctrine()11 (service: doctrine)
Simply inject doctrine instead of fetching it from the container.

getUser()12 (service: security.token_storage)
Listing 34-14
1
2
3
4
5
$user = null;
$token = $tokenStorage->getToken();
if (null !== $token && is_object($token->getUser())) {
$user = $token->getUser();
}
isGranted()13 (service: security.authorization_checker)

Listing 34-15
$authChecker->isGranted($attributes, $object);
redirect()14
Listing 34-16
1
2
3
use Symfony\Component\HttpFoundation\RedirectResponse;
return new RedirectResponse($url, $status);
render()15 (service: templating)

Listing 34-17
$templating->renderResponse($view, $parameters, $response);
renderView()16 (service: templating)

Listing 34-18
$templating->render($view, $parameters);
stream()17 (service: templating)

Listing 34-19
1
2
3
4
5
6
7
8
use Symfony\Component\HttpFoundation\StreamedResponse;
$templating = $this->templating;
$callback = function () use ($templating, $view, $parameters) {
$templating->stream($view, $parameters);
}
return new StreamedResponse($callback);
getRequest has been deprecated. Instead, have an argument to your controller action method
called Request $request. The order of the parameters is not important, but the typehint must
be provided.
11.
12.
13.
14.
15.
16.
17.
http://api.symfony.com/2.7/Symfony/Bundle/FrameworkBundle/Controller/Controller.html#method_getDoctrine
http://api.symfony.com/2.7/Symfony/Bundle/FrameworkBundle/Controller/Controller.html#method_getUser
http://api.symfony.com/2.7/Symfony/Bundle/FrameworkBundle/Controller/Controller.html#method_isGranted
http://api.symfony.com/2.7/Symfony/Bundle/FrameworkBundle/Controller/Controller.html#method_redirect
http://api.symfony.com/2.7/Symfony/Bundle/FrameworkBundle/Controller/Controller.html#method_render
http://api.symfony.com/2.7/Symfony/Bundle/FrameworkBundle/Controller/Controller.html#method_renderView
http://api.symfony.com/2.7/Symfony/Bundle/FrameworkBundle/Controller/Controller.html#method_stream

Chapter 35
How to Upload Files
Instead of handling file uploading yourself, you may consider using the VichUploaderBundle1
community bundle. This bundle provides all the common operations (such as file renaming, saving
and deleting) and it's tightly integrated with Doctrine ORM, MongoDB ODM, PHPCR ODM and
Propel.
Imagine that you have a Product entity in your application and you want to add a PDF brochure for
each product. To do so, add a new property called brochure in the Product entity:
Listing 35-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
// src/AppBundle/Entity/Product.php
namespace AppBundle\Entity;
use Doctrine\ORM\Mapping as ORM;
use Symfony\Component\Validator\Constraints as Assert;
class Product
{
// ...
/**
* @ORM\Column(type="string")
*
* @Assert\NotBlank(message="Please, upload the product brochure as a PDF file.")
* @Assert\File(mimeTypes={ "application/pdf" })
*/
private $brochure;
public function getBrochure()
{
return $this->brochure;
}
public function setBrochure($brochure)
{
$this->brochure = $brochure;
return $this;
}
}
1. https://github.com/dustin10/VichUploaderBundle

Chapter 35: How to Upload Files | 122
Note that the type of the brochure column is string instead of binary or blob because it just stores
the PDF file name instead of the file contents.
Then, add a new brochure field to the form that manages the Product entity:
Listing 35-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
// src/AppBundle/Form/ProductType.php
namespace AppBundle\Form;
use Symfony\Component\Form\AbstractType;
use Symfony\Component\Form\FormBuilderInterface;
use Symfony\Component\OptionsResolver\OptionsResolver;
class ProductType extends AbstractType
{
public function buildForm(FormBuilderInterface $builder, array $options)
{
$builder
// ...
->add('brochure', 'file', array('label' => 'Brochure (PDF file)'))
// ...
;
}
public function configureOptions(OptionsResolver $resolver)
{
$resolver->setDefaults(array(
'data_class' => 'AppBundle\Entity\Product',
));
}
public function getName()
{
return 'product';
}
}
Now, update the template that renders the form to display the new brochure field (the exact template
code to add depends on the method used by your application to customize form rendering):
Listing 35-3
1
2
3
4
5
6
7
8
{# app/Resources/views/product/new.html.twig #}
<h1>Adding a new product</h1>
{{ form_start(form) }}
{# ... #}
{{ form_row(form.brochure) }}
{{ form_end(form) }}
Finally, you need to update the code of the controller that handles the form:
Listing 35-4
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
// src/AppBundle/Controller/ProductController.php
namespace AppBundle\ProductController;
use
use
use
use
use
Sensio\Bundle\FrameworkExtraBundle\Configuration\Route;
Symfony\Component\HttpFoundation\Request;
AppBundle\Entity\Product;
AppBundle\Form\ProductType;
class ProductController extends Controller

{
/**
* @Route("/product/new", name="app_product_new")
*/
public function newAction(Request $request)
{
$product = new Product();
$form = $this->createForm(new ProductType(), $product);
$form->handleRequest($request);

20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
if ($form->isSubmitted() && $form->isValid()) {

// $file stores the uploaded PDF file
/** @var Symfony\Component\HttpFoundation\File\UploadedFile $file */
$file = $product->getBrochure();
// Generate a unique name for the file before saving it

$fileName = md5(uniqid()).'.'.$file->guessExtension();
// Move the file to the directory where brochures are stored
$file->move(
$this->getParameter('brochures_directory'),
$fileName
);
// Update the 'brochure' property to store the PDF file name
// instead of its contents
$product->setBrochure($fileName);
// ... persist the $product variable or any other work
return $this->redirect($this->generateUrl('app_product_list'));
}
return $this->render('product/new.html.twig', array(
'form' => $form->createView(),
));
}
}
Now, create the brochures_directory parameter that was used in the controller to specify the
directory in which the brochures should be stored:
Listing 35-5
1
2
3
4
5
# ...
parameters:
brochures_directory: '%kernel.root_dir%/../web/uploads/brochures'
There are some important things to consider in the code of the above controller:
1. When the form is uploaded, the brochure property contains the whole PDF file contents. Since
this property stores just the file name, you must set its new value before persisting the changes
of the entity;
2. In Symfony applications, uploaded files are objects of the UploadedFile2 class. This class provides
methods for the most common operations when dealing with uploaded files;
3. A well-known security best practice is to never trust the input provided by users. This also
applies to the files uploaded by your visitors. The UploadedFile class provides methods to get the
original file extension (getExtension()3), the original file size (getClientSize()4) and the original file
name (getClientOriginalName()5). However, they are considered not safe because a malicious user
could tamper that information. That's why it's always better to generate a unique name and use
the guessExtension()6 method to let Symfony guess the right extension according to the file MIME
type;
You can use the following code to link to the PDF brochure of a product:
Listing 35-6
<a href="{{ asset('uploads/brochures/' ~ product.brochure) }}">View brochure (PDF)</a>
2.
3.
4.
5.
6.
http://api.symfony.com/2.7/Symfony/Component/HttpFoundation/File/UploadedFile.html
http://api.symfony.com/2.7/Symfony/Component/HttpFoundation/File/UploadedFile.html#method_getExtension
http://api.symfony.com/2.7/Symfony/Component/HttpFoundation/File/UploadedFile.html#method_getClientSize
http://api.symfony.com/2.7/Symfony/Component/HttpFoundation/File/UploadedFile.html#method_getClientOriginalName
http://api.symfony.com/2.7/Symfony/Component/HttpFoundation/File/UploadedFile.html#method_guessExtension

When creating a form to edit an already persisted item, the file form type still expects a File7
instance. As the persisted entity now contains only the relative file path, you first have to concatenate
the configured upload path with the stored filename and create a new File class:
Listing 35-7
1
2
3
4
5
6
use Symfony\Component\HttpFoundation\File\File;
// ...
$product->setBrochure(
new File($this->getParameter('brochures_directory').'/'.$product->getBrochure())
);
Creating an Uploader Service

To avoid logic in controllers, making them big, you can extract the upload logic to a separate service:
Listing 35-8
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
// src/AppBundle/FileUploader.php
namespace AppBundle;
use Symfony\Component\HttpFoundation\File\UploadedFile;
class FileUploader
{
private $targetDir;
public function __construct($targetDir)
{
$this->targetDir = $targetDir;
}
public function upload(UploadedFile $file)
{
$fileName = md5(uniqid()).'.'.$file->guessExtension();
$file->move($this->targetDir, $fileName);
return $fileName;
}
}
Then, define a service for this class:

Listing 35-9
1
2
3
4
5
6
services:
# ...
app.brochure_uploader:
class: AppBundle\FileUploader
arguments: ['%brochures_directory%']
Now you're ready to use this service in the controller:

Listing 35-10
1
2
3
4
5
6
7
8
9
10
11
// src/AppBundle/Controller/ProductController.php
// ...
{
// ...
if ($form->isValid()) {
$file = $product->getBrochure();
$fileName = $this->get('app.brochure_uploader')->upload($file);
7. http://api.symfony.com/2.7/Symfony/Component/HttpFoundation/File/File.html

12
13
14
15
16
17
18
$product->setBrochure($fileName);
// ...
}
// ...
}
Using a Doctrine Listener

If you are using Doctrine to store the Product entity, you can create a Doctrine listener to automatically
upload the file when persisting the entity:
Listing 35-11
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
// src/AppBundle/EventListener/BrochureUploadListener.php
use
use
use
use
use
Symfony\Component\HttpFoundation\File\UploadedFile;
Doctrine\ORM\Event\LifecycleEventArgs;
Doctrine\ORM\Event\PreUpdateEventArgs;
AppBundle\Entity\Product;
AppBundle\FileUploader;
class BrochureUploadListener
{
private $uploader;
public function __construct(FileUploader $uploader)
{
$this->uploader = $uploader;
}
public function prePersist(LifecycleEventArgs $args)
{
$entity = $args->getEntity();
$this->uploadFile($entity);
}
public function preUpdate(PreUpdateEventArgs $args)
{
$this->uploadFile($entity);
}
private function uploadFile($entity)
{
// upload only works for Product entities
if (!$entity instanceof Product) {
return;
}
$file = $entity->getBrochure();
// only upload new files

if (!$file instanceof UploadedFile) {
return;
}
$fileName = $this->uploader->upload($file);
$entity->setBrochure($fileName);
}
}
Now, register this class as a Doctrine listener:

Listing 35-12
1
2
3
4
5
6
7
8
9
services:
# ...
app.doctrine_brochure_listener:
class: AppBundle\EventListener\BrochureUploadListener
arguments: ['@app.brochure_uploader']
tags:
- { name: doctrine.event_listener, event: prePersist }
- { name: doctrine.event_listener, event: preUpdate }
This listeners is now automatically executed when persisting a new Product entity. This way, you can
remove everything related to uploading from the controller.
This listener can also create the File instance based on the path when fetching entities from the
database:
Listing 35-13
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
// ...
use Symfony\Component\HttpFoundation\File\File;
// ...
class BrochureUploadListener
{
// ...
public function postLoad(LifecycleEventArgs $args)
{
$fileName = $entity->getBrochure();
$entity->setBrochure(new File($this->targetPath.'/'.$fileName));
}
}
After adding these lines, configure the listener to also listen for the postLoad event.

Chapter 36
How to Optimize your Development

Environment for Debugging
When you work on a Symfony project on your local machine, you should use the dev environment
(app_dev.php front controller). This environment configuration is optimized for two main purposes:
Give the developer accurate feedback whenever something goes wrong (web debug toolbar, nice
exception pages, profiler, ...);
Be as similar as possible as the production environment to avoid problems when deploying the
project.
Disabling the Bootstrap File and Class Caching

And to make the production environment as fast as possible, Symfony creates big PHP files in your
cache containing the aggregation of PHP classes your project needs for every request. However, this
behavior can confuse your IDE or your debugger. This recipe shows you how you can tweak this caching
mechanism to make it friendlier when you need to debug code that involves Symfony classes.
The app_dev.php front controller reads as follows by default:
Listing 36-1
1
2
3
4
5
6
7
8
// ...
$loader = require_once __DIR__.'/../app/bootstrap.php.cache';
$kernel = new AppKernel('dev', true);
$kernel->loadClassCache();
$request = Request::createFromGlobals();
To make your debugger happier, disable all PHP class caches by removing the call to
loadClassCache() and by replacing the require statements like below:
Listing 36-2
1
2
3
// ...
// $loader = require_once __DIR__.'/../app/bootstrap.php.cache';

Chapter 36: How to Optimize your Development Environment for Debugging | 128
4
5
6
7
8
9
$loader = require_once __DIR__.'/../app/autoload.php';

$kernel = new AppKernel('dev', true);
// $kernel->loadClassCache();
$request = Request::createFromGlobals();
If you disable the PHP caches, don't forget to revert after your debugging session.
Some IDEs do not like the fact that some classes are stored in different locations. To avoid problems, you
can either tell your IDE to ignore the PHP cache files, or you can change the extension used by Symfony
for these files:
Listing 36-3
$kernel->loadClassCache('classes', '.php.cache');

Chapter 36: How to Optimize your Development Environment for Debugging | 129
Chapter 37
How to Deploy a Symfony Application
Deploying can be a complex and varied task depending on the setup and the requirements of
your application. This article is not a step-by-step guide, but is a general list of the most common
requirements and ideas for deployment.
Symfony Deployment Basics

The typical steps taken while deploying a Symfony application include:
1. Upload your code to the production server;
2. Install your vendor dependencies (typically done via Composer and may be done before
uploading);
3. Running database migrations or similar tasks to update any changed data structures;
4. Clearing (and optionally, warming up) your cache.
A deployment may also include other tasks, such as:
Tagging a particular version of your code as a release in your source control repository;
Creating a temporary staging area to build your updated setup "offline";
Running any tests available to ensure code and/or server stability;
Removal of any unnecessary files from the web/ directory to keep your production environment
clean;
Clearing of external cache systems (like Memcached1 or Redis2).
How to Deploy a Symfony Application

There are several ways you can deploy a Symfony application. Start with a few basic deployment
strategies and build up from there.
1. http://memcached.org/
2. http://redis.io/

Chapter 37: How to Deploy a Symfony Application | 130
Basic File Transfer

The most basic way of deploying an application is copying the files manually via ftp/scp (or similar
method). This has its disadvantages as you lack control over the system as the upgrade progresses. This
method also requires you to take some manual steps after transferring the files (see Common PostDeployment Tasks)
Using Source Control

If you're using source control (e.g. Git or SVN), you can simplify by having your live installation also be a
copy of your repository. When you're ready to upgrade it is as simple as fetching the latest updates from
your source control system.
This makes updating your files easier, but you still need to worry about manually taking other steps (see
Common Post-Deployment Tasks).
Using Build Scripts and other Tools

There are also tools to help ease the pain of deployment. Some of them have been specifically tailored to
the requirements of Symfony.
Capistrano3 with Symfony plugin4
Capistrano5 is a remote server automation and deployment tool written in Ruby. Symfony plugin6 is
a plugin to ease Symfony related tasks, inspired by Capifony7 (which works only with Capistrano 2
)
sf2debpkg8
Helps you build a native Debian package for your Symfony project.
Magallanes9
This Capistrano-like deployment tool is built in PHP, and may be easier for PHP developers to
extend for their needs.
Fabric10
This Python-based library provides a basic suite of operations for executing local or remote shell
commands and uploading/downloading files.
Deployer11
This is another native PHP rewrite of Capistrano, with some ready recipes for Symfony.
Bundles
There are some bundles that add deployment features12 directly into your Symfony console.
Basic scripting
You can of course use shell, Ant13 or any other build tool to script the deploying of your project.
3. http://capistranorb.com/
4. https://github.com/capistrano/symfony/
5. http://capistranorb.com/
6. https://github.com/capistrano/symfony/
7. http://capifony.org/
8. https://github.com/liip/sf2debpkg
9. https://github.com/andres-montanez/Magallanes
10. http://www.fabfile.org/
11. http://deployer.org/
12. http://knpbundles.com/search?q=deploy
13. http://blog.sznapka.pl/deploying-symfony2-applications-with-ant

Platform as a Service Providers

The Symfony Cookbook includes detailed articles for some of the most well-known Platform as a
Service (PaaS) providers:
Microsoft Azure
Heroku
Platform.sh
Common Post-Deployment Tasks

After deploying your actual source code, there are a number of common things you'll need to do:
A) Check Requirements
Check if your server meets the requirements by running:
Listing 37-1
$ php app/check.php
B) Configure your app/config/parameters.yml File

This file should not be deployed, but managed through the automatic utilities provided by Symfony.
C) Install/Update your Vendors

Your vendors can be updated before transferring your source code (i.e. update the vendor/ directory,
then transfer that with your source code) or afterwards on the server. Either way, just update your
vendors as you normally do:
Listing 37-2
$ composer install --no-dev --optimize-autoloader
The --optimize-autoloader flag improves Composer's autoloader performance significantly by

building a "class map". The --no-dev flag ensures that development packages are not installed in
the production environment.
If you get a "class not found" error during this step, you may need to run export
SYMFONY_ENV=prod before running this command so that the post-install-cmd scripts run in
the prod environment.
D) Clear your Symfony Cache

Make sure you clear (and warm-up) your Symfony cache:
Listing 37-3
$ php app/console cache:clear --env=prod --no-debug
E) Dump your Assetic Assets

If you're using Assetic, you'll also want to dump your assets:
Listing 37-4

F) Other Things!
There may be lots of other things that you need to do, depending on your setup:
Running any database migrations

Clearing your APC cache
Running assets:install (already taken care of in composer
Add/edit CRON jobs
Pushing assets to a CDN
...
install)
Application Lifecycle: Continuous Integration, QA, etc

While this entry covers the technical details of deploying, the full lifecycle of taking code from
development up to production may have a lot more steps (think deploying to staging, QA (Quality
Assurance), running tests, etc).
The use of staging, testing, QA, continuous integration, database migrations and the capability to roll
back in case of failure are all strongly advised. There are simple and more complex tools and one can
make the deployment as easy (or sophisticated) as your environment requires.
Don't forget that deploying your application also involves updating any dependency (typically via
Composer), migrating your database, clearing your cache and other potential things like pushing assets
to a CDN (see Common Post-Deployment Tasks).

Chapter 38
Deploying to Microsoft Azure Website Cloud

This step by step cookbook describes how to deploy a small Symfony web application to the Microsoft
Azure Website cloud platform. It will explain how to set up a new Azure website including configuring
the right PHP version and global environment variables. The document also shows how to you can
leverage Git and Composer to deploy your Symfony application to the cloud.
Setting up the Azure Website

To set up a new Microsoft Azure Website, first sign up with Azure1 or sign in with your credentials. Once
you're connected to your Azure Portal2 interface, scroll down to the bottom and select the New panel.
On this panel, click Web Site and choose Custom Create:
1. https://signup.live.com/signup.aspx
2. https://manage.windowsazure.com

Chapter 38: Deploying to Microsoft Azure Website Cloud | 134
Step 1: Create Web Site

Here, you will be prompted to fill in some basic information.
For the URL, enter the URL that you would like to use for your Symfony application, then pick Create
new web hosting plan in the region you want. By default, a free 20 MB SQL database is selected in the
database dropdown list. In this tutorial, the Symfony app will connect to a MySQL database. Pick the
Create a new MySQL database option in the dropdown list. You can keep the DefaultConnection
string name. Finally, check the box Publish from source control to enable a Git repository and go to
the next step.
Step 2: New MySQL Database

On this step, you will be prompted to set up your MySQL database storage with a database name and a
region. The MySQL database storage is provided by Microsoft in partnership with ClearDB. Choose the
same region you selected for the hosting plan configuration in the previous step.
Agree to the terms and conditions and click on the right arrow to continue.

Step 3: Where Is your Source Code

Now, on the third step, select a Local Git repository item and click on the right arrow to configure your
Azure Website credentials.
Step 4: New Username and Password

Great! You're now on the final step. Create a username and a secure password: these will become
essential identifiers to connect to the FTP server and also to push your application code to the Git
repository.
Congratulations! Your Azure Website is now up and running. You can check it by browsing to the
Website url you configured in the first step. You should see the following display in your web browser:

The Microsoft Azure portal also provides a complete control panel for the Azure Website.
Your Azure Website is ready! But to run a Symfony site, you need to configure just a few additional
things.
Configuring the Azure Website for Symfony

This section of the tutorial details how to configure the correct version of PHP to run Symfony. It also
shows you how to enable some mandatory PHP extensions and how to properly configure PHP for a
production environment.
Configuring the latest PHP Runtime

Even though Symfony only requires PHP 5.3.9 to run, it's always recommended to use the most recent
PHP version whenever possible. PHP 5.3 is no longer supported by the PHP core team, but you can
update it easily in Azure.
To update your PHP version on Azure, go to the Configure tab of the control panel and select the version
you want.
Click the Save button in the bottom bar to save your changes and restart the web server.
Choosing a more recent PHP version can greatly improve runtime performance. PHP 5.5 ships with
a new built-in PHP accelerator called OPCache that replaces APC. On an Azure Website, OPCache
is already enabled and there is no need to install and set up APC.
The following screenshot shows the output of a phpinfo3 script run from an Azure Website to
verify that PHP 5.5 is running with OPCache enabled.
Tweaking php.ini Configuration Settings

Microsoft Azure allows you to override the php.ini global configuration settings by creating a custom
.user.ini file under the project root directory (site/wwwroot).
Listing 38-1
1
2
3
4
; .user.ini
expose_php = Off
memory_limit = 256M
upload_max_filesize = 10M
3. http://php.net/manual/en/function.phpinfo.php

None of these settings needs to be overridden. The default PHP configuration is already pretty good, so
this is just an example to show how you can easily tweak PHP internal settings by uploading your custom
.ini file.
You can either manually create this file on your Azure Website FTP server under the site/wwwroot
directory or deploy it with Git. You can get your FTP server credentials from the Azure Website Control
panel under the Dashboard tab on the right sidebar. If you want to use Git, simply put your .user.ini
file at the root of your local repository and push your commits to your Azure Website repository.
This cookbook has a section dedicated to explaining how to configure your Azure Website Git
repository and how to push the commits to be deployed. See Deploying from Git. You can also learn
more about configuring PHP internal settings on the official PHP MSDN documentation4 page.
Enabling the PHP intl Extension

This is the tricky part of the guide! At the time of writing this cookbook, Microsoft Azure Website
provided the intl extension, but it's not enabled by default. To enable the intl extension, there is no
need to upload any DLL files as the php_intl.dll file already exists on Azure. In fact, this file just
needs to be moved into the custom website extension directory.
The Microsoft Azure team is currently working on enabling the intl PHP extension by default. In
the near future, the following steps will no longer be necessary.
To get the php_intl.dll file under your site/wwwroot directory, simply access the online Kudu
tool by browsing to the following URL:
Listing 38-2
https://[your-website-name].scm.azurewebsites.net
Kudu is a set of tools to manage your application. It comes with a file explorer, a command line prompt,
a log stream and a configuration settings summary page. Of course, this section can only be accessed if
you're logged in to your main Azure Website account.
4. http://blogs.msdn.com/b/silverlining/archive/2012/07/10/configuring-php-in-windows-azure-websites-with-user-ini-files.aspx

From the Kudu front page, click on the Debug Console navigation item in the main menu and choose
CMD. This should open the Debug Console page that shows a file explorer and a console prompt
below.
In the console prompt, type the following three commands to copy the original php_intl.dll
extension file into a custom website ext/ directory. This new directory must be created under the main
directory site/wwwroot.
Listing 38-3
1
2
3
$ cd site\wwwroot
$ mkdir ext
$ copy "D:\Program Files (x86)\PHP\v5.5\ext\php_intl.dll" ext
The whole process and output should look like this:
To complete the activation of the php_intl.dll extension, you must tell Azure Website to load it
from the newly created ext directory. This can be done by registering a global PHP_EXTENSIONS
environment variable from the Configure tab of the main Azure Website Control panel.
In the app settings section, register the PHP_EXTENSIONS environment variable with the value
ext\php_intl.dll as shown in the screenshot below:

Hit "save" to confirm your changes and restart the web server. The PHP Intl extension should now
be available in your web server environment. The following screenshot of a phpinfo5 page verifies the
intl extension is properly enabled:
Great! The PHP environment setup is now complete. Next, you'll learn how to configure the Git
repository and push code to production. You'll also learn how to install and configure the Symfony app
after it's deployed.
Deploying from Git

First, make sure Git is correctly installed on your local machine using the following command in your
terminal:
Listing 38-4
$ git --version
Get your Git from the git-scm.com6 website and follow the instructions to install and configure it on
your local machine.
In the Azure Website Control panel, browse the Deployment tab to get the Git repository URL where
you should push your code:
5. http://php.net/manual/en/function.phpinfo.php
6. http://git-scm.com/download

Now, you'll want to connect your local Symfony application with this remote Git repository on Azure
Website. If your Symfony application is not yet stored with Git, you must first create a Git repository
in your Symfony application directory with the git init command and commit to it with the git
commit command.
Also, make sure your Symfony repository has a .gitignore file at its root directory with at least the
following contents:
Listing 38-5
1
2
3
4
5
6
7
8
9
10
11
12
13
14
/app/bootstrap.php.cache
/app/cache/*
/app/config/parameters.yml
/app/logs/*
!app/cache/.gitkeep
!app/logs/.gitkeep
/app/SymfonyRequirements.php
/build/
/vendor/
/bin/
/composer.phar
/web/app_dev.php
/web/bundles/
/web/config.php
The .gitignore file asks Git not to track any of the files and directories that match these patterns. This
means these files won't be deployed to the Azure Website.
Now, from the command line on your local machine, type the following at the root of your Symfony
project:
Listing 38-6
1
2
$ git remote add azure https://<username>@<your-website-name>.scm.azurewebsites.net:443/<your-website-name>.git

$ git push azure master
Don't forget to replace the values enclosed by < and > with your custom settings displayed in the
Deployment tab of your Azure Website panel. The git remote command connects the Azure Website
remote Git repository and assigns an alias to it with the name azure. The second git push command
pushes all your commits to the remote master branch of your remote azure Git repository.
The deployment with Git should produce an output similar to the screenshot below:

The code of the Symfony application has now been deployed to the Azure Website which you can browse
from the file explorer of the Kudu application. You should see the app/, src/ and web/ directories
under your site/wwwroot directory on the Azure Website filesystem.
Configure the Symfony Application

PHP has been configured and your code has been pushed with Git. The last step is to configure the
application and install the third party dependencies it requires that aren't tracked by Git. Switch back to
the online Console of the Kudu application and execute the following commands in it:
Listing 38-7
1
2
3
$ cd site\wwwroot
$ curl -sS https://getcomposer.org/installer | php
$ php -d extension=php_intl.dll composer.phar install
The curl command retrieves and downloads the Composer command line tool and installs it at the
root of the site/wwwroot directory. Then, running the Composer install command downloads and
installs all necessary third-party libraries.
This may take a while depending on the number of third-party dependencies you've configured in your
composer.json file.
The -d switch allows you to quickly override/add any php.ini settings. In this command, we are
forcing PHP to use the intl extension, because it is not enabled by default in Azure Website at
the moment. Soon, this -d option will no longer be needed since Microsoft will enable the intl
extension by default.
At the end of the composer install command, you will be prompted to fill in the values of some
Symfony settings like database credentials, locale, mailer credentials, CSRF token protection, etc. These
parameters come from the app/config/parameters.yml.dist file.

The most important thing in this cookbook is to correctly set up your database settings. You can get your
MySQL database settings on the right sidebar of the Azure Website Dashboard panel. Simply click on
the View Connection Strings link to make them appear in a pop-in.
The displayed MySQL database settings should be something similar to the code below. Of course, each
value depends on what you've already configured.
Listing 38-8
Database=mysymfonyMySQL;Data Source=eu-cdbr-azure-north-c.cloudapp.net;User Id=bff2481a5b6074;Password=bdf50b42
Switch back to the console and answer the prompted questions and provide the following answers. Don't
forget to adapt the values below with your real values from the MySQL connection string.
Listing 38-9
1
2
3
4
5
6
7
database_driver: pdo_mysql
database_host: u-cdbr-azure-north-c.cloudapp.net
database_port: null
database_name: mysymfonyMySQL
database_user: bff2481a5b6074
database_password: bdf50b42
// ...
Don't forget to answer all the questions. It's important to set a unique random string for the secret
variable. For the mailer configuration, Azure Website doesn't provide a built-in mailer service. You
should consider configuring the host-name and credentials of some other third-party mailing service if
your application needs to send emails.
Your Symfony application is now configured and should be almost operational. The final step is to build
the database schema. This can easily be done with the command line interface if you're using Doctrine.
In the online Console tool of the Kudu application, run the following command to mount the tables into
your MySQL database.
Listing 38-10
$ php app/console doctrine:schema:update --force
This command builds the tables and indexes for your MySQL database. If your Symfony application is
more complex than a basic Symfony Standard Edition, you may have additional commands to execute
for setup (see How to Deploy a Symfony Application).
Make sure that your application is running by browsing the app.php front controller with your web
browser and the following URL:
Listing 38-11
http://<your-website-name>.azurewebsites.net/web/app.php
If Symfony is correctly installed, you should see the front page of your Symfony application showing.
Configure the Web Server

At this point, the Symfony application has been deployed and works perfectly on the Azure Website.
However, the web folder is still part of the URL, which you definitely don't want. But don't worry! You
can easily configure the web server to point to the web folder and remove the web in the URL (and
guarantee that nobody can access files outside of the web directory.)
To do this, create and deploy (see previous section about Git) the following web.config file. This file
must be located at the root of your project next to the composer.json file. This file is the Microsoft IIS
Server equivalent to the well-known .htaccess file from Apache. For a Symfony application, configure
it with the following content:
Listing 38-12
1
2
3
4
5


<?xml version="1.0" encoding="UTF-8"?>
<configuration>
<system.webServer>
<rewrite>

6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
<rules>
<clear />
<rule name="BlockAccessToPublic" patternSyntax="Wildcard" stopProcessing="true">
<match url="*" />
<conditions logicalGrouping="MatchAll" trackAllCaptures="false">
<add input="{URL}" pattern="/web/*" />
</conditions>
<action type="CustomResponse" statusCode="403" statusReason="Forbidden: Access is denied."
statusDescription="You do not have permission to view this directory or page using the credentials that you
supplied." />
</rule>
<rule name="RewriteAssetsToPublic" stopProcessing="true">
<match url="^(.*)(\.css|\.js|\.jpg|\.png|\.gif)$" />
</conditions>
<action type="Rewrite" url="web/{R:0}" />
</rule>
<rule name="RewriteRequestsToPublic" stopProcessing="true">
<match url="^(.*)$" />
</conditions>
<action type="Rewrite" url="web/app.php/{R:0}" />
</rule>
</rules>
</rewrite>
</system.webServer>
</configuration>
As you can see, the latest rule RewriteRequestsToPublic is responsible for rewriting any URLs to
the web/app.php front controller which allows you to skip the web/ folder in the URL. The first rule
called BlockAccessToPublic matches all URL patterns that contain the web/ folder and serves a 403
Forbidden HTTP response instead. This example is based on Benjamin Eberlei's sample you can find
on GitHub in the SymfonyAzureEdition7 bundle.
Deploy this file under the site/wwwroot directory of the Azure Website and browse to your application
without the web/app.php segment in the URL.
Conclusion
Nice work! You've now deployed your Symfony application to the Microsoft Azure Website Cloud
platform. You also saw that Symfony can be easily configured and executed on a Microsoft IIS web server.
The process is simple and easy to implement. And as a bonus, Microsoft is continuing to reduce the
number of steps needed so that deployment becomes even easier.
7. https://github.com/beberlei/symfony-azure-edition/

Chapter 39
Deploying to Heroku Cloud

This step by step cookbook describes how to deploy a Symfony web application to the Heroku cloud
platform. Its contents are based on the original article1 published by Heroku.
Setting up
To set up a new Heroku website, first sign up with Heroku2 or sign in with your credentials. Then
download and install the Heroku Toolbelt3 on your local computer.
You can also check out the getting Started with PHP on Heroku4 guide to gain more familiarity with the
specifics of working with PHP applications on Heroku.
Preparing your Application

Deploying a Symfony application to Heroku doesn't require any change in its code, but it requires some
minor tweaks to its configuration.
By default, the Symfony app will log into your application's app/log/ directory. This is not ideal as
Heroku uses an ephemeral file system5. On Heroku, the best way to handle logging is using Logplex6. And
the best way to send log data to Logplex is by writing to STDERR or STDOUT. Luckily, Symfony uses the
excellent Monolog library for logging. So, a new log destination is just a change to a config file away.
Open the app/config/config_prod.yml file, locate the monolog/handlers/nested section
(or create it if it doesn't exist yet) and change the value of path from "%kernel.logs_dir%/
%kernel.environment%.log" to "php://stderr":
Listing 39-1
1
2
monolog:
1. https://devcenter.heroku.com/articles/getting-started-with-symfony2
2. https://signup.heroku.com/signup/dc
3. https://devcenter.heroku.com/articles/getting-started-with-php#set-up
4. https://devcenter.heroku.com/articles/getting-started-with-php
5. https://devcenter.heroku.com/articles/dynos#ephemeral-filesystem
6. https://devcenter.heroku.com/articles/logplex

Chapter 39: Deploying to Heroku Cloud | 147
3
4
5
6
7
8
# ...
handlers:
# ...
nested:
# ...
path: 'php://stderr'
Once the application is deployed, run heroku logs --tail to keep the stream of logs from Heroku
open in your terminal.
Creating a new Application on Heroku

To create a new Heroku application that you can push to, use the CLI create command:
Listing 39-2
1
2
3
4
5
$ heroku create
Creating mighty-hamlet-1981 in organization heroku... done, stack is cedar
http://mighty-hamlet-1981.herokuapp.com/ | [email protected]:mighty-hamlet-1981.git
Git remote heroku added
You are now ready to deploy the application as explained in the next section.
Deploying your Application on Heroku

Before your first deploy, you need to do just three more things, which are explained below:
1. Create a Procfile
2. Set the Environment to prod
3. Push your Code to Heroku
1) Create a Procfile
By default, Heroku will launch an Apache web server together with PHP to serve applications. However,
two special circumstances apply to Symfony applications:
1. The document root is in the web/ directory and not in the root directory of the application;
2. The Composer bin-dir, where vendor binaries (and thus Heroku's own boot scripts) are placed,
is bin/ , and not the default vendor/bin.
Vendor binaries are usually installed to vendor/bin by Composer, but sometimes (e.g. when
running a Symfony Standard Edition project!), the location will be different. If in doubt, you can
always run composer config bin-dir to figure out the right location.
Create a new file called Procfile (without any extension) at the root directory of the application and
add just the following content:
Listing 39-3
web: bin/heroku-php-apache2 web/
If you prefer to use Nginx, which is also available on Heroku, you can create a configuration file for
it and point to it from your Procfile as described in the Heroku documentation7:
Listing 39-4
web: bin/heroku-php-nginx -C nginx_app.conf web/
7. https://devcenter.heroku.com/articles/custom-php-settings#nginx

If you prefer working on the command console, execute the following commands to create the
Procfile file and to add it to the repository:
Listing 39-5
1
2
3
4
5
$ echo "web: bin/heroku-php-apache2 web/" > Procfile

$ git add .
$ git commit -m "Procfile for Apache and PHP"
[master 35075db] Procfile for Apache and PHP
1 file changed, 1 insertion(+)
2) Set the Environment to prod

During a deployment, Heroku runs composer install --no-dev to install all the dependencies
your application requires. However, typical post-install-commands8 in composer.json, e.g. to install
assets or clear (or pre-warm) caches, run using Symfony's dev environment by default.
This is clearly not what you want - the app runs in "production" (even if you use it just for an experiment,
or as a staging environment), and so any build steps should use the same prod environment as well.
Thankfully, the solution to this problem is very simple: Symfony will pick up an environment variable
named SYMFONY_ENV and use that environment if nothing else is explicitly set. As Heroku exposes
all config vars9 as environment variables, you can issue a single command to prepare your app for a
deployment:
Listing 39-6
$ heroku config:set SYMFONY_ENV=prod
Be aware that dependencies from composer.json listed in the require-dev section are never
installed during a deploy on Heroku. This may cause problems if your Symfony environment relies
on such packages. The solution is to move these packages from require-dev to the require
section.
3) Push your Code to Heroku

Next up, it's finally time to deploy your application to Heroku. If you are doing this for the very first time,
you may see a message such as the following:
Listing 39-7
1
2
3
The authenticity of host 'heroku.com (50.19.85.132)' can't be established.

RSA key fingerprint is 8b:48:5e:67:0e:c9:16:47:32:f2:87:0c:1f:c8:60:ad.
Are you sure you want to continue connecting (yes/no)?
In this case, you need to confirm by typing yes and hitting <Enter> key - ideally after you've verified
that the RSA key fingerprint is correct10.
Then, deploy your application executing this command:
Listing 39-8
1
2
3
4
5
6
7
8
9
10
$ git push heroku master

Initializing repository, done.
Counting objects: 130, done.
Delta compression using up to 4 threads.
Compressing objects: 100% (107/107), done.
Writing objects: 100% (130/130), 70.88 KiB | 0 bytes/s, done.
Total 130 (delta 17), reused 0 (delta 0)
-----> PHP app detected
8. https://getcomposer.org/doc/articles/scripts.md
9. https://devcenter.heroku.com/articles/config-vars
10. https://devcenter.heroku.com/articles/git-repository-ssh-fingerprints

11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
-----> Setting up runtime environment...

- PHP 5.5.12
- Apache 2.4.9
- Nginx 1.4.6
-----> Installing PHP extensions:
- opcache (automatic; bundled, using 'ext-opcache.ini')
-----> Installing dependencies...
Composer version 64ac32fca9e64eb38e50abfadc6eb6f2d0470039 2014-05-24 20:57:50
Loading composer repositories with package information
Installing dependencies from lock file
- ...
Generating optimized autoload files
Creating the "app/config/parameters.yml" file
Clearing the cache for the dev environment with debug true
Installing assets using the hard copy option
Installing assets for Symfony\Bundle\FrameworkBundle into web/bundles/framework
Installing assets for Acme\DemoBundle into web/bundles/acmedemo
Installing assets for Sensio\Bundle\DistributionBundle into web/bundles/sensiodistribution
-----> Building runtime environment...
-----> Discovering process types
Procfile declares types -> web
-----> Compressing... done, 61.5MB
-----> Launching... done, v3
http://mighty-hamlet-1981.herokuapp.com/ deployed to Heroku
To [email protected]:mighty-hamlet-1981.git
* [new branch]
master -> master
And that's it! If you now open your browser, either by manually pointing it to the URL heroku create
gave you, or by using the Heroku Toolbelt, the application will respond:
Listing 39-9
1
2
$ heroku open
Opening mighty-hamlet-1981... done
You should be seeing your Symfony application in your browser.

If you take your first steps on Heroku using a fresh installation of the Symfony Standard Edition,
you may run into a 404 page not found error. This is because the route for / is defined by the
AcmeDemoBundle, but the AcmeDemoBundle is only loaded in the dev environment (check out
your AppKernel class). Try opening /app/example from the AppBundle.
Custom Compile Steps

If you wish to execute additional custom commands during a build, you can leverage Heroku's custom
compile steps11. Imagine you want to remove the dev front controller from your production environment
on Heroku in order to avoid a potential vulnerability. Adding a command to remove web/app_dev.php
to Composer's post-install-commands12 would work, but it also removes the controller in your local
development environment on each composer install or composer update respectively. Instead,
you can add a custom Composer command13 named compile (this key name is a Heroku convention)
11. https://devcenter.heroku.com/articles/php-support#custom-compile-step
12. https://getcomposer.org/doc/articles/scripts.md
13. https://getcomposer.org/doc/articles/scripts.md#writing-custom-commands

to the scripts section of your composer.json. The listed commands hook into Heroku's deploy
process:
Listing 39-10
1
2
3
4
5
6
7
{
"scripts": {
"compile": [
"rm web/app_dev.php"
]
}
}
This is also very useful to build assets on the production system, e.g. with Assetic:
Listing 39-11
1
2
3
4
5
6
7
{
"scripts": {
"compile": [
"app/console assetic:dump"
]
}
}
Node.js Dependencies
Building assets may depend on node packages, e.g. uglifyjs or uglifycss for asset minification.
Installing node packages during the deploy requires a node installation. But currently, Heroku
compiles your app using the PHP buildpack, which is auto-detected by the presence of a
composer.json file, and does not include a node installation. Because the Node.js buildpack has
a higher precedence than the PHP buildpack (see Heroku buildpacks14), adding a package.json
listing your node dependencies makes Heroku opt for the Node.js buildpack instead:
Listing 39-12
1
2
3
4
5
6
7
8
9
10
{
"name": "myApp",
"engines": {
"node": "0.12.x"
},
"dependencies": {
"uglifycss": "*",
"uglify-js": "*"
}
}
With the next deploy, Heroku compiles your app using the Node.js buildpack and your npm
packages become installed. On the other hand, your composer.json is now ignored. To compile
your app with both buildpacks, Node.js and PHP, you need to use both buildpacks. To override
buildpack auto-detection, you need to explicitly set the buildpack:
Listing 39-13
1
2
3
4
5
6
7
8
$ heroku buildpacks:set heroku/nodejs

Buildpack set. Next release on your-application will use heroku/nodejs.
Run git push heroku master to create a new release using this buildpack.
$ heroku buildpacks:set heroku/php --index 2
Buildpack set. Next release on your-application will use:
1. heroku/nodejs
2. heroku/php
Run git push heroku master to create a new release using these buildpacks.
With the next deploy, you can benefit from both buildpacks. This setup also enables your Heroku
environment to make use of node based automatic build tools like Grunt15 or gulp16.
14. https://devcenter.heroku.com/articles/buildpacks
15. http://gruntjs.com
16. http://gulpjs.com

Chapter 40
Deploying to Platform.sh
This step-by-step cookbook describes how to deploy a Symfony web application to Platform.sh1. You can
read more about using Symfony with Platform.sh on the official Platform.sh documentation2.
Deploy an Existing Site

In this guide, it is assumed your codebase is already versioned with Git.
Get a Project on Platform.sh

You need to subscribe to a Platform.sh project3. Choose the development plan and go through the
checkout process. Once your project is ready, give it a name and choose: Import an existing site.
Prepare Your Application

To deploy your Symfony application on Platform.sh, you simply need to add a .platform.app.yaml
at the root of your Git repository which will tell Platform.sh how to deploy your application (read more
about Platform.sh configuration files4).
Listing 40-1
1
2
3
4
5
6
7
8
9
10
11
12
# .platform.app.yaml
# This file describes an application. You can have multiple applications
# in the same project.
# The name of this app. Must be unique within a project.
name: myphpproject
# The type of the application to build.
type: php:5.6
build:
flavor: symfony
1. https://platform.sh
2. https://docs.platform.sh/toolstacks/symfony/symfony-getting-started
3. https://marketplace.commerceguys.com/platform/buy-now
4. https://docs.platform.sh/reference/configuration-files

Chapter 40: Deploying to Platform.sh | 152
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
# The relationships of the application with services or other applications.

# The left-hand side is the name of the relationship as it will be exposed
# to the application in the PLATFORM_RELATIONSHIPS variable. The right-hand
# side is in the form `<service name>:<endpoint name>`.
relationships:
database: 'mysql:mysql'
# The configuration of app when it is exposed to the web.
web:
# The public directory of the app, relative to its root.
document_root: '/web'
# The front-controller script to send non-static requests to.
passthru: '/app.php'
# The size of the persistent disk of the application (in MB).
disk: 2048
# The mounts that will be performed when the package is deployed.
mounts:
'/app/cache': 'shared:files/cache'
'/app/logs': 'shared:files/logs'
# The hooks that will be performed when the package is deployed.
hooks:
build: |
rm web/app_dev.php
app/console --env=prod assetic:dump --no-debug
deploy: |
app/console --env=prod cache:clear
For best practices, you should also add a .platform folder at the root of your Git repository which
contains the following files:
Listing 40-2
1
2
3
4
5
# .platform/routes.yaml
"http://{default}/":
type: upstream
# the first part should be your project name
upstream: 'myphpproject:php'
Listing 40-3
1
2
3
4
# .platform/services.yaml
mysql:
type: mysql
disk: 2048
An example of these configurations can be found on GitHub5. The list of available services6 can be found
on the Platform.sh documentation.
Configure Database Access

Platform.sh overrides your database specific configuration via importing the following file (it's your role
to add this file to your code base):
Listing 40-4
1
2
3
4
5
6
7
8
9
// app/config/parameters_platform.php
<?php
$relationships = getenv("PLATFORM_RELATIONSHIPS");
if (!$relationships) {
return;
}
$relationships = json_decode(base64_decode($relationships), true);
5. https://github.com/platformsh/platformsh-examples
6. https://docs.platform.sh/reference/configuration-files/#configure-services

10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
foreach ($relationships['database'] as $endpoint) {

if (empty($endpoint['query']['is_master'])) {
continue;
}
$container->setParameter('database_driver', 'pdo_' . $endpoint['scheme']);
$container->setParameter('database_host', $endpoint['host']);
$container->setParameter('database_port', $endpoint['port']);
$container->setParameter('database_name', $endpoint['path']);
$container->setParameter('database_user', $endpoint['username']);
$container->setParameter('database_password', $endpoint['password']);
$container->setParameter('database_path', '');
}
# Store session into /tmp.

ini_set('session.save_path', '/tmp/sessions');
Make sure this file is listed in your imports:

Listing 40-5
1
2
3
imports:
- { resource: parameters_platform.php }
Deploy your Application

Now you need to add a remote to Platform.sh in your Git repository (copy the command that you see on
the Platform.sh web UI):
Listing 40-6
$ git remote add platform [PROJECT-ID]@git.[CLUSTER].platform.sh:[PROJECT-ID].git
PROJECT-ID
Unique identifier of your project. Something like kjh43kbobssae

CLUSTER
Server location where your project is deployed. It can be eu or us

Commit the Platform.sh specific files created in the previous section:
Listing 40-7
1
2
3
$ git add .platform.app.yaml .platform/*

$ git add app/config/config.yml app/config/parameters_platform.php
$ git commit -m "Adding Platform.sh configuration files."
Push your code base to the newly added remote:

Listing 40-8
$ git push platform master
That's it! Your application is being deployed on Platform.sh and you'll soon be able to access it in your
browser.
Every code change that you do from now on will be pushed to Git in order to redeploy your environment
on Platform.sh.
More information about migrating your database and files7 can be found on the Platform.sh
documentation.
7. https://docs.platform.sh/toolstacks/php/symfony/migrate-existing-site/

Deploy a new Site

You can start a new Platform.sh project8. Choose the development plan and go through the checkout
process.
Once your project is ready, give it a name and choose: Create a new site. Choose the Symfony stack and
a starting point such as Standard.
That's it! Your Symfony application will be bootstrapped and deployed. You'll soon be able to see it in
your browser.
8. https://marketplace.commerceguys.com/platform/buy-now

Chapter 41
Deploying to fortrabbit
This step-by-step cookbook describes how to deploy a Symfony web application to fortrabbit1. You can
read more about using Symfony with fortrabbit on the official fortrabbit Symfony install guide2.
Setting up fortrabbit
Before getting started, you should have done a few things on the fortrabbit side:
Sign up3;
Add an SSH key to your Account (to deploy via Git);
Create an App.
Preparing your Application

You don't need to change any code to deploy a Symfony application to fortrabbit. But it requires some
minor tweaks to its configuration.
Configure Logging
Per default Symfony logs to a file. Modify the app/config/config_prod.yml file to redirect it to
error_log4:
Listing 41-1
1
2
3
4
5
6
monolog:
# ...
handlers:
nested:
type: error_log
1. https://www.fortrabbit.com
2. https://help.fortrabbit.com/install-symfony
3. https://dashboard.fortrabbit.com
4. http://php.net/manual/en/function.error-log.php

Chapter 41: Deploying to fortrabbit | 156
Configuring Database Access & Session Handler

You can use the fortrabbit App Secrets to attain your database credentials. Create the file app/config/
config_prod_secrets.php with the following contents:
Listing 41-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
// get the path to the secrects.json file

$secrets = getenv("APP_SECRETS")
if (!$secrets) {
return;
}
// read the file and decode json to an array
$secrets = json_decode(file_get_contents($secrets), true);
// set database parameters to the container
if (isset($secrets['MYSQL'])) {
$container->setParameter('database_driver', 'pdo_mysql');
$container->setParameter('database_host', $secrets['MYSQL']['HOST']);
$container->setParameter('database_name', $secrets['MYSQL']['DATABASE']);
$container->setParameter('database_user', $secrets['MYSQL']['USER']);
$container->setParameter('database_password', $secrets['MYSQL']['PASSWORD']);
}
// check if the Memcache component is present
if (isset($secrets['MEMCACHE'])) {
$memcache = $secrets['MEMCACHE'];
$handlers = array();
foreach (range(1, $memcache['COUNT']) as $num) {
$handlers[] = $memcache['HOST'.$num].':'.$memcache['PORT'.$num];
}
// apply ini settings

ini_set('session.save_handler', 'memcached');
ini_set('session.save_path', implode(',', $handlers));
if ("2" === $memcache['COUNT']) {
ini_set('memcached.sess_number_of_replicas', 1);
ini_set('memcached.sess_consistent_hash', 1);
ini_set('memcached.sess_binary', 1);
}
}
Make sure this file is imported into the main config file:
Listing 41-3
1
2
3
4
5
6
7
8
9
10
11
imports:
- { resource: config_prod_secrets.php }
# ..
framework:
session:
# set handler_id to null to use default session handler from php.ini (memcached)
handler_id: ~
# ..
Configuring the Environment in the Dashboard

PHP Settings
The PHP version and enabled extensions are configuable under the PHP settings of your App within the
fortrabbit Dashboard.

Environment Variables
Set the SYMFONY_ENV environment variable to prod to make sure the right config files get loaded. ENV
vars are configuable in fortrabbit Dashboard as well.
Document Root
The document root is configuable for every custom domain you setup for your App. The default is
/htdocs, but for Symfony you probably want to change it to /htdocs/web. You also do so in the
fortrabbit Dashboard under Domain settings.
Deploying to fortrabbit
It is assumed that your codebase is under version-control with Git and dependencies are managed with
Composer (locally).
Every time you push to fortrabbit composer install runs before your code gets deployed. To finetune the
deployment behavior put a fortrabbit.yml5. deployment file (optional) in the project root.
Add fortrabbit as a (additional) Git remote and add your configuration changes:
Listing 41-4
1
2
3
$ git remote add fortrabbit [email protected]:<your-app>.git

$ git add composer.json composer.lock
$ git add app/config/config_prod_secrets.php
Commit and push

Listing 41-5
1
2
$ git commit -m 'fortrabbit config'

$ git push fortrabbit master -u
Replace <your-app> with the name of your fortrabbit App.
Listing 41-6
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
Commit received, starting build of branch master
B U I L D
Checksum:
def1bb29911a62de26b1ddac6ef97fc76a5c647b
Deployment file:
fortrabbit.yml
Pre-script:
not found
0ms
Composer:
- - Loading composer repositories with package information
Installing dependencies (including require-dev) from lock file
Nothing to install or update
Generating autoload files
5. https://help.fortrabbit.com/deployment-file-v2

24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
- - 172ms
Post-script:
not found
0ms
R E L E A S E
Packaging:
930ms
Revision:
1455788127289043421.def1bb29911a62de26b1ddac6ef97fc76a5c647b
Size:
9.7MB
Uploading:
500ms
Build & release done in 1625ms, now queued for final distribution.
The first git push takes much longer as all composer dependencies get downloaded. All
subsequent deploys are done within seconds.
That's it! Your application is being deployed on fortrabbit. More information about database migrations
and tunneling6 can be found in the fortrabbit documentation.
6. https://help.fortrabbit.com/install-symfony-2#toc-migrate-amp-other-database-commands

Chapter 42
How to use Doctrine Extensions:

Timestampable, Sluggable, Translatable, etc.
Doctrine2 is very flexible, and the community has already created a series of useful Doctrine extensions
to help you with common entity-related tasks.
One library in particular - the DoctrineExtensions1 library - provides integration functionality for
Sluggable2, Translatable3, Timestampable4, Loggable5, Tree6 and Sortable7 behaviors.
The usage for each of these extensions is explained in that repository.
However, to install/activate each extension you must register and activate an Event Listener. To do this,
you have two options:
1. Use the StofDoctrineExtensionsBundle8, which integrates the above library.
2. Implement this services directly by following the documentation for integration with Symfony:
Install Gedmo Doctrine2 extensions in Symfony29
1. https://github.com/Atlantic18/DoctrineExtensions
2. https://github.com/Atlantic18/DoctrineExtensions/blob/master/doc/sluggable.md
3. https://github.com/Atlantic18/DoctrineExtensions/blob/master/doc/translatable.md
4. https://github.com/Atlantic18/DoctrineExtensions/blob/master/doc/timestampable.md
5. https://github.com/Atlantic18/DoctrineExtensions/blob/master/doc/loggable.md
6. https://github.com/Atlantic18/DoctrineExtensions/blob/master/doc/tree.md
7. https://github.com/Atlantic18/DoctrineExtensions/blob/master/doc/sortable.md
8. https://github.com/stof/StofDoctrineExtensionsBundle
9. https://github.com/Atlantic18/DoctrineExtensions/blob/master/doc/symfony2.md

Chapter 42: How to use Doctrine Extensions: Timestampable, Sluggable, Translatable, etc. | 160
Chapter 43
How to Register Event Listeners and

Subscribers
Doctrine packages a rich event system that fires events when almost anything happens inside the
system. For you, this means that you can create arbitrary services and tell Doctrine to notify those
objects whenever a certain action (e.g. prePersist) happens within Doctrine. This could be useful, for
example, to create an independent search index whenever an object in your database is saved.
Doctrine defines two types of objects that can listen to Doctrine events: listeners and subscribers. Both
are very similar, but listeners are a bit more straightforward. For more, see The Event System1 on
Doctrine's website.
The Doctrine website also explains all existing events that can be listened to.
Configuring the Listener/Subscriber

To register a service to act as an event listener or subscriber you just have to tag it with the appropriate
name. Depending on your use-case, you can hook a listener into every DBAL connection and ORM entity
manager or just into one specific DBAL connection and all the entity managers that use this connection.
Listing 43-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
doctrine:
dbal:
default_connection: default
connections:
default:
driver: pdo_sqlite
memory: true
services:
my.listener:
class: AppBundle\EventListener\SearchIndexer
tags:
- { name: doctrine.event_listener, event: postPersist }
my.listener2:
class: AppBundle\EventListener\SearchIndexer2
1. http://docs.doctrine-project.org/projects/doctrine-orm/en/latest/reference/events.html

Chapter 43: How to Register Event Listeners and Subscribers | 161
16
17
18
19
20
21
tags:
- { name: doctrine.event_listener, event: postPersist, connection: default }
my.subscriber:
class: AppBundle\EventListener\SearchIndexerSubscriber
tags:
- { name: doctrine.event_subscriber, connection: default }
Creating the Listener Class

In the previous example, a service my.listener was configured as a Doctrine listener on the event
postPersist. The class behind that service must have a postPersist method, which will be called
when the event is dispatched:
Listing 43-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
// src/AppBundle/EventListener/SearchIndexer.php
use Doctrine\ORM\Event\LifecycleEventArgs;
use AppBundle\Entity\Product;
class SearchIndexer
{
public function postPersist(LifecycleEventArgs $args)
{
// only act on some "Product" entity

if (!$entity instanceof Product) {
return;
}
$entityManager = $args->getEntityManager();
// ... do something with the Product
}
}
In each event, you have access to a LifecycleEventArgs object, which gives you access to both the
entity object of the event and the entity manager itself.
One important thing to notice is that a listener will be listening for all entities in your application. So, if
you're interested in only handling a specific type of entity (e.g. a Product entity but not a BlogPost
entity), you should check for the entity's class type in your method (as shown above).
In Doctrine 2.4, a feature called Entity Listeners was introduced. It is a lifecycle listener class used
for an entity. You can read about it in the Doctrine Documentation2.
Creating the Subscriber Class

A Doctrine event subscriber must implement the Doctrine\Common\EventSubscriber interface
and have an event method for each event it subscribes to:
Listing 43-3
1
2
3
4
5
// src/AppBundle/EventListener/SearchIndexerSubscriber.php
use Doctrine\Common\EventSubscriber;
use Doctrine\ORM\Event\LifecycleEventArgs;
2. http://docs.doctrine-project.org/projects/doctrine-orm/en/latest/reference/events.html#entity-listeners

6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
// for Doctrine 2.4: Doctrine\Common\Persistence\Event\LifecycleEventArgs;

use AppBundle\Entity\Product;
class SearchIndexerSubscriber implements EventSubscriber
{
public function getSubscribedEvents()
{
return array(
'postPersist',
'postUpdate',
);
}
public function postUpdate(LifecycleEventArgs $args)
{
$this->index($args);
}
public function postPersist(LifecycleEventArgs $args)
{
$this->index($args);
}
public function index(LifecycleEventArgs $args)
{
// perhaps you only want to act on some "Product" entity

if ($entity instanceof Product) {
$entityManager = $args->getEntityManager();
// ... do something with the Product
}
}
}
Doctrine event subscribers can not return a flexible array of methods to call for the events like the
Symfony event subscriber can. Doctrine event subscribers must return a simple array of the event
names they subscribe to. Doctrine will then expect methods on the subscriber with the same name
as each subscribed event, just as when using an event listener.
For a full reference, see chapter The Event System3 in the Doctrine documentation.
3. http://docs.doctrine-project.org/projects/doctrine-orm/en/latest/reference/events.html

Chapter 44
How to Use Doctrine DBAL
This article is about the Doctrine DBAL. Typically, you'll work with the higher level Doctrine ORM
layer, which simply uses the DBAL behind the scenes to actually communicate with the database.
To read more about the Doctrine ORM, see "Databases and Doctrine".
The Doctrine1 Database Abstraction Layer (DBAL) is an abstraction layer that sits on top of PDO2 and
offers an intuitive and flexible API for communicating with the most popular relational databases. In
other words, the DBAL library makes it easy to execute queries and perform other database actions.
Read the official Doctrine DBAL Documentation3 to learn all the details and capabilities of Doctrine's
DBAL library.
To get started, configure the database connection parameters:

Listing 44-1
doctrine:
dbal:
driver:
pdo_mysql
dbname:
Symfony
user:
root
password: null
charset: UTF8
server_version: 5.6
1
2
3
4
5
6
7
8
9
For full DBAL configuration options, or to learn how to configure multiple connections, see Doctrine
DBAL Configuration.
You can then access the Doctrine DBAL connection by accessing the database_connection service:
Listing 44-2
1
2
3
class UserController extends Controller

{
public function indexAction()
1. http://www.doctrine-project.org
2. http://www.php.net/pdo
3. http://docs.doctrine-project.org/projects/doctrine-dbal/en/latest/index.html

Chapter 44: How to Use Doctrine DBAL | 164
4
5
6
7
8
9
10
{
$conn = $this->get('database_connection');
$users = $conn->fetchAll('SELECT * FROM users');
// ...
}
}
Registering custom Mapping Types

You can register custom mapping types through Symfony's configuration. They will be added to all
configured connections. For more information on custom mapping types, read Doctrine's Custom
Mapping Types4 section of their documentation.
Listing 44-3
1
2
3
4
5
6
doctrine:
dbal:
types:
custom_first: AppBundle\Type\CustomFirst
custom_second: AppBundle\Type\CustomSecond
Registering custom Mapping Types in the SchemaTool

The SchemaTool is used to inspect the database to compare the schema. To achieve this task, it needs to
know which mapping type needs to be used for each database types. Registering new ones can be done
through the configuration.
Now, map the ENUM type (not supported by DBAL by default) to the string mapping type:
Listing 44-4
1
2
3
4
5
doctrine:
dbal:
mapping_types:
enum: string
4. http://docs.doctrine-project.org/projects/doctrine-dbal/en/latest/reference/types.html#custom-mapping-types

Chapter 44: How to Use Doctrine DBAL | 165
Chapter 45
How to Generate Entities from an Existing

Database
When starting work on a brand new project that uses a database, two different situations comes
naturally. In most cases, the database model is designed and built from scratch. Sometimes, however,
you'll start with an existing and probably unchangeable database model. Fortunately, Doctrine comes
with a bunch of tools to help generate model classes from your existing database.
As the Doctrine tools documentation1 says, reverse engineering is a one-time process to get started on
a project. Doctrine is able to convert approximately 70-80% of the necessary mapping information
based on fields, indexes and foreign key constraints. Doctrine can't discover inverse associations,
inheritance types, entities with foreign keys as primary keys or semantical operations on associations
such as cascade or lifecycle events. Some additional work on the generated entities will be necessary
afterwards to design each to fit your domain model specificities.
This tutorial assumes you're using a simple blog application with the following two tables: blog_post
and blog_comment. A comment record is linked to a post record thanks to a foreign key constraint.
Listing 45-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
CREATE TABLE `blog_post` (

ìd` bigint(20) NOT NULL AUTO_INCREMENT,
`title` varchar(100) COLLATE utf8_unicode_ci NOT NULL,
`content` longtext COLLATE utf8_unicode_ci NOT NULL,
`created_at` datetime NOT NULL,
PRIMARY KEY (ìd`)
) ENGINE=InnoDB AUTO_INCREMENT=1 DEFAULT CHARSET=utf8 COLLATE=utf8_unicode_ci;
CREATE TABLE `blog_comment` (
ìd` bigint(20) NOT NULL AUTO_INCREMENT,
`post_id` bigint(20) NOT NULL,
àuthor` varchar(20) COLLATE utf8_unicode_ci NOT NULL,
`content` longtext COLLATE utf8_unicode_ci NOT NULL,
`created_at` datetime NOT NULL,
PRIMARY KEY (ìd`),
KEY `blog_comment_post_id_idx` (`post_id`),
1. http://docs.doctrine-project.org/projects/doctrine-orm/en/latest/reference/tools.html#reverse-engineering

Chapter 45: How to Generate Entities from an Existing Database | 166
17
18
CONSTRAINT `blog_post_id` FOREIGN KEY (`post_id`) REFERENCES `blog_post` (ìd`) ON DELETE CASCADE
) ENGINE=InnoDB AUTO_INCREMENT=1 DEFAULT CHARSET=utf8 COLLATE=utf8_unicode_ci;
Before diving into the recipe, be sure your database connection parameters are correctly setup in the
app/config/parameters.yml file (or wherever your database configuration is kept) and that you
have initialized a bundle that will host your future entity class. In this tutorial it's assumed that an
AcmeBlogBundle exists and is located under the src/Acme/BlogBundle folder.
The first step towards building entity classes from an existing database is to ask Doctrine to introspect
the database and generate the corresponding metadata files. Metadata files describe the entity class to
generate based on table fields.
Listing 45-2
$ php app/console doctrine:mapping:import --force AcmeBlogBundle xml
This command line tool asks Doctrine to introspect the database and generate the XML metadata
files under the src/Acme/BlogBundle/Resources/config/doctrine folder of your bundle. This
generates two files: BlogPost.orm.xml and BlogComment.orm.xml.
It's also possible to generate the metadata files in YAML format by changing the last argument to
yml.
The generated BlogPost.orm.xml metadata file looks as follows:

Listing 45-3
1
2
3
4
5
6
7
8
9
10
11
<?xml version="1.0" encoding="utf-8"?>

<doctrine-mapping xmlns="http://doctrine-project.org/schemas/orm/doctrine-mapping"
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://doctrine-project.org/schemas/
orm/doctrine-mapping http://doctrine-project.org/schemas/orm/doctrine-mapping.xsd">
<entity name="Acme\BlogBundle\Entity\BlogPost" table="blog_post">
<id name="id" type="bigint" column="id">
<generator strategy="IDENTITY"/>
</id>
<field name="title" type="string" column="title" length="100" nullable="false"/>
<field name="content" type="text" column="content" nullable="false"/>
<field name="createdAt" type="datetime" column="created_at" nullable="false"/>
</entity>
</doctrine-mapping>
Once the metadata files are generated, you can ask Doctrine to build related entity classes by executing
the following two commands.
Listing 45-4
1
2
$ php app/console doctrine:mapping:convert annotation ./src

$ php app/console doctrine:generate:entities AcmeBlogBundle
The first command generates entity classes with annotation mappings. But if you want to use YAML or
XML mapping instead of annotations, you should execute the second command only.
If you want to use annotations, you must remove the XML (or YAML) files after running these two
commands. This is necessary as it is not possible to mix mapping configuration formats
For example, the newly created BlogComment entity class looks as follow:
Listing 45-5
1
2
3
4
5
6
// src/Acme/BlogBundle/Entity/BlogComment.php
namespace Acme\BlogBundle\Entity;
/**

7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
* Acme\BlogBundle\Entity\BlogComment
*
* @ORM\Table(name="blog_comment")
* @ORM\Entity
*/
class BlogComment
{
/**
* @var integer $id
*
* @ORM\Column(name="id", type="bigint")
* @ORM\Id
* @ORM\GeneratedValue(strategy="IDENTITY")
*/
private $id;
/**
* @var string $author
*
* @ORM\Column(name="author", type="string", length=100, nullable=false)
*/
private $author;
/**
* @var text $content
*
* @ORM\Column(name="content", type="text", nullable=false)
*/
private $content;
/**
* @var datetime $createdAt
*
* @ORM\Column(name="created_at", type="datetime", nullable=false)
*/
private $createdAt;
/**
* @var BlogPost
*
* @ORM\ManyToOne(targetEntity="BlogPost")
* @ORM\JoinColumn(name="post_id", referencedColumnName="id")
*/
private $post;
}
As you can see, Doctrine converts all table fields to pure private and annotated class properties. The most
impressive thing is that it also discovered the relationship with the BlogPost entity class based on the
foreign key constraint. Consequently, you can find a private $post property mapped with a BlogPost
entity in the BlogComment entity class.
If you want to have a one-to-many relationship, you will need to add it manually into the entity or
to the generated XML or YAML files. Add a section on the specific entities for one-to-many defining
the inversedBy and the mappedBy pieces.
The generated entities are now ready to be used. Have fun!

Chapter 46
How to Work with multiple Entity Managers

and Connections
You can use multiple Doctrine entity managers or connections in a Symfony application. This is
necessary if you are using different databases or even vendors with entirely different sets of entities. In
other words, one entity manager that connects to one database will handle some entities while another
entity manager that connects to another database might handle the rest.
Using multiple entity managers is pretty easy, but more advanced and not usually required. Be sure
you actually need multiple entity managers before adding in this layer of complexity.
The following configuration code shows how you can configure two entity managers:
Listing 46-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
doctrine:
dbal:
default_connection: default
connections:
default:
driver:
pdo_mysql
host:
'%database_host%'
port:
'%database_port%'
dbname:
'%database_name%'
user:
'%database_user%'
password: '%database_password%'
charset: UTF8
customer:
driver:
pdo_mysql
host:
'%database_host2%'
port:
'%database_port2%'
dbname:
'%database_name2%'
user:
'%database_user2%'
password: '%database_password2%'
charset: UTF8
orm:
default_entity_manager: default
entity_managers:
default:

Chapter 46: How to Work with multiple Entity Managers and Connections | 169
26
27
28
29
30
31
32
33
connection: default
mappings:
AppBundle: ~
AcmeStoreBundle: ~
customer:
connection: customer
mappings:
AcmeCustomerBundle: ~
In this case, you've defined two entity managers and called them default and customer. The
default entity manager manages entities in the AppBundle and AcmeStoreBundle, while the
customer entity manager manages entities in the AcmeCustomerBundle. You've also defined two
connections, one for each entity manager.
When working with multiple connections and entity managers, you should be explicit about which
configuration you want. If you do omit the name of the connection or entity manager, the default
(i.e. default) is used.
When working with multiple connections to create your databases:

Listing 46-2
# Play only with "default" connection

$ php app/console doctrine:database:create
1
2
3
4
5
# Play only with "customer" connection

$ php app/console doctrine:database:create --connection=customer
When working with multiple entity managers to update your schema:

Listing 46-3
# Play only with "default" mappings

1
2
3
4
5
# Play only with "customer" mappings

$ php app/console doctrine:schema:update --force --em=customer
If you do omit the entity manager's name when asking for it, the default entity manager (i.e. default) is
returned:
Listing 46-4
1
2
3
4
5
6
7
8
9
10
11
12
13
14

{
{
// All three return the "default" entity manager
$em = $this->get('doctrine')->getManager();
$em = $this->get('doctrine')->getManager('default');
$em = $this->get('doctrine.orm.default_entity_manager');
// Both of these return the "customer" entity manager

$customerEm = $this->get('doctrine')->getManager('customer');
$customerEm = $this->get('doctrine.orm.customer_entity_manager');
}
}
You can now use Doctrine just as you did before - using the default entity manager to persist and fetch
entities that it manages and the customer entity manager to persist and fetch its entities.
The same applies to repository calls:
Listing 46-5
1
2
3
4
5

{
{
// Retrieves a repository managed by the "default" em

6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
$products = $this->get('doctrine')
->getRepository('AcmeStoreBundle:Product')
->findAll()
;
// Explicit way to deal with the "default" em

$products = $this->get('doctrine')
->getRepository('AcmeStoreBundle:Product', 'default')
->findAll()
;
// Retrieves a repository managed by the "customer" em
$customers = $this->get('doctrine')
->getRepository('AcmeCustomerBundle:Customer', 'customer')
->findAll()
;
}
}

Chapter 47
How to Register custom DQL Functions

Doctrine allows you to specify custom DQL functions. For more information on this topic, read
Doctrine's cookbook article "DQL User Defined Functions1".
In Symfony, you can register your custom DQL functions as follows:
Listing 47-1
1
2
3
4
5
6
7
8
9
10
11
12
doctrine:
orm:
# ...
dql:
string_functions:
test_string: AppBundle\DQL\StringFunction
second_string: AppBundle\DQL\SecondStringFunction
numeric_functions:
test_numeric: AppBundle\DQL\NumericFunction
datetime_functions:
test_datetime: AppBundle\DQL\DatetimeFunction
1. http://docs.doctrine-project.org/projects/doctrine-orm/en/latest/cookbook/dql-user-defined-functions.html

Chapter 47: How to Register custom DQL Functions | 172
Chapter 48
How to Define Relationships with Abstract

Classes and Interfaces
One of the goals of bundles is to create discreet bundles of functionality that do not have many (if any)
dependencies, allowing you to use that functionality in other applications without including unnecessary
items.
Doctrine 2.2 includes a new utility called the ResolveTargetEntityListener, that functions by
intercepting certain calls inside Doctrine and rewriting targetEntity parameters in your metadata
mapping at runtime. It means that in your bundle you are able to use an interface or abstract class in your
mappings and expect correct mapping to a concrete entity at runtime.
This functionality allows you to define relationships between different entities without making them hard
dependencies.
Background
Suppose you have an InvoiceBundle which provides invoicing functionality and a CustomerBundle that
contains customer management tools. You want to keep these separated, because they can be used in
other systems without each other, but for your application you want to use them together.
In this case, you have an Invoice entity with a relationship to a non-existent object, an
InvoiceSubjectInterface. The goal is to get the ResolveTargetEntityListener to replace
any mention of the interface with a real object that implements that interface.
Set up
This article uses the following two basic entities (which are incomplete for brevity) to explain how to set
up and use the ResolveTargetEntityListener.
A Customer entity:
Listing 48-1

Chapter 48: How to Define Relationships with Abstract Classes and Interfaces | 173
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
// src/Acme/AppBundle/Entity/Customer.php
namespace Acme\AppBundle\Entity;
use Acme\CustomerBundle\Entity\Customer as BaseCustomer;
use Acme\InvoiceBundle\Model\InvoiceSubjectInterface;
/**
* @ORM\Entity
* @ORM\Table(name="customer")
*/
class Customer extends BaseCustomer implements InvoiceSubjectInterface
{
// In this example, any methods defined in the InvoiceSubjectInterface
// are already implemented in the BaseCustomer
}
An Invoice entity:
Listing 48-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
// src/Acme/InvoiceBundle/Entity/Invoice.php
namespace Acme\InvoiceBundle\Entity;
use Doctrine\ORM\Mapping AS ORM;
use Acme\InvoiceBundle\Model\InvoiceSubjectInterface;
/**
* Represents an Invoice.
*
* @ORM\Entity
* @ORM\Table(name="invoice")
*/
class Invoice
{
/**
* @ORM\ManyToOne(targetEntity="Acme\InvoiceBundle\Model\InvoiceSubjectInterface")
* @var InvoiceSubjectInterface
*/
protected $subject;
}
An InvoiceSubjectInterface:
Listing 48-3
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
// src/Acme/InvoiceBundle/Model/InvoiceSubjectInterface.php
namespace Acme\InvoiceBundle\Model;
/**
* An interface that the invoice Subject object should implement.
* In most circumstances, only a single object should implement
* this interface as the ResolveTargetEntityListener can only
* change the target to a single object.
*/
interface InvoiceSubjectInterface
{
// List any additional methods that your InvoiceBundle
// will need to access on the subject so that you can
// be sure that you have access to those methods.
/**
* @return string
*/
public function getName();
}
Next, you need to configure the listener, which tells the DoctrineBundle about the replacement:
Listing 48-4

1
2
3
4
5
6
7
doctrine:
# ...
orm:
# ...
resolve_target_entities:
Acme\InvoiceBundle\Model\InvoiceSubjectInterface: Acme\AppBundle\Entity\Customer
Final Thoughts
With the ResolveTargetEntityListener, you are able to decouple your bundles, keeping them
usable by themselves, but still being able to define relationships between different objects. By using this
method, your bundles will end up being easier to maintain independently.

Chapter 49
How to Provide Model Classes for several

Doctrine Implementations
When building a bundle that could be used not only with Doctrine ORM but also the CouchDB ODM,
MongoDB ODM or PHPCR ODM, you should still only write one model class. The Doctrine bundles
provide a compiler pass to register the mappings for your model classes.
For non-reusable bundles, the easiest option is to put your model classes in the default locations:
Entity for the Doctrine ORM or Document for one of the ODMs. For reusable bundles, rather
than duplicate model classes just to get the auto-mapping, use the compiler pass.
New in version 2.3: The base mapping compiler pass was introduced in Symfony 2.3. The Doctrine
bundles support it from DoctrineBundle >= 1.3.0, MongoDBBundle >= 3.0.0, PHPCRBundle >= 1.0.0
and the (unversioned) CouchDBBundle supports the compiler pass since the CouchDB Mapping
Compiler Pass pull request1 was merged.
New in version 2.6: Support for defining namespace aliases was introduced in Symfony 2.6. It is safe
to define the aliases with older versions of Symfony as the aliases are the last argument to
createXmlMappingDriver and are ignored by PHP if that argument doesn't exist.
In your bundle class, write the following code to register the compiler pass. This one is written for the
CmfRoutingBundle, so parts of it will need to be adapted for your case:
Listing 49-1
1
2
3
4
5
6
7
8
9
10
11
12
use
use
use
use
Doctrine\Bundle\DoctrineBundle\DependencyInjection\Compiler\DoctrineOrmMappingsPass;
Doctrine\Bundle\MongoDBBundle\DependencyInjection\Compiler\DoctrineMongoDBMappingsPass;
Doctrine\Bundle\CouchDBBundle\DependencyInjection\Compiler\DoctrineCouchDBMappingsPass;
Doctrine\Bundle\PHPCRBundle\DependencyInjection\Compiler\DoctrinePhpcrMappingsPass;
class CmfRoutingBundle extends Bundle

{
public function build(ContainerBuilder $container)
{
parent::build($container);
// ...
1. https://github.com/doctrine/DoctrineCouchDBBundle/pull/27

Chapter 49: How to Provide Model Classes for several Doctrine Implementations | 176
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
$modelDir = realpath(__DIR__.'/Resources/config/doctrine/model');
$mappings = array(
$modelDir => 'Symfony\Cmf\RoutingBundle\Model',
);
$ormCompilerClass =
'Doctrine\Bundle\DoctrineBundle\DependencyInjection\Compiler\DoctrineOrmMappingsPass';
if (class_exists($ormCompilerClass)) {
$container->addCompilerPass(
DoctrineOrmMappingsPass::createXmlMappingDriver(
$mappings,
array('cmf_routing.model_manager_name'),
'cmf_routing.backend_type_orm',
array('CmfRoutingBundle' => 'Symfony\Cmf\RoutingBundle\Model')
));
}
$mongoCompilerClass =
'Doctrine\Bundle\MongoDBBundle\DependencyInjection\Compiler\DoctrineMongoDBMappingsPass';
if (class_exists($mongoCompilerClass)) {
DoctrineMongoDBMappingsPass::createXmlMappingDriver(
$mappings,
'cmf_routing.backend_type_mongodb',
));
}
$couchCompilerClass =
'Doctrine\Bundle\CouchDBBundle\DependencyInjection\Compiler\DoctrineCouchDBMappingsPass';
if (class_exists($couchCompilerClass)) {
DoctrineCouchDBMappingsPass::createXmlMappingDriver(
$mappings,
'cmf_routing.backend_type_couchdb',
));
}
$phpcrCompilerClass =
'Doctrine\Bundle\PHPCRBundle\DependencyInjection\Compiler\DoctrinePhpcrMappingsPass';
if (class_exists($phpcrCompilerClass)) {
DoctrinePhpcrMappingsPass::createXmlMappingDriver(
$mappings,
'cmf_routing.backend_type_phpcr',
));
}
}
}
Note the class_exists2 check. This is crucial, as you do not want your bundle to have a hard
dependency on all Doctrine bundles but let the user decide which to use.
The compiler pass provides factory methods for all drivers provided by Doctrine: Annotations, XML,
Yaml, PHP and StaticPHP. The arguments are:
A map/hash of absolute directory path to namespace;
An array of container parameters that your bundle uses to specify the name of the Doctrine manager
that it is using. In the example above, the CmfRoutingBundle stores the manager name that's being
used under the cmf_routing.model_manager_name parameter. The compiler pass will append the parameter
2. http://php.net/manual/en/function.class-exists.php

Doctrine is using to specify the name of the default manager. The first parameter found is used and
the mappings are registered with that manager;
An optional container parameter name that will be used by the compiler pass to determine if this
Doctrine type is used at all. This is relevant if your user has more than one type of Doctrine bundle
installed, but your bundle is only used with one type of Doctrine;
A map/hash of aliases to namespace. This should be the same convention used by Doctrine automapping. In the example above, this allows the user to call $om->getRepository('CmfRoutingBundle:Route').
The factory method is using the SymfonyFileLocator of Doctrine, meaning it will only see XML
and YML mapping files if they do not contain the full namespace as the filename. This is by design:
the SymfonyFileLocator simplifies things by assuming the files are just the "short" version of the
class as their filename (e.g. BlogPost.orm.xml)
If you also need to map a base class, you can register a compiler pass with the
DefaultFileLocator like this. This code is taken from the DoctrineOrmMappingsPass and
adapted to use the DefaultFileLocator instead of the SymfonyFileLocator:
Listing 49-2
1
2
3
4
5
6
7
8
9
10
11
12
13
private function buildMappingCompilerPass()

{
$arguments = array(array(realpath(__DIR__ . '/Resources/config/doctrine-base')), '.orm.xml');
$locator = new Definition('Doctrine\Common\Persistence\Mapping\Driver\DefaultFileLocator',
$arguments);
$driver = new Definition('Doctrine\ORM\Mapping\Driver\XmlDriver', array($locator));
return new DoctrineOrmMappingsPass(
$driver,
array('Full\Namespace'),
array('your_bundle.manager_name'),
'your_bundle.orm_enabled'
);
}
Note that you do not need to provide a namespace alias unless your users are expected to ask
Doctrine for the base classes.
Now place your mapping file into /Resources/config/doctrine-base with the fully qualified
class name, separated by . instead of \, for example Other.Namespace.Model.Name.orm.xml.
You may not mix the two as otherwise the SymfonyFileLocator will get confused.
Adjust accordingly for the other Doctrine implementations.

Chapter 50
How to Implement a Simple Registration Form

Creating a registration form is pretty easy - it really means just creating a form that will update some
User model object (a Doctrine entity in this example) and then save it.
The popular FOSUserBundle1 provides a registration form, reset password form and other user
management functionality.
If you don't already have a User entity and a working login system, first start with How to Load Security
Users from the Database (the Entity Provider).
Your User entity will probably at least have the following fields:
username
This will be used for logging in, unless you instead want your user to login via email (in that case,
this field is unnecessary).
email
A nice piece of information to collect. You can also allow users to login via email.
password
The encoded password.

plainPassword
This field is not persisted: (notice no @ORM\Column above it). It temporarily stores the plain password
from the registration form. This field can be validated and is then used to populate the password field.
With some validation added, your class may look something like this:
Listing 50-1
1
2
3
4
5
6
7
// src/AppBundle/Entity/User.php
use
use
use
use
Doctrine\ORM\Mapping as ORM;
Symfony\Component\Validator\Constraints as Assert;
Symfony\Bridge\Doctrine\Validator\Constraints\UniqueEntity;
Symfony\Component\Security\Core\User\UserInterface;

Chapter 50: How to Implement a Simple Registration Form | 179
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
/**
* @ORM\Entity
* @UniqueEntity(fields="email", message="Email already taken")
* @UniqueEntity(fields="username", message="Username already taken")
*/
class User implements UserInterface
{
/**
* @ORM\Id
* @ORM\Column(type="integer")
* @ORM\GeneratedValue(strategy="AUTO")
*/
private $id;
/**
* @ORM\Column(type="string", length=255, unique=true)
* @Assert\NotBlank()
* @Assert\Email()
*/
private $email;
/**
*/
private $username;
/**
* @Assert\Length(max=4096)
*/
private $plainPassword;
/**
* The below length depends on the "algorithm" you use for encoding
* the password, but this works well with bcrypt.
*
* @ORM\Column(type="string", length=64)
*/
private $password;
// other properties and methods
public function getEmail()
{
return $this->email;
}
public function setEmail($email)
{
$this->email = $email;
}
public function getUsername()
{
return $this->username;
}
public function setUsername($username)
{
$this->username = $username;
}
public function getPlainPassword()
{
return $this->plainPassword;
}
public function setPlainPassword($password)
{

79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
$this->plainPassword = $password;
}
public function setPassword($password)
{
$this->password = $password;
}
public function getSalt()
{
// The bcrypt algorithm doesn't require a separate salt.
// You *may* need a real salt if you choose a different encoder.
return null;
}
// other methods, including security methods like getRoles()

}
The UserInterface2 requires a few other methods and your security.yml file needs to be
configured properly to work with the User entity. For a more complete example, see the Entity Provider
article.
Why the 4096 Password Limit?

Notice that the plainPassword field has a max length of 4096 characters. For security purposes
(CVE-2013-57503), Symfony limits the plain password length to 4096 characters when encoding it.
Adding this constraint makes sure that your form will give a validation error if anyone tries a superlong password.
You'll need to add this constraint anywhere in your application where your user submits a plaintext
password (e.g. change password form). The only place where you don't need to worry about this is
your login form, since Symfony's Security component handles this for you.
Create a Form for the Entity

Next, create the form for the User entity:
Listing 50-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
// src/AppBundle/Form/UserType.php
class UserType extends AbstractType
{
{
$builder
->add('email', 'email')
->add('username', 'text')
->add('plainPassword', 'repeated', array(
'type' => 'password',
'first_options' => array('label' => 'Password'),
'second_options' => array('label' => 'Repeat Password'),
)
);
}
2. http://api.symfony.com/2.7/Symfony/Component/Security/Core/User/UserInterface.html
3. https://symfony.com/blog/cve-2013-5750-security-issue-in-fosuserbundle-login-form

23
24
25
26
27
28
29
30
31
32
33
34

{
'data_class' => 'AppBundle\Entity\User',
));
}
{
return 'user';
}
}
There are just three fields: email, username and plainPassword (repeated to confirm the entered
password).
To explore more things about the Form component, read the chapter about forms in the book.
Handling the Form Submission

Next, you need a controller to handle the form rendering and submission. If the form is submitted, the
controller performs the validation and saves the data into the database:
Listing 50-3
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
// src/AppBundle/Controller/RegistrationController.php
use
use
use
use
use
AppBundle\Form\UserType;
AppBundle\Entity\User;
Sensio\Bundle\FrameworkExtraBundle\Configuration\Route;
class RegistrationController extends Controller

{
/**
* @Route("/register", name="user_registration")
*/
public function registerAction(Request $request)
{
// 1) build the form
$user = new User();
$form = $this->createForm(new UserType(), $user);
// 2) handle the submit (will only happen on POST)

if ($form->isSubmitted() && $form->isValid()) {
// 3) Encode the password (you could also do this via Doctrine listener)
$password = $this->get('security.password_encoder')
->encodePassword($user, $user->getPlainPassword());
$user->setPassword($password);
// 4) save the User!
$em = $this->getDoctrine()->getManager();
$em->persist($user);
$em->flush();
// ... do any other work - like sending them an email, etc
// maybe set a "flash" success message for the user
return $this->redirectToRoute('replace_with_some_route');
}

40
41
42
43
44
45
46
'registration/register.html.twig',
array('form' => $form->createView())
);
}
}
To define the algorithm used to encode the password in step 3 configure the encoder in the security
configuration:
Listing 50-4
# app/config/security.yml
security:
encoders:
AppBundle\Entity\User: bcrypt
1
2
3
4
In this case the recommended bcrypt algorithm is used. To learn more about how to encode the users
password have a look into the security chapter.
If you decide to NOT use annotation routing (shown above), then you'll need to create a route to
this controller:
Listing 50-5
1
2
3
4
user_registration:
path:
/register
defaults: { _controller: AppBundle:Registration:register }
Next, create the template:

Listing 50-6
1
2
3
4
5
6
7
8
9
10
{# app/Resources/views/registration/register.html.twig #}
{{ form_row(form.username) }}
{{ form_row(form.email) }}
{{ form_row(form.plainPassword.first) }}
{{ form_row(form.plainPassword.second) }}
<button type="submit">Register!</button>
See How to Customize Form Rendering for more details.
Update your Database Schema

If you've updated the User entity during this tutorial, you have to update your database schema using
this command:
Listing 50-7
That's it! Head to /register to try things out!
Having a Registration form with only Email (no Username)

If you want your users to login via email and you don't need a username, then you can remove it from
your User entity entirely. Instead, make getUsername() return the email property:

Listing 50-8
1
2
3
4
5
6
7
8
9
10
11
12
13
14
// ...
class User implements UserInterface
{
// ...
{
return $this->email;
}
// ...
}
Next, just update the providers section of your security.yml file so that Symfony knows how to
load your users via the email property on login. See Using a Custom Query to Load the User.
Adding a "accept terms" Checkbox

Sometimes, you want a "Do you accept the terms and conditions" checkbox on your registration form.
The only trick is that you want to add this field to your form without adding an unnecessary new
termsAccepted property to your User entity that you'll never need.
To do this, add a termsAccepted field to your form, but set its mapped option to false:
Listing 50-9
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
// src/AppBundle/Form/UserType.php
// ...
use Symfony\Component\Validator\Constraints\IsTrue;
class UserType extends AbstractType
{
{
$builder
->add('email', 'email');
// ...
->add('termsAccepted', 'checkbox', array(
'mapped' => false,
'constraints' => new IsTrue(),
))
);
}
}
The constraints option is also used, which allows us to add validation, even though there is no
termsAccepted property on User.

Chapter 51
How to Use PdoSessionHandler to Store

Sessions in the Database
There was a backwards-compatibility break in Symfony 2.6: the database schema changed slightly.
See Symfony 2.6 Changes for details.
The default Symfony session storage writes the session information to files. Most medium to large
websites use a database to store the session values instead of files, because databases are easier to use and
scale in a multiple web server environment.
Symfony has a built-in solution for database session storage called PdoSessionHandler1. To use it,
you just need to change some parameters in the main configuration file:
Listing 51-1
1
2
3
4
5
6
7
8
9
10
11
12
13
framework:
session:
# ...
handler_id: session.handler.pdo
services:
session.handler.pdo:
class:
Symfony\Component\HttpFoundation\Session\Storage\Handler\PdoSessionHandler
public:
false
arguments:
- 'mysql:dbname=mydatabase'
- { db_username: myuser, db_password: mypassword }
Configuring the Table and Column Names

This will expect a sessions table with a number of different columns. The table name, and all of the
column names, can be configured by passing a second array argument to PdoSessionHandler:
1. http://api.symfony.com/2.7/Symfony/Component/HttpFoundation/Session/Storage/Handler/PdoSessionHandler.html

Chapter 51: How to Use PdoSessionHandler to Store Sessions in the Database | 185
Listing 51-2
1
2
3
4
5
6
7
8
9
services:
# ...
class:
public:
false
arguments:
- 'mysql:dbname=mydatabase'
- { db_table: sessions, db_username: myuser, db_password: mypassword }
New in version 2.6: The db_lifetime_col was introduced in Symfony 2.6. Prior to 2.6, this column
did not exist.
These are parameters that you must configure:
db_table (default sessions):
The name of the session table in your database;
db_id_col (default sess_id):
The name of the id column in your session table (VARCHAR(128));

db_data_col (default sess_data):
The name of the value column in your session table (BLOB);

db_time_col (default sess_time):
The name of the time column in your session table (INTEGER);

db_lifetime_col (default sess_lifetime):
The name of the lifetime column in your session table (INTEGER).
Sharing your Database Connection Information

With the given configuration, the database connection settings are defined for the session storage
connection only. This is OK when you use a separate database for the session data.
But if you'd like to store the session data in the same database as the rest of your project's data, you
can use the connection settings from the parameters.yml file by referencing the database-related
parameters defined there:
Listing 51-3
1
2
3
4
5
6
7
services:
class:
public:
false
arguments:
- 'mysql:host=%database_host%;port=%database_port%;dbname=%database_name%'
- { db_username: '%database_user%', db_password: '%database_password%' }
Preparing the Database to Store Sessions

Before storing sessions in the database, you must create the table that stores the information. The
following sections contain some examples of the SQL statements you may use for your specific database
engine.

Schema Changes needed when Upgrading to Symfony 2.6

If you use the PdoSessionHandler prior to Symfony 2.6 and upgrade, you'll need to make a few
changes to your session table:
A new session lifetime (sess_lifetime by default) integer column needs to be added;
The data column (sess_data by default) needs to be changed to a BLOB type.
Check the SQL statements below for more details.
To keep the old (2.5 and earlier) functionality, change your class name to use
LegacyPdoSessionHandler instead of PdoSessionHandler (the legacy class was added in
Symfony 2.6.2).
MySQL
Listing 51-4
1
2
3
4
5
6
CREATE TABLE `sessions` (

`sess_id` VARBINARY(128) NOT NULL PRIMARY KEY,
`sess_data` BLOB NOT NULL,
`sess_time` INTEGER UNSIGNED NOT NULL,
`sess_lifetime` MEDIUMINT NOT NULL
) COLLATE utf8_bin, ENGINE = InnoDB;
A BLOB column type can only store up to 64 kb. If the data stored in a user's session exceeds this, an
exception may be thrown or their session will be silently reset. Consider using a MEDIUMBLOB if you
need more space.
PostgreSQL
Listing 51-5
1
2
3
4
5
6
CREATE TABLE sessions (

sess_id VARCHAR(128) NOT NULL PRIMARY KEY,
sess_data BYTEA NOT NULL,
sess_time INTEGER NOT NULL,
sess_lifetime INTEGER NOT NULL
);
Microsoft SQL Server

Listing 51-6
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
CREATE TABLE [dbo].[sessions](

[sess_id] [nvarchar](255) NOT NULL,
[sess_data] [ntext] NOT NULL,
[sess_time] [int] NOT NULL,
[sess_lifetime] [int] NOT NULL,
PRIMARY KEY CLUSTERED(
[sess_id] ASC
) WITH (
PAD_INDEX = OFF,
STATISTICS_NORECOMPUTE = OFF,
IGNORE_DUP_KEY = OFF,
ALLOW_ROW_LOCKS = ON,
ALLOW_PAGE_LOCKS = ON
) ON [PRIMARY]
) ON [PRIMARY] TEXTIMAGE_ON [PRIMARY]

If the session data doesn't fit in the data column, it might get truncated by the database engine. To
make matters worse, when the session data gets corrupted, PHP ignores the data without giving a
warning.
If the application stores large amounts of session data, this problem can be solved by increasing the
column size (use BLOB or even MEDIUMBLOB). When using MySQL as the database engine, you can
also enable the strict SQL mode2 to get noticed when such an error happens.
2. https://dev.mysql.com/doc/refman/5.7/en/sql-mode.html

Chapter 52
How to Use MongoDbSessionHandler to Store

Sessions in a MongoDB Database
The default Symfony session storage writes the session information to files. Some medium to large
websites use a NoSQL database called MongoDB to store the session values instead of files, because
databases are easier to use and scale in a multi-webserver environment.
Symfony
has
built-in
solution
for
NoSQL
database
session
storage
called
MongoDbSessionHandler1. MongoDB is an open-source document database that provides high

performance, high availability and automatic scaling. This article assumes that you have already installed
and configured a MongoDB server2. To use it, you just need to change/add some parameters in the main
configuration file:
Listing 52-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
framework:
session:
# ...
handler_id: session.handler.mongo
cookie_lifetime: 2592000 # optional, it is set to 30 days here
gc_maxlifetime: 2592000 # optional, it is set to 30 days here
services:
# ...
mongo_client:
class: MongoClient
# if using a username and password
arguments: ['mongodb://%mongodb_username%:%mongodb_password%@%mongodb_host%:27017']
# if not using a username and password
arguments: ['mongodb://%mongodb_host%:27017']
session.handler.mongo:
class: Symfony\Component\HttpFoundation\Session\Storage\Handler\MongoDbSessionHandler
arguments: ['@mongo_client', '%mongo.session.options%']
The parameters used above should be defined somewhere in your application, often in your main
parameters configuration:
Listing 52-2
1. http://api.symfony.com/2.7/Symfony/Component/HttpFoundation/Session/Storage/Handler/MongoDbSessionHandler.html
2. http://docs.mongodb.org/manual/installation/

Chapter 52: How to Use MongoDbSessionHandler to Store Sessions in a MongoDB Database | 189
1
2
3
4
5
6
7
8
9
parameters:
# ...
mongo.session.options:
database: session_db # your MongoDB database name
collection: session # your MongoDB collection name
mongodb_host: 1.2.3.4 # your MongoDB server's IP
mongodb_username: my_username
mongodb_password: my_password
Setting Up the MongoDB Collection

Because MongoDB uses dynamic collection schemas, you do not need to do anything to initialize your
session collection. However, you may want to add an index to improve garbage collection performance.
From the MongoDB shell3:
Listing 52-3
1
2
use session_db
db.session.ensureIndex( { "expires_at": 1 }, { expireAfterSeconds: 0 } )
3. http://docs.mongodb.org/v2.2/tutorial/getting-started-with-the-mongo-shell/

Chapter 52: How to Use MongoDbSessionHandler to Store Sessions in a MongoDB Database | 190
Chapter 53
Console Commands
The Doctrine2 ORM integration offers several console commands under the doctrine namespace. To
view the command list you can use the list command:
Listing 53-1
$ php app/console list doctrine
A list of available commands will print out. You can find out more information about any of these
commands (or any Symfony command) by running the help command. For example, to get details
about the doctrine:database:create task, run:
Listing 53-2
$ php app/console help doctrine:database:create
Some notable or interesting tasks include:

doctrine:ensure-production-settings - checks to see if the current environment is
configured efficiently for production. This should always be run in the prod environment:
Listing 53-3
$ php app/console doctrine:ensure-production-settings --env=prod
doctrine:mapping:import - allows Doctrine to introspect an existing database and create

mapping information. For more information, see How to Generate Entities from an Existing
Database.
doctrine:mapping:info - tells you all of the entities that Doctrine is aware of and whether or
not there are any basic errors with the mapping.
doctrine:query:dql and doctrine:query:sql - allow you to execute DQL or SQL queries
directly from the command line.

Chapter 53: Console Commands | 191
Chapter 54
How to Send an Email

Sending emails is a classic task for any web application and one that has special complications and
potential pitfalls. Instead of recreating the wheel, one solution to send emails is to use the
SwiftmailerBundle, which leverages the power of the Swift Mailer1 library. This bundle comes with the
Symfony Standard Edition.
Configuration
To use Swift Mailer, you'll need to configure it for your mail server.
Instead of setting up/using your own mail server, you may want to use a hosted mail provider such
as Mandrill2, SendGrid3, Amazon SES4 or others. These give you an SMTP server, username and
password (sometimes called keys) that can be used with the Swift Mailer configuration.
In a standard Symfony installation, some swiftmailer configuration is already included:

Listing 54-1
1
2
3
4
5
6
swiftmailer:
transport: '%mailer_transport%'
host:
'%mailer_host%'
username: '%mailer_user%'
password: '%mailer_password%'
These values (e.g. %mailer_transport%), are reading from the parameters that are set in the
parameters.yml file. You can modify the values in that file, or set the values directly here.
The following configuration attributes are available:
1.
2.
3.
4.
transport (smtp, mail, sendmail,
or gmail)
username
password
http://swiftmailer.org/
https://mandrill.com/
https://sendgrid.com/
http://aws.amazon.com/ses/

Chapter 54: How to Send an Email | 192
host
port
encryption (tls,
or ssl)
auth_mode (plain, login,
or cram-md5)
spool
type
path
(how to queue the messages, file or memory is supported, see How to Spool Emails)
(where to store the messages)
delivery_address
disable_delivery
(an email address where to send ALL emails)

(set to true to disable delivery completely)
Sending Emails
The Swift Mailer library works by creating, configuring and then sending Swift_Message objects. The
"mailer" is responsible for the actual delivery of the message and is accessible via the mailer service.
Overall, sending an email is pretty straightforward:
Listing 54-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29

{
$message = \Swift_Message::newInstance()
->setSubject('Hello Email')
->setFrom('[email protected]')
->setTo('[email protected]')
->setBody(
$this->renderView(
// app/Resources/views/Emails/registration.html.twig
'Emails/registration.html.twig',
),
'text/html'
)
/*
* If you also want to include a plaintext version of the message
->addPart(
$this->renderView(
'Emails/registration.txt.twig',
),
'text/plain'
)
*/
;
$this->get('mailer')->send($message);
return $this->render(...);
}
To keep things decoupled, the email body has been stored in a template and rendered with the
renderView() method. The registration.html.twig template might look something like this:
Listing 54-3
1
2
3
4
5
6
7
8
9
10
11
12
{# app/Resources/views/Emails/registration.html.twig #}
<h3>You did it! You registered!</h3>
Hi {{ name }}! You're successfully registered.
{# example, assuming you have a route named "login" #}

To login, go to: <a href="{{ url('login') }}">...</a>.
Thanks!
{# Makes an absolute URL to the /images/logo.png file #}

<img src="{{ absolute_url(asset('images/logo.png')) }}">

New in version 2.7: The absolute_url() function was introduced in Symfony 2.7. Prior to 2.7, the
asset() function has an argument to enable returning an absolute URL.
The $message object supports many more options, such as including attachments, adding HTML
content, and much more. Fortunately, Swift Mailer covers the topic of Creating Messages5 in great detail
in its documentation.
Several other cookbook articles are available related to sending emails in Symfony:
How to Use Gmail to Send Emails
How to Work with Emails during Development
How to Spool Emails
5. http://swiftmailer.org/docs/messages.html

Chapter 55
How to Use Gmail to Send Emails

During development, instead of using a regular SMTP server to send emails, you might find using Gmail
easier and more practical. The SwiftmailerBundle makes it really easy.
In the development configuration file, change the transport setting to gmail and set the username
and password to the Google credentials:
Listing 55-1
1
2
3
4
5
swiftmailer:
transport: gmail
username: your_gmail_username
password: your_gmail_password
It's more convenient to configure these options in the parameters.yml file:

Listing 55-2
Listing 55-3
1
2
3
4
5
parameters:
# ...
mailer_user:
your_gmail_username
mailer_password: your_gmail_password
1
2
3
4
5
swiftmailer:
transport: gmail
username: '%mailer_user%'
password: '%mailer_password%'
Redefining the Default Configuration Parameters

The gmail transport is simply a shortcut that uses the smtp transport and sets these options:
Option
Value
encryption
ssl

Chapter 55: How to Use Gmail to Send Emails | 195
Option
Value
auth_mode
login
host
smtp.gmail.com
If your application uses tls encryption or oauth authentication, you must override the default options
by defining the encryption and auth_mode parameters.
If your Gmail account uses 2-Step-Verification, you must generate an App password1 and use it as the
value of the mailer_password parameter. You must also ensure that you allow less secure apps to
access your Gmail account2.
See the Swiftmailer configuration reference for more details.
1. https://support.google.com/accounts/answer/185833
2. https://support.google.com/accounts/answer/6010255

Chapter 55: How to Use Gmail to Send Emails | 196
Chapter 56
How to Use the Cloud to Send Emails

Requirements for sending emails from a production system differ from your development setup as you
don't want to be limited in the number of emails, the sending rate or the sender address. Thus, using
Gmail or similar services is not an option. If setting up and maintaining your own reliable mail server
causes you a headache there's a simple solution: Leverage the cloud to send your emails.
This cookbook shows how easy it is to integrate Amazon's Simple Email Service (SES)1 into Symfony.
You can use the same technique for other mail services, as most of the time there is nothing more to
it than configuring an SMTP endpoint for Swift Mailer.
In the Symfony configuration, change the Swift Mailer settings transport, host, port and
encryption according to the information provided in the SES console2. Create your individual SMTP
credentials in the SES console and complete the configuration with the provided username and
password:
Listing 56-1
1
2
3
4
5
6
7
8
swiftmailer:
transport: smtp
host:
email-smtp.us-east-1.amazonaws.com
port:
587 # different ports are available, see SES console
encryption: tls # TLS encryption is required
username:
AWS_SES_SMTP_USERNAME # to be created in the SES console
password:
AWS_SES_SMTP_PASSWORD # to be created in the SES console
The port and encryption keys are not present in the Symfony Standard Edition configuration by
default, but you can simply add them as needed.
And that's it, you're ready to start sending emails through the cloud!
1. http://aws.amazon.com/ses
2. https://console.aws.amazon.com/ses

Chapter 56: How to Use the Cloud to Send Emails | 197
If you are using the Symfony Standard Edition, configure the parameters in parameters.yml
and use them in your configuration files. This allows for different Swift Mailer configurations for
each installation of your application. For instance, use Gmail during development and the cloud in
production.
Listing 56-2
1
2
3
4
5
6
7
8
9
parameters:
# ...
mailer_transport: smtp
mailer_host:
email-smtp.us-east-1.amazonaws.com
mailer_port:
587 # different ports are available, see SES console
mailer_encryption: tls # TLS encryption is required
mailer_user:
AWS_SES_SMTP_USERNAME # to be created in the SES console
mailer_password:
AWS_SES_SMTP_PASSWORD # to be created in the SES console
If you intend to use Amazon SES, please note the following:

You have to sign up to Amazon Web Services (AWS)3;
Every sender address used in the From or Return-Path (bounce address) header needs to be
confirmed by the owner. You can also confirm an entire domain;
Initially you are in a restricted sandbox mode. You need to request production access before
being allowed to send to arbitrary recipients;
SES may be subject to a charge.
3. http://aws.amazon.com

Chapter 56: How to Use the Cloud to Send Emails | 198
Chapter 57
How to Work with Emails during Development

When developing an application which sends email, you will often not want to actually send the email
to the specified recipient during development. If you are using the SwiftmailerBundle with Symfony,
you can easily achieve this through configuration settings without having to make any changes to
your application's code at all. There are two main choices when it comes to handling email during
development: (a) disabling the sending of email altogether or (b) sending all email to a specific address
(with optional exceptions).
Disabling Sending
You can disable sending email by setting the disable_delivery option to true. This is the default in
the test environment in the Standard distribution. If you do this in the test specific config then email
will not be sent when you run tests, but will continue to be sent in the prod and dev environments:
Listing 57-1
1
2
3
# app/config/config_test.yml
swiftmailer:
disable_delivery: true
If you'd also like to disable deliver in the dev environment, simply add this same configuration to the
config_dev.yml file.
Sending to a Specified Address

You can also choose to have all email sent to a specific address, instead of the address actually specified
when sending the message. This can be done via the delivery_address option:
Listing 57-2
1
2
3
swiftmailer:
delivery_address: '[email protected]'
Now, suppose you're sending an email to [email protected].

Listing 57-3

Chapter 57: How to Work with Emails during Development | 199
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17

{
->setBody(
$this->renderView(
'HelloBundle:Hello:email.txt.twig',
)
)
;
}
In the dev environment, the email will instead be sent to [email protected]. Swift Mailer will add
an extra header to the email, X-Swift-To, containing the replaced address, so you can still see who it
would have been sent to.
In addition to the to addresses, this will also stop the email being sent to any CC and BCC addresses
set for it. Swift Mailer will add additional headers to the email with the overridden addresses in them.
These are X-Swift-Cc and X-Swift-Bcc for the CC and BCC addresses respectively.
Sending to a Specified Address but with Exceptions

Suppose you want to have all email redirected to a specific address, (like in the above scenario to
[email protected]). But then you may want email sent to some specific email addresses to go through
after all, and not be redirected (even if it is in the dev environment). This can be done by adding the
delivery_whitelist option:
Listing 57-4
1
2
3
4
5
6
7
8
swiftmailer:
delivery_address: [email protected]
delivery_whitelist:
# all email addresses matching these regexes will be delivered
# like normal, as well as being sent to [email protected]
- '/@specialdomain\.com$/'
- '/âdmin@mydomain\.com$/'
In the above example all email messages will be redirected to [email protected] and messages sent
to the [email protected] address or to any email address belonging to the domain
specialdomain.com will also be delivered as normal.
Viewing from the Web Debug Toolbar

You can view any email sent during a single response when you are in the dev environment using the
web debug toolbar. The email icon in the toolbar will show how many emails were sent. If you click it, a
report will open showing the details of the sent emails.
If you're sending an email and then immediately redirecting to another page, the web debug toolbar will
not display an email icon or a report on the next page.
Instead, you can set the intercept_redirects option to true in the config_dev.yml file, which
will cause the redirect to stop and allow you to open the report with details of the sent emails.

Listing 57-5
1
2
3
web_profiler:
intercept_redirects: true
Alternatively, you can open the profiler after the redirect and search by the submit URL used on
the previous request (e.g. /contact/handle). The profiler's search feature allows you to load the
profiler information for any past requests.

Chapter 58
How to Spool Emails

When you are using the SwiftmailerBundle to send an email from a Symfony application, it will default
to sending the email immediately. You may, however, want to avoid the performance hit of the
communication between Swift Mailer and the email transport, which could cause the user to wait for
the next page to load while the email is sending. This can be avoided by choosing to "spool" the emails
instead of sending them directly. This means that Swift Mailer does not attempt to send the email but
instead saves the message to somewhere such as a file. Another process can then read from the spool and
take care of sending the emails in the spool. Currently only spooling to file or memory is supported by
Swift Mailer.
Spool Using Memory

When you use spooling to store the emails to memory, they will get sent right before the kernel
terminates. This means the email only gets sent if the whole request got executed without any unhandled
exception or any errors. To configure swiftmailer with the memory option, use the following
configuration:
Listing 58-1
1
2
3
4
swiftmailer:
# ...
spool: { type: memory }
Spool Using Files

When you use the filesystem for spooling, Symfony creates a folder in the given path for each mail service
(e.g. "default" for the default service). This folder will contain files for each email in the spool. So make
sure this directory is writable by Symfony (or your webserver/php)!
In order to use the spool with files, use the following configuration:
Listing 58-2
1
2
3
swiftmailer:
# ...

Chapter 58: How to Spool Emails | 202
4
5
6
spool:
type: file
path: /path/to/spooldir
If you want to store the spool somewhere with your project directory, remember that you can use
the %kernel.root_dir% parameter to reference the project's root:
Listing 58-3
path: '%kernel.root_dir%/spool'
Now, when your app sends an email, it will not actually be sent but instead added to the spool. Sending
the messages from the spool is done separately. There is a console command to send the messages in the
spool:
Listing 58-4
$ php app/console swiftmailer:spool:send --env=prod
It has an option to limit the number of messages to be sent:

Listing 58-5
$ php app/console swiftmailer:spool:send --message-limit=10 --env=prod
You can also set the time limit in seconds:

Listing 58-6
$ php app/console swiftmailer:spool:send --time-limit=10 --env=prod
Of course you will not want to run this manually in reality. Instead, the console command should be
triggered by a cron job or scheduled task and run at a regular interval.

Chapter 58: How to Spool Emails | 203
Chapter 59
How to Test that an Email is Sent in a

Functional Test
Sending emails with Symfony is pretty straightforward thanks to the SwiftmailerBundle, which leverages
the power of the Swift Mailer1 library.
To functionally test that an email was sent, and even assert the email subject, content or any other
headers, you can use the Symfony Profiler.
Start with an easy controller action that sends an email:
Listing 59-1
1
2
3
4
5
6
7
8
9
10
11
12
13
public function sendEmailAction($name)

{
->setBody('You should see me from the profiler!')
;
}
Don't forget to enable the profiler as explained in How to Use the Profiler in a Functional Test.
In your functional test, use the swiftmailer collector on the profiler to get information about the
messages sent on the previous request:
Listing 59-2
1
2
3
4
// src/AppBundle/Tests/Controller/MailControllerTest.php
use Symfony\Bundle\FrameworkBundle\Test\WebTestCase;
class MailControllerTest extends WebTestCase
1. http://swiftmailer.org/

Chapter 59: How to Test that an Email is Sent in a Functional Test | 204
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
{
public function testMailIsSentAndContentIsOk()
{
$client = static::createClient();
// Enable the profiler for the next request (it does nothing if the profiler is not available)
$client->enableProfiler();
$crawler = $client->request('POST', '/path/to/above/action');
$mailCollector = $client->getProfile()->getCollector('swiftmailer');
// Check that an email was sent

$this->assertEquals(1, $mailCollector->getMessageCount());
$collectedMessages = $mailCollector->getMessages();
$message = $collectedMessages[0];
// Asserting email data

$this->assertInstanceOf('Swift_Message', $message);
$this->assertEquals('Hello Email', $message->getSubject());
$this->assertEquals('[email protected]', key($message->getFrom()));
$this->assertEquals('[email protected]', key($message->getTo()));
$this->assertEquals(
'You should see me from the profiler!',
$message->getBody()
);
}
}

Chapter 59: How to Test that an Email is Sent in a Functional Test | 205
Chapter 60
How to Create Event Listeners and Subscribers

During the execution of a Symfony application, lots of event notifications are triggered. Your application
can listen to these notifications and respond to them by executing any piece of code.
Internal events provided by Symfony itself are defined in the KernelEvents1 class. Third-party bundles
and libraries also trigger lots of events and your own application can trigger custom events.
All the examples shown in this article use the same KernelEvents::EXCEPTION event for consistency
purposes. In your own application, you can use any event and even mix several of them in the same
subscriber.
Creating an Event Listener

The most common way to listen to an event is to register an event listener:
Listing 60-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
// src/AppBundle/EventListener/ExceptionListener.php
use Symfony\Component\HttpKernel\Event\GetResponseForExceptionEvent;
use Symfony\Component\HttpKernel\Exception\HttpExceptionInterface;
class ExceptionListener
{
public function onKernelException(GetResponseForExceptionEvent $event)
{
// You get the exception object from the received event
$exception = $event->getException();
$message = sprintf(
'My Error says: %s with code: %s',
$exception->getMessage(),
$exception->getCode()
);
// Customize your response object to display the exception details

$response = new Response();
$response->setContent($message);
1. http://api.symfony.com/2.7/Symfony/Component/HttpKernel/KernelEvents.html

Chapter 60: How to Create Event Listeners and Subscribers | 206
24
25
26
27
28
29
30
31
32
33
34
35
36
// HttpExceptionInterface is a special type of exception that

// holds status code and header details
if ($exception instanceof HttpExceptionInterface) {
$response->setStatusCode($exception->getStatusCode());
$response->headers->replace($exception->getHeaders());
} else {
$response->setStatusCode(Response::HTTP_INTERNAL_SERVER_ERROR);
}
// Send the modified response object to the event
$event->setResponse($response);
}
}
Each event receives a slightly different type of $event object. For the kernel.exception event,
it is GetResponseForExceptionEvent2. To see what type of object each event listener receives,
see KernelEvents3 or the documentation about the specific event you're listening to.
Now that the class is created, you just need to register it as a service and notify Symfony that it is a
"listener" on the kernel.exception event by using a special "tag":
Listing 60-2
services:
app.exception_listener:
class: AppBundle\EventListener\ExceptionListener
tags:
- { name: kernel.event_listener, event: kernel.exception }
1
2
3
4
5
6
There is an optional tag attribute called method which defines which method to execute when the
event is triggered. By default the name of the method is on + "camel-cased event name". If the event
is kernel.exception the method executed by default is onKernelException().
The other optional tag attribute is called priority, which defaults to 0 and it controls the order
in which listeners are executed (the highest the priority, the earlier a listener is executed). This is
useful when you need to guarantee that one listener is executed before another. The priorities of
the internal Symfony listeners usually range from -255 to 255 but your own listeners can use any
positive or negative integer.
Creating an Event Subscriber

Another way to listen to events is via an event subscriber, which is a class that defines one or more
methods that listen to one or various events. The main difference with the event listeners is that
subscribers always know which events they are listening to.
In a given subscriber, different methods can listen to the same event. The order in which methods are
executed is defined by the priority parameter of each method (the higher the priority the earlier the
method is called). To learn more about event subscribers, read The EventDispatcher Component.
The following example shows an event subscriber that defines several methods which listen to the same
kernel.exception event:
Listing 60-3
1
2
// src/AppBundle/EventSubscriber/ExceptionSubscriber.php
namespace AppBundle\EventSubscriber;
2. http://api.symfony.com/2.7/Symfony/Component/HttpKernel/Event/GetResponseForExceptionEvent.html
3. http://api.symfony.com/2.7/Symfony/Component/HttpKernel/KernelEvents.html

3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
use Symfony\Component\EventDispatcher\EventSubscriberInterface;
use Symfony\Component\HttpKernel\Event\GetResponseForExceptionEvent;
use Symfony\Component\HttpKernel\KernelEvents;
class ExceptionSubscriber implements EventSubscriberInterface
{
public static function getSubscribedEvents()
{
// return the subscribed events, their methods and priorities
return array(
KernelEvents::EXCEPTION => array(
array('processException', 10),
array('logException', 0),
array('notifyException', -10),
)
);
}
public function processException(GetResponseForExceptionEvent $event)
{
// ...
}
public function logException(GetResponseForExceptionEvent $event)
{
// ...
}
public function notifyException(GetResponseForExceptionEvent $event)
{
// ...
}
}
Now, you just need to register the class as a service and add the kernel.event_subscriber tag to
tell Symfony that this is an event subscriber:
Listing 60-4
1
2
3
4
5
6
services:
app.exception_subscriber:
class: AppBundle\EventSubscriber\ExceptionSubscriber
tags:
- { name: kernel.event_subscriber }
Request Events, Checking Types

A single page can make several requests (one master request, and then multiple sub-requests - typically
by Embedding Controllers). For the core Symfony events, you might need to check to see if the event is
for a "master" request or a "sub request":
Listing 60-5
1
2
3
4
5
6
7
8
9
10
11
12
13
14
// src/AppBundle/EventListener/RequestListener.php
use Symfony\Component\HttpKernel\Event\GetResponseEvent;
use Symfony\Component\HttpKernel\HttpKernel;
use Symfony\Component\HttpKernel\HttpKernelInterface;
class RequestListener
{
public function onKernelRequest(GetResponseEvent $event)
{
if (!$event->isMasterRequest()) {
// don't do anything if it's not the master request
return;

15
16
17
18
19
// ...
}
}
Certain things, like checking information on the real request, may not need to be done on the sub-request
listeners.
Listeners or Subscribers
Listeners and subscribers can be used in the same application indistinctly. The decision to use either of
them is usually a matter of personal taste. However, there are some minor advantages for each of them:
Subscribers are easier to reuse because the knowledge of the events is kept in the class rather than
in the service definition. This is the reason why Symfony uses subscribers internally;
Listeners are more flexible because bundles can enable or disable each of them conditionally
depending on some configuration value.
Debugging Event Listeners

New in version 2.6: The debug:event-dispatcher command was introduced in Symfony 2.6.
You can find out what listeners are registered in the event dispatcher using the console. To show all
events and their listeners, run:
Listing 60-6
$ php app/console debug:event-dispatcher
You can get registered listeners for a particular event by specifying its name:
Listing 60-7
$ php app/console debug:event-dispatcher kernel.exception

Chapter 61
How to Set Up Before and After Filters

It is quite common in web application development to need some logic to be executed just before or just
after your controller actions acting as filters or hooks.
In symfony1, this was achieved with the preExecute and postExecute methods. Most major frameworks
have similar methods but there is no such thing in Symfony. The good news is that there is a much better
way to interfere with the Request -> Response process using the EventDispatcher component.
Token Validation Example

Imagine that you need to develop an API where some controllers are public but some others are restricted
to one or some clients. For these private features, you might provide a token to your clients to identify
themselves.
So, before executing your controller action, you need to check if the action is restricted or not. If it is
restricted, you need to validate the provided token.
Please note that for simplicity in this recipe, tokens will be defined in config and neither database
setup nor authentication via the Security component will be used.
Before Filters with the kernel.controller Event

First, store some basic token configuration using config.yml and the parameters key:
Listing 61-1
1
2
3
4
5
parameters:
tokens:
client1: pass1
client2: pass2

Chapter 61: How to Set Up Before and After Filters | 210
Tag Controllers to Be Checked

A kernel.controller listener gets notified on every request, right before the controller is executed.
So, first, you need some way to identify if the controller that matches the request needs token validation.
A clean and easy way is to create an empty interface and make the controllers implement it:
Listing 61-2
1
2
3
4
5
6
interface TokenAuthenticatedController
{
// ...
}
A controller that implements this interface simply looks like this:

Listing 61-3
1
2
3
4
5
6
7
8
9
10
11
12
13
use AppBundle\Controller\TokenAuthenticatedController;
class FooController extends Controller implements TokenAuthenticatedController
{
// An action that needs authentication
public function barAction()
{
// ...
}
}
Creating an Event Listener

Next, you'll need to create an event listener, which will hold the logic that you want executed before your
controllers. If you're not familiar with event listeners, you can learn more about them at How to Create
Event Listeners and Subscribers:
Listing 61-4
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
// src/AppBundle/EventListener/TokenListener.php
use AppBundle\Controller\TokenAuthenticatedController;
use Symfony\Component\HttpKernel\Exception\AccessDeniedHttpException;
use Symfony\Component\HttpKernel\Event\FilterControllerEvent;
class TokenListener
{
private $tokens;
public function __construct($tokens)
{
$this->tokens = $tokens;
}
public function onKernelController(FilterControllerEvent $event)
{
$controller = $event->getController();
/*
* $controller passed can be either a class or a Closure.
* This is not usual in Symfony but it may happen.
* If it is a class, it comes in array format
*/
if (!is_array($controller)) {
return;
}
if ($controller[0] instanceof TokenAuthenticatedController) {
$token = $event->getRequest()->query->get('token');

32
33
34
35
36
37
if (!in_array($token, $this->tokens)) {
throw new AccessDeniedHttpException('This action needs a valid token!');
}
}
}
}
Registering the Listener

Finally, register your listener as a service and tag it as an event listener. By listening on
kernel.controller, you're telling Symfony that you want your listener to be called just before any
controller is executed.
Listing 61-5
1
2
3
4
5
6
7
services:
app.tokens.action_listener:
class: AppBundle\EventListener\TokenListener
arguments: ['%tokens%']
tags:
- { name: kernel.event_listener, event: kernel.controller, method: onKernelController }
With this configuration, your TokenListener onKernelController method will be executed on

each
request.
If
the
controller
that
is
about
to
be
executed
implements
TokenAuthenticatedController, token authentication is applied. This lets you have a "before"
filter on any controller that you want.
After Filters with the kernel.response Event

In addition to having a "hook" that's executed before your controller, you can also add a hook that's
executed after your controller. For this example, imagine that you want to add a sha1 hash (with a salt
using that token) to all responses that have passed this token authentication.
Another core Symfony event - called kernel.response - is notified on every request, but after the
controller returns a Response object. Creating an "after" listener is as easy as creating a listener class and
registering it as a service on this event.
For example, take the TokenListener from the previous example and first record the authentication
token inside the request attributes. This will serve as a basic flag that this request underwent token
authentication:
Listing 61-6
1
2
3
4
5
6
7
8
9
10
11
12
13
14
public function onKernelController(FilterControllerEvent $event)

{
// ...
if ($controller[0] instanceof TokenAuthenticatedController) {
$token = $event->getRequest()->query->get('token');
if (!in_array($token, $this->tokens)) {
throw new AccessDeniedHttpException('This action needs a valid token!');
}
// mark the request as having passed token authentication

$event->getRequest()->attributes->set('auth_token', $token);
}
}
Now, add another method to this class - onKernelResponse - that looks for this flag on the request
object and sets a custom header on the response if it's found:
Listing 61-7

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
// add the new use statement at the top of your file

use Symfony\Component\HttpKernel\Event\FilterResponseEvent;
public function onKernelResponse(FilterResponseEvent $event)
{
// check to see if onKernelController marked this as a token "auth'ed" request
if (!$token = $event->getRequest()->attributes->get('auth_token')) {
return;
}
$response = $event->getResponse();
// create a hash and set it as a response header

$hash = sha1($response->getContent().$token);
$response->headers->set('X-CONTENT-HASH', $hash);
}
Finally, a second "tag" is needed in the service definition to notify Symfony that the onKernelResponse
event should be notified for the kernel.response event:
Listing 61-8
1
2
3
4
5
6
7
8
services:
app.tokens.action_listener:
class: AppBundle\EventListener\TokenListener
arguments: ['%tokens%']
tags:
- { name: kernel.event_listener, event: kernel.controller, method: onKernelController }
- { name: kernel.event_listener, event: kernel.response, method: onKernelResponse }
That's it! The TokenListener is now notified before every controller is executed
(onKernelController) and after every controller returns a response (onKernelResponse). By
making specific controllers implement the TokenAuthenticatedController interface, your listener
knows which controllers it should take action on. And by storing a value in the request's "attributes" bag,
the onKernelResponse method knows to add the extra header. Have fun!

Chapter 62
How to Extend a Class without Using

Inheritance
To allow multiple classes to add methods to another one, you can define the magic __call() method
in the class you want to be extended like this:
Listing 62-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
class Foo
{
// ...
public function __call($method, $arguments)
{
// create an event named 'foo.method_is_not_found'
$event = new HandleUndefinedMethodEvent($this, $method, $arguments);
$this->dispatcher->dispatch('foo.method_is_not_found', $event);
// no listener was able to process the event? The method does not exist
if (!$event->isProcessed()) {
throw new \Exception(sprintf('Call to undefined method %s::%s.', get_class($this), $method));
}
// return the listener returned value
return $event->getReturnValue();
}
}
This uses a special HandleUndefinedMethodEvent that should also be created. This is a generic class
that could be reused each time you need to use this pattern of class extension:
Listing 62-2
1
2
3
4
5
6
7
8
9
10
11
use Symfony\Component\EventDispatcher\Event;
class HandleUndefinedMethodEvent extends Event
{
protected $subject;
protected $method;
protected $arguments;
protected $returnValue;
protected $isProcessed = false;
public function __construct($subject, $method, $arguments)

Chapter 62: How to Extend a Class without Using Inheritance | 214
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
{
$this->subject = $subject;
$this->method = $method;
$this->arguments = $arguments;
}
public function getSubject()
{
return $this->subject;
}
public function getMethod()
{
return $this->method;
}
public function getArguments()
{
return $this->arguments;
}
/**
* Sets the value to return and stops other listeners from being notified
*/
public function setReturnValue($val)
{
$this->returnValue = $val;
$this->isProcessed = true;
$this->stopPropagation();
}
public function getReturnValue()
{
return $this->returnValue;
}
public function isProcessed()
{
return $this->isProcessed;
}
}
Next, create a class that will listen to the foo.method_is_not_found event and add the method
bar():
Listing 62-3
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
class Bar
{
public function onFooMethodIsNotFound(HandleUndefinedMethodEvent $event)
{
// only respond to the calls to the 'bar' method
if ('bar' != $event->getMethod()) {
// allow another listener to take care of this unknown method
return;
}
// the subject object (the foo instance)

$foo = $event->getSubject();
// the bar method arguments
$arguments = $event->getArguments();
// ... do something
// set the return value
$event->setReturnValue($someValue);
}
}

Finally, add the new bar method to the Foo class by registering an instance of Bar with the
foo.method_is_not_found event:
Listing 62-4
1
2
$bar = new Bar();

$dispatcher->addListener('foo.method_is_not_found', array($bar, 'onFooMethodIsNotFound'));

Chapter 63
How to Customize a Method Behavior without

Using Inheritance
Doing something before or after a Method Call

If you want to do something just before, or just after a method is called, you can dispatch an event
respectively at the beginning or at the end of the method:
Listing 63-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
class Foo
{
// ...
public function send($foo, $bar)
{
// do something before the method
$event = new FilterBeforeSendEvent($foo, $bar);
$this->dispatcher->dispatch('foo.pre_send', $event);
// get $foo and $bar from the event, they may have been modified
$foo = $event->getFoo();
$bar = $event->getBar();
// the real method implementation is here
$ret = ...;
// do something after the method
$event = new FilterSendReturnValue($ret);
$this->dispatcher->dispatch('foo.post_send', $event);
return $event->getReturnValue();
}
}
In this example, two events are thrown: foo.pre_send, before the method is executed, and
foo.post_send after the method is executed. Each uses a custom Event class to communicate
information to the listeners of the two events. These event classes would need to be created by you and
should allow, in this example, the variables $foo, $bar and $ret to be retrieved and set by the listeners.
Chapter 63: How to Customize a Method Behavior without Using Inheritance | 217
For example, assuming the FilterSendReturnValue has a setReturnValue method, one listener
might look like this:
Listing 63-2
1
2
3
4
5
6
7
public function onFooPostSend(FilterSendReturnValue $event)

{
$ret = $event->getReturnValue();
// modify the original ``$ret`` value
$event->setReturnValue($ret);
}

Chapter 63: How to Customize a Method Behavior without Using Inheritance | 218
Chapter 64
How to use Expressions in Security, Routing,

Services, and Validation
In Symfony 2.4, a powerful ExpressionLanguage component was added to Symfony. This allows us to
add highly customized logic inside configuration.
The Symfony Framework leverages expressions out of the box in the following ways:
Configuring services;
Route matching conditions;
Checking security (explained below) and access controls with allow_if;
Validation.
For more information about how to create and work with expressions, see The Expression Syntax.
Security: Complex Access Controls with Expressions

In addition to a role like ROLE_ADMIN, the isGranted method also accepts an Expression1 object:
Listing 64-1
1
2
3
4
5
6
7
8
9
10
11
use Symfony\Component\ExpressionLanguage\Expression;
// ...
{
$this->denyAccessUnlessGranted(new Expression(
'"ROLE_ADMIN" in roles or (user and user.isSuperAdmin())'
));
// ...
}
In this example, if the current user has ROLE_ADMIN or if the current user object's isSuperAdmin()
method returns true, then access will be granted (note: your User object may not have an
isSuperAdmin method, that method is invented for this example).
1. http://api.symfony.com/2.7/Symfony/Component/ExpressionLanguage/Expression.html

Chapter 64: How to use Expressions in Security, Routing, Services, and Validation | 219
This uses an expression and you can learn more about the expression language syntax, see The Expression
Syntax.
Inside the expression, you have access to a number of variables:
user
The user object (or the string anon if you're not authenticated).
roles
The array of roles the user has, including from the role hierarchy but not including the
IS_AUTHENTICATED_* attributes (see the functions below).
object
The object (if any) that's passed as the second argument to isGranted.
token
The token object.

trust_resolver
The AuthenticationTrustResolverInterface2, object: you'll probably use the is_* functions below instead.
Additionally, you have access to a number of functions inside the expression:
is_authenticated
Returns true if the user is authenticated via "remember-me" or authenticated "fully" - i.e. returns true
if the user is "logged in".
is_anonymous
Equal to using IS_AUTHENTICATED_ANONYMOUSLY with the isGranted function.

is_remember_me
Similar, but not equal to IS_AUTHENTICATED_REMEMBERED, see below.

is_fully_authenticated
Similar, but not equal to IS_AUTHENTICATED_FULLY, see below.

has_role
Checks to see if the user has the given role - equivalent to an expression like 'ROLE_ADMIN'
in roles.
2. http://api.symfony.com/2.7/Symfony/Component/Security/Core/Authentication/AuthenticationTrustResolverInterface.html

is_remember_me is different than checking IS_AUTHENTICATED_REMEMBERED

The is_remember_me and is_authenticated_fully functions are similar to using
IS_AUTHENTICATED_REMEMBERED and IS_AUTHENTICATED_FULLY with the isGranted
function - but they are not the same. The following shows the difference:
Listing 64-2
1
2
3
4
5
6
7
8
9
use Symfony\Component\ExpressionLanguage\Expression;
// ...
$ac = $this->get('security.authorization_checker');
$access1 = $ac->isGranted('IS_AUTHENTICATED_REMEMBERED');
$access2 = $ac->isGranted(new Expression(
'is_remember_me() or is_fully_authenticated()'
));
Here, $access1 and $access2 will be the same value. Unlike the behavior of
IS_AUTHENTICATED_REMEMBERED and IS_AUTHENTICATED_FULLY, the is_remember_me
function only returns true if the user is authenticated via a remember-me cookie and
is_fully_authenticated only returns true if the user has actually logged in during this session
(i.e. is full-fledged).

Chapter 65
How to Customize Form Rendering

Symfony gives you a wide variety of ways to customize how a form is rendered. In this guide, you'll learn
how to customize every possible part of your form with as little effort as possible whether you use Twig
or PHP as your templating engine.
Form Rendering Basics

Recall that the label, error and HTML widget of a form field can easily be rendered by using the
form_row Twig function or the row PHP helper method:
Listing 65-1
{{ form_row(form.age) }}
You can also render each of the three parts of the field individually:
Listing 65-2
1
2
3
4
5
<div>
{{ form_label(form.age) }}
{{ form_errors(form.age) }}
{{ form_widget(form.age) }}
</div>
In both cases, the form label, errors and HTML widget are rendered by using a set of markup that ships
standard with Symfony. For example, both of the above templates would render:
Listing 65-3
1
2
3
4
5
6
7
<div>
<label for="form_age">Age</label>
<ul>
<li>This field is required</li>
</ul>
<input type="number" id="form_age" name="form[age]" />
</div>
To quickly prototype and test a form, you can render the entire form with just one line:
Listing 65-4
1
2
3
{# renders all fields #}

{{ form_widget(form) }}

Chapter 65: How to Customize Form Rendering | 222
4
5
{# renders all fields *and* the form start and end tags #}
{{ form(form) }}
The remainder of this recipe will explain how every part of the form's markup can be modified at
several different levels. For more information about form rendering in general, see Rendering a Form in a
Template.
What are Form Themes?

Symfony uses form fragments - a small piece of a template that renders just one part of a form - to render
each part of a form - field labels, errors, input text fields, select tags, etc.
The fragments are defined as blocks in Twig and as template files in PHP.
A theme is nothing more than a set of fragments that you want to use when rendering a form. In other
words, if you want to customize one portion of how a form is rendered, you'll import a theme which
contains a customization of the appropriate form fragments.
Symfony comes with four built-in form themes that define each and every fragment needed to render
every part of a form:
form_div_layout.html.twig1, wraps each form field inside a <div> element.
form_table_layout.html.twig2, wraps the entire form inside a <table> element and each form field
inside a <tr> element.
bootstrap_3_layout.html.twig3, wraps each form field inside a <div> element with the appropriate
CSS classes to apply the default Bootstrap 3 CSS framework4 styles.
bootstrap_3_horizontal_layout.html.twig5, it's similar to the previous theme, but the CSS classes
applied are the ones used to display the forms horizontally (i.e. the label and the widget in the same
row).
When you use the Bootstrap form themes and render the fields manually, calling form_label()
for a checkbox/radio field doesn't show anything. Due to Bootstrap internals, the label is already
shown by form_widget().
In the next section you will learn how to customize a theme by overriding some or all of its fragments.
For example, when the widget of an integer type field is rendered, an input number field is generated
Listing 65-5
renders:
Listing 65-6
<input type="number" id="form_age" name="form[age]" required="required" value="33" />
Internally, Symfony uses the integer_widget fragment to render the field. This is because the field
type is integer and you're rendering its widget (as opposed to its label or errors).
In Twig that would default to the block integer_widget from the form_div_layout.html.twig6
template.
1. https://github.com/symfony/symfony/blob/master/src/Symfony/Bridge/Twig/Resources/views/Form/form_div_layout.html.twig
2. https://github.com/symfony/symfony/blob/master/src/Symfony/Bridge/Twig/Resources/views/Form/form_table_layout.html.twig
3. https://github.com/symfony/symfony/blob/master/src/Symfony/Bridge/Twig/Resources/views/Form/bootstrap_3_layout.html.twig
4. http://getbootstrap.com/
5. https://github.com/symfony/symfony/blob/master/src/Symfony/Bridge/Twig/Resources/views/Form/bootstrap_3_horizontal_layout.html.twig

In PHP it would rather be the integer_widget.html.php file located in the FrameworkBundle/

Resources/views/Form folder.
The default implementation of the integer_widget fragment looks like this:
Listing 65-7
1
2
3
4
5
{# form_div_layout.html.twig #}
{% block integer_widget %}
{% set type = type|default('number') %}
{{ block('form_widget_simple') }}
{% endblock integer_widget %}
As you can see, this fragment itself renders another fragment - form_widget_simple:
Listing 65-8
1
2
3
4
5
{# form_div_layout.html.twig #}
{% block form_widget_simple %}
{% set type = type|default('text') %}
<input type="{{ type }}" {{ block('widget_attributes') }} {% if value is not empty %}value="{{ value }}"
{% endif %}/>
{% endblock form_widget_simple %}
The point is, the fragments dictate the HTML output of each part of a form. To customize the form
output, you just need to identify and override the correct fragment. A set of these form fragment
customizations is known as a form "theme". When rendering a form, you can choose which form
theme(s) you want to apply.
In Twig a theme is a single template file and the fragments are the blocks defined in this file.
In PHP a theme is a folder and the fragments are individual template files in this folder.
Knowing which Block to Customize

In this example, the customized fragment name is integer_widget because you want to override
the HTML widget for all integer field types. If you need to customize textarea fields, you
would customize textarea_widget.
As you can see, the fragment name is a combination of the field type and which part of the field is
being rendered (e.g. widget, label, errors, row). As such, to customize how errors are rendered
for just input text fields, you should customize the text_errors fragment.
More commonly, however, you'll want to customize how errors are displayed across all fields.
You can do this by customizing the form_errors fragment. This takes advantage of field type
inheritance. Specifically, since the text type extends from the form type, the Form component
will first look for the type-specific fragment (e.g. text_errors) before falling back to its parent
fragment name if it doesn't exist (e.g. form_errors).
For more information on this topic, see Form Fragment Naming.
Form Theming
To see the power of form theming, suppose you want to wrap every input number field with a div tag.
The key to doing this is to customize the integer_widget fragment.
Form Theming in Twig

When customizing the form field block in Twig, you have two options on where the customized form
block can live:

Method
Pros
Cons
Inside the same template as the

form
Quick and easy
Can't be reused in other templates
Inside a separate template
Can be reused by many

templates
Requires an extra template to be

created
Both methods have the same effect but are better in different situations.
Method 1: Inside the same Template as the Form

The easiest way to customize the integer_widget block is to customize it directly in the template
that's actually rendering the form.
Listing 65-9
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
{% extends 'base.html.twig' %}
{% form_theme form _self %}
<div class="integer_widget">
</div>
{% endblock %}
{% block content %}
{# ... render the form #}
{{ form_row(form.age) }}
{% endblock %}
By using the special {% form_theme form _self %} tag, Twig looks inside the same template for
any overridden form blocks. Assuming the form.age field is an integer type field, when its widget is
rendered, the customized integer_widget block will be used.
The disadvantage of this method is that the customized form block can't be reused when rendering other
forms in other templates. In other words, this method is most useful when making form customizations
that are specific to a single form in your application. If you want to reuse a form customization across
several (or all) forms in your application, read on to the next section.
Method 2: Inside a separate Template

You can also choose to put the customized integer_widget form block in a separate template entirely.
The code and end-result are the same, but you can now re-use the form customization across many
templates:
Listing 65-10
1
2
3
4
5
6
7
{# app/Resources/views/form/fields.html.twig #}
</div>
{% endblock %}
Now that you've created the customized form block, you need to tell Symfony to use it. Inside the
template where you're actually rendering your form, tell Symfony to use the template via the
form_theme tag:
Listing 65-11

1
2
3
{% form_theme form 'form/fields.html.twig' %}

When the form.age widget is rendered, Symfony will use the integer_widget block from the new
template and the input tag will be wrapped in the div element specified in the customized block.
Multiple Templates
A form can also be customized by applying several templates. To do this, pass the name of all the
templates as an array using the with keyword:
Listing 65-12
1
2
3
{% form_theme form with ['common.html.twig', 'form/fields.html.twig'] %}
{# ... #}
The templates can also be located in different bundles, use the functional name to reference these
templates, e.g. AcmeFormExtraBundle:form:fields.html.twig.
Child Forms
You can also apply a form theme to a specific child of your form:
Listing 65-13
{% form_theme form.child 'form/fields.html.twig' %}
This is useful when you want to have a custom theme for a nested form that's different than the one of
your main form. Just specify both your themes:
Listing 65-14
1
2
3
{% form_theme form 'form/fields.html.twig' %}

{% form_theme form.child 'form/fields_child.html.twig' %}
Form Theming in PHP

When using PHP as a templating engine, the only method to customize a fragment is to create a new
template file - this is similar to the second method used by Twig.
The template file must be named after the fragment. You must create a integer_widget.html.php
file in order to customize the integer_widget fragment.
Listing 65-15
1
2
3
4
5
6
7
8


<?php echo $view['form']->block(
$form,
'form_widget_simple',
array('type' => isset($type) ? $type : "number")
) ?>
</div>
Now that you've created the customized form template, you need to tell Symfony to use it. Inside the
template where you're actually rendering your form, tell Symfony to use the theme via the setTheme
helper method:
Listing 65-16
1
2
3
<?php $view['form']->setTheme($form, array(':form')); ?>

<?php $view['form']->widget($form['age']) ?>

form.age widget is rendered, Symfony will use the customized

integer_widget.html.php template and the input tag will be wrapped in the div element.
When
the
If you want to apply a theme to a specific child form, pass it to the setTheme method:
Listing 65-17
<?php $view['form']->setTheme($form['child'], ':form'); ?>
The :form syntax is based on the functional names for templates: Bundle:Directory. As the
form directory lives in the app/Resources/views directory, the Bundle part is empty, resulting
in :form.
Referencing base Form Blocks (Twig specific)

So far, to override a particular form block, the best method is to copy the default block from
form_div_layout.html.twig7, paste it into a different template, and then customize it. In many cases, you
can avoid doing this by referencing the base block when customizing it.
This is easy to do, but varies slightly depending on if your form block customizations are in the same
template as the form or a separate template.
Referencing Blocks from inside the same Template as the Form

Import the blocks by adding a use tag in the template where you're rendering the form:
Listing 65-18
{% use 'form_div_layout.html.twig' with integer_widget as base_integer_widget %}
Now, when the blocks from form_div_layout.html.twig8 are imported, the integer_widget block is
called base_integer_widget. This means that when you redefine the integer_widget block, you
can reference the default markup via base_integer_widget:
Listing 65-19
1
2
3
4
5
{{ block('base_integer_widget') }}
</div>
{% endblock %}
Referencing base Blocks from an external Template

If your form customizations live inside an external template, you can reference the base block by using
the parent() Twig function:
Listing 65-20
1
2
3
4
5
6
7
8
{# app/Resources/views/Form/fields.html.twig #}
{% extends 'form_div_layout.html.twig' %}
{{ parent() }}
</div>
{% endblock %}

It is not possible to reference the base block when using PHP as the templating engine. You have to
manually copy the content from the base block to your new template file.
Making Application-wide Customizations

If you'd like a certain form customization to be global to your application, you can accomplish this by
making the form customizations in an external template and then importing it inside your application
configuration.
Twig
By
following configuration, any customized form blocks

fields.html.twig template will be used globally when a form is rendered.
Listing 65-21
1
2
3
4
5
using
the
inside
the
form/
twig:
form_themes:
- 'form/fields.html.twig'
# ...
By default, Twig uses a div layout when rendering forms. Some people, however, may prefer to render
forms in a table layout. Use the form_table_layout.html.twig resource to use such a layout:
Listing 65-22
1
2
3
4
5
twig:
form_themes:
- 'form_table_layout.html.twig'
# ...
If you only want to make the change in one template, add the following line to your template file rather
than adding the template as a resource:
Listing 65-23
{% form_theme form 'form_table_layout.html.twig' %}
Note that the form variable in the above code is the form view variable that you passed to your template.
PHP
By using the following configuration, any customized form fragments inside the app/Resources/
views/Form folder will be used globally when a form is rendered.
Listing 65-24
1
2
3
4
5
6
7
framework:
templating:
form:
resources:
- 'AppBundle:Form'
# ...
By default, the PHP engine uses a div layout when rendering forms. Some people, however, may prefer to
render forms in a table layout. Use the FrameworkBundle:FormTable resource to use such a layout:
Listing 65-25
1
2
3
framework:
templating:

4
5
6
form:
resources:
- 'FrameworkBundle:FormTable'
If you only want to make the change in one template, add the following line to your template file rather
than adding the template as a resource:
Listing 65-26
<?php $view['form']->setTheme($form, array('FrameworkBundle:FormTable')); ?>
Note that the $form variable in the above code is the form view variable that you passed to your
template.
How to Customize an individual Field

So far, you've seen the different ways you can customize the widget output of all text field types. You can
also customize individual fields. For example, suppose you have two text fields in a product form name and description - but you only want to customize one of the fields. This can be accomplished
by customizing a fragment whose name is a combination of the field's id attribute and which part of the
field is being customized. For example, to customize the name field only:
Listing 65-27
1
2
3
4
5
6
7
8
9

{% block _product_name_widget %}
<div class="text_widget">
</div>
{% endblock %}
{{ form_widget(form.name) }}
Here, the _product_name_widget fragment defines the template to use for the field whose id is
product_name (and name is product[name]).
The product portion of the field is the form name, which may be set manually or generated
automatically based on your form type name (e.g. ProductType equates to product). If you're
not sure what your form name is, just view the source of your generated form.
If you want to change the product or name portion of the block name _product_name_widget
you can set the block_name option in your form type:
Listing 65-28
1
2
3
4
5
6
7
8
9
10
{
// ...
$builder->add('name', 'text', array(
'block_name' => 'custom_name',
));
}
Then the block name will be _product_custom_name_widget.
You can also override the markup for an entire field row using the same method:
Listing 65-29
1
2

3
4
5
6
7
8
9
10
11
{% block _product_name_row %}
<div class="name_row">
{{ form_label(form) }}
{{ form_errors(form) }}
</div>
{% endblock %}
{{ form_row(form.name) }}
How to Customize a Collection Prototype

When using a collection of forms, the prototype can be overridden with a completely custom prototype
by overriding a block. For example, if your form field is named tasks, you will be able to change the
widget for each task as follows:
Listing 65-30
1
2
3
4
5
6
7
8

{% block _tasks_entry_widget %}
<tr>
<td>{{ form_widget(form.task) }}</td>
<td>{{ form_widget(form.dueDate) }}</td>
</tr>
{% endblock %}
Not only can you override the rendered widget, but you can also change the complete form row or the
label as well. For the tasks field given above, the block names would be the following:
Part of the Form Block Name
label
_tasks_entry_label
widget
_tasks_entry_widget
row
_tasks_entry_row
Other common Customizations

So far, this recipe has shown you several different ways to customize a single piece of how a form is
rendered. The key is to customize a specific fragment that corresponds to the portion of the form you
want to control (see naming form blocks).
In the next sections, you'll see how you can make several common form customizations. To apply these
customizations, use one of the methods described in the Form Theming section.
Customizing Error Output

The Form component only handles how the validation errors are rendered, and not the actual
validation error messages. The error messages themselves are determined by the validation
constraints you apply to your objects. For more information, see the chapter on validation.
There are many different ways to customize how errors are rendered when a form is submitted with
errors. The error messages for a field are rendered when you use the form_errors helper:
Listing 65-31
{{ form_errors(form.age) }}

By default, the errors are rendered inside an unordered list:

Listing 65-32
1
2
3
<ul>
<li>This field is required</li>
</ul>
To override how errors are rendered for all fields, simply copy, paste and customize the form_errors
fragment.
Listing 65-33
1
2
3
4
5
6
7
8
9
10
11
12
{# form_errors.html.twig #}
{% block form_errors %}
{% spaceless %}
{% if errors|length > 0 %}
<ul>
{% for error in errors %}
<li>{{ error.message }}</li>
{% endfor %}
</ul>
{% endif %}
{% endspaceless %}
{% endblock form_errors %}
See Form Theming for how to apply this customization.
You can also customize the error output for just one specific field type. To customize only the markup
used for these errors, follow the same directions as above but put the contents in a relative _errors
block (or file in case of PHP templates). For example: text_errors (or text_errors.html.php).
See Form Fragment Naming to find out which specific block or file you have to customize.
Certain errors that are more global to your form (i.e. not specific to just one field) are rendered separately,
usually at the top of your form:
Listing 65-34
To customize only the markup used for these errors, follow the same directions as above, but now check
if the compound variable is set to true. If it is true, it means that what's being currently rendered is a
collection of fields (e.g. a whole form), and not just an individual field.
Listing 65-35
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
{# form_errors.html.twig #}
{% block form_errors %}
{% spaceless %}
{% if errors|length > 0 %}
{% if compound %}
<ul>
{% for error in errors %}
<li>{{ error.message }}</li>
{% endfor %}
</ul>
{% else %}
{# ... display the errors for a single field #}
{% endif %}
{% endif %}
{% endspaceless %}
{% endblock form_errors %}

Customizing the "Form Row"

When you can manage it, the easiest way to render a form field is via the form_row function, which
renders the label, errors and HTML widget of a field. To customize the markup used for rendering all
form field rows, override the form_row fragment. For example, suppose you want to add a class to the
div element around each row:
Listing 65-36
1
2
3
4
5
6
7
8
{# form_row.html.twig #}
{% block form_row %}
<div class="form_row">
{{ form_label(form) }}
</div>
{% endblock form_row %}
Adding a "Required" Asterisk to Field Labels

If you want to denote all of your required fields with a required asterisk (*), you can do this by
customizing the form_label fragment.
In Twig, if you're making the form customization inside the same template as your form, modify the use
tag and add the following:
Listing 65-37
1
2
3
4
5
6
7
8
9
{% use 'form_div_layout.html.twig' with form_label as base_form_label %}

{% block form_label %}
{{ block('base_form_label') }}
{% if required %}
<span class="required" title="This field is required">*</span>
{% endif %}
{% endblock %}
In Twig, if you're making the form customization inside a separate template, use the following:
Listing 65-38
1
2
3
4
5
6
7
8
9
{% block form_label %}
{{ parent() }}
{% if required %}
{% endif %}
{% endblock %}
When using PHP as a templating engine you have to copy the content from the original template:
Listing 65-39
1
2
3
4
5
6
7
8
9
10



<?php if ($required) { $label_attr['class'] = trim((isset($label_attr['class']) ? $label_attr['class'] :
'').' required'); } ?>
<?php if (!$compound) { $label_attr['for'] = $id; } ?>
<?php if (!$label) { $label = $view['form']->humanize($name); } ?>
<label <?php foreach ($label_attr as $k => $v) { printf('%s="%s" ', $view->escape($k), $view->escape($v)); }
?>><?php echo $view->escape($view['translator']->trans($label, array(), $translation_domain)) ?></label>

11
12


<?php if ($required) : ?>
<?php endif ?>
Using CSS only

By default, label tags of required fields are rendered with a required CSS class. Thus, you can
also add an asterisk using CSS only:
Listing 65-40
1
2
3
label.required:before {
content: "* ";
}
Adding "help" Messages

You can also customize your form widgets to have an optional "help" message.
In Twig, if you're making the form customization inside the same template as your form, modify the use
tag and add the following:
Listing 65-41
1
2
3
4
5
6
7
8
9
{% use 'form_div_layout.html.twig' with form_widget_simple as base_form_widget_simple %}

{{ block('base_form_widget_simple') }}
{% if help is defined %}
<span class="help-block">{{ help }}</span>
{% endif %}
{% endblock %}
In Twig, if you're making the form customization inside a separate template, use the following:
Listing 65-42
1
2
3
4
5
6
7
8
9
{{ parent() }}
{% if help is defined %}
<span class="help-block">{{ help }}</span>
{% endif %}
{% endblock %}
When using PHP as a templating engine you have to copy the content from the original template:
Listing 65-43
1
2
3
4
5
6
7
8
9
10
11



<input
type="<?php echo isset($type) ? $view->escape($type) : 'text' ?>"
<?php if (!empty($value)): ?>value="<?php echo $view->escape($value) ?>"<?php endif ?>
<?php echo $view['form']->block($form, 'widget_attributes') ?>
/>

<?php if (isset($help)) : ?>

12
13
<span class="help"><?php echo $view->escape($help) ?></span>

<?php endif ?>
To render a help message below a field, pass in a help variable:

Listing 65-44
{{ form_widget(form.title, {'help': 'foobar'}) }}
Using Form Variables

Most of the functions available for rendering different parts of a form (e.g. the form widget, form label,
form errors, etc.) also allow you to make certain customizations directly. Look at the following example:
Listing 65-45
1
2
{# render a widget, but add a "foo" class to it #}

{{ form_widget(form.name, { 'attr': {'class': 'foo'} }) }}
The array passed as the second argument contains form "variables". For more details about this concept
in Twig, see More about Form Variables.

Chapter 66
How to Use Data Transformers

Data transformers are used to translate the data for a field into a format that can be displayed in a form
(and back on submit). They're already used internally for many field types. For example, the date field
type can be rendered as a yyyy-MM-dd-formatted input textbox. Internally, a data transformer converts
the starting DateTime value of the field into the yyyy-MM-dd string to render the form, and then back
into a DateTime object on submit.
When a form field has the inherit_data option set, Data Transformers won't be applied to that
field.
Simple Example: Transforming String Tags from User Input to an Array

Suppose you have a Task form with a tags text type:
Listing 66-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
// src/AppBundle/Form/TaskType.php
namespace AppBundle\Form\Type;
// ...
class TaskType extends AbstractType
{
{
$builder->add('tags', 'text')
}
{
'data_class' => 'AppBundle\Entity\Task',
));
}

Chapter 66: How to Use Data Transformers | 235
22
23
// ...
}
Internally the tags are stored as an array, but displayed to the user as a simple comma seperated string
to make them easier to edit.
This is a perfect time to attach a custom data transformer to the tags field. The easiest way to do this is
with the CallbackTransformer1 class:
Listing 66-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
use Symfony\Component\Form\CallbackTransformer;
// ...
{
{
$builder->add('tags', 'text');
$builder->get('tags')
->addModelTransformer(new CallbackTransformer(
function ($tagsAsArray) {
// transform the array to a string
return implode(', ', $tagsAsArray);
},
function ($tagsAsString) {
// transform the string back to an array
return explode(', ', $tagsAsString);
}
))
;
}
// ...
}
The CallbackTransformer takes two callback functions as arguments. The first transforms the
original value into a format that'll be used to render the field. The second does the reverse: it transforms
the submitted value back into the format you'll use in your code.
addModelTransformer() method accepts any object that implements

DataTransformerInterface2 - so you can create your own classes, instead of putting all the
The
logic in the form (see the next section).
You can also add the transformer, right when adding the field by changing the format slightly:
Listing 66-3
1
2
3
4
5
$builder->add(
$builder
->create('tags', 'text')
->addModelTransformer(...)
);
1. http://api.symfony.com/2.7/Symfony/Component/Form/CallbackTransformer.html
2. http://api.symfony.com/2.7/Symfony/Component/Form/DataTransformerInterface.html

Harder Example: Transforming an Issue Number into an Issue Entity

Say you have a many-to-one relation from the Task entity to an Issue entity (i.e. each Task has an optional
foreign key to its related Issue). Adding a listbox with all possible issues could eventually get really long
and take a long time to load. Instead, you decide you want to add a textbox, where the user can simply
enter the issue number.
Start by setting up the text field like normal:
Listing 66-4
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
// ...
{
{
$builder
->add('description', 'textarea')
->add('issue', 'text')
;
}
{
'data_class' => 'AppBundle\Entity\Task'
));
}
// ...
}
Good start! But if you stopped here and submitted the form, the Task's issue property would be a string
(e.g. "55"). How can you transform this into an Issue entity on submit?
Creating the Transformer

You could use the CallbackTransformer like earlier. But since this is a bit more complex, creating a
new transformer class will keep the TaskType form class simpler.
Create an IssueToNumberTransformer class: it will be responsible for converting to and from the
issue number and the Issue object:
Listing 66-5
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
// src/AppBundle/Form/DataTransformer/IssueToNumberTransformer.php
namespace AppBundle\Form\DataTransformer;
use
use
use
use
AppBundle\Entity\Issue;
Doctrine\Common\Persistence\ObjectManager;
Symfony\Component\Form\DataTransformerInterface;
Symfony\Component\Form\Exception\TransformationFailedException;
class IssueToNumberTransformer implements DataTransformerInterface

{
private $manager;
public function __construct(ObjectManager $manager)
{
$this->manager = $manager;
}
/**
* Transforms an object (issue) to a string (number).
*
* @param Issue|null $issue

22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
* @return string
*/
public function transform($issue)
{
if (null === $issue) {
return '';
}
return $issue->getId();
}
/**
* Transforms a string (number) to an object (issue).
*
* @param string $issueNumber
* @return Issue|null
* @throws TransformationFailedException if object (issue) is not found.
*/
public function reverseTransform($issueNumber)
{
// no issue number? It's optional, so that's ok
if (!$issueNumber) {
return;
}
$issue = $this->manager
->getRepository('AppBundle:Issue')
// query for the issue with this id
->find($issueNumber)
;
if (null === $issue) {
// causes a validation error
// this message is not shown to the user
// see the invalid_message option
throw new TransformationFailedException(sprintf(
'An issue with number "%s" does not exist!',
$issueNumber
));
}
return $issue;
}
}
Just like in the first example, a transformer has two directions. The transform() method is responsible
for converting the data used in your code to a format that can be rendered in your form (e.g. an
Issue object to its id, a string). The reverseTransform() method does the reverse: it converts the
submitted value back into the format you want (e.g. convert the id back to the Issue object).
To cause a validation error, throw a TransformationFailedException3. But the message you pass
to this exception won't be shown to the user. You'll set that message with the invalid_message option
(see below).
When null is passed to the transform() method, your transformer should return an equivalent
value of the type it is transforming to (e.g. an empty string, 0 for integers or 0.0 for floats).
3. http://api.symfony.com/2.7/Symfony/Component/Form/Exception/TransformationFailedException.html

Using the Transformer

Next, you need to instantiate the IssueToNumberTransformer class from inside TaskType and
add it to the issue field. But to do that, you'll need an instance of the entity manager (because
IssueToNumberTransformer needs this).
No problem! Just add a __construct() function to TaskType and force this to be passed in. Then,
you can easily create and add the transformer:
Listing 66-6
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
use AppBundle\Form\DataTransformer\IssueToNumberTransformer;
use Doctrine\Common\Persistence\ObjectManager;
// ...
{
private $manager;
{
}
{
$builder
->add('issue', 'text', array(
// validation message if the data transformer fails
'invalid_message' => 'That is not a valid issue number',
));
// ...
$builder->get('issue')
->addModelTransformer(new IssueToNumberTransformer($this->manager));
}
// ...
}
Now, when you create your TaskType, you'll need to pass in the entity manager:
Listing 66-7
1
2
3
4
5
// e.g. in a controller somewhere

$manager = $this->getDoctrine()->getManager();
$form = $this->createForm(new TaskType($manager), $task);
// ...
To make this step easier (especially if TaskType is embedded into other form type classes), you
might choose to register your form type as a service.
Cool, you're done! Your user will be able to enter an issue number into the text field and it will
be transformed back into an Issue object. This means that, after a successful submission, the Form
component will pass a real Issue object to Task::setIssue() instead of the issue number.
If the issue isn't found, a form error will be created for that field and its error message can be controlled
with the invalid_message field option.

Be careful when adding your transformers. For example, the following is wrong, as the transformer
would be applied to the entire form, instead of just this field:
Listing 66-8
// THIS IS WRONG - TRANSFORMER WILL BE APPLIED TO THE ENTIRE FORM

// see above example for correct code
$builder->add('issue', 'text')
->addModelTransformer($transformer);
Creating a Reusable issue_selector Field

In the above example, you applied the transformer to a normal text field. But if you do this
transformation a lot, it might be better to create a custom field type. that does this automatically.
First, create the custom field type class:
Listing 66-9
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
// src/AppBundle/Form/IssueSelectorType.php
use
use
use
use
use
AppBundle\Form\DataTransformer\IssueToNumberTransformer;
Doctrine\Common\Persistence\ObjectManager;
Symfony\Component\Form\AbstractType;
Symfony\Component\Form\FormBuilderInterface;
Symfony\Component\OptionsResolver\OptionsResolver;
class IssueSelectorType extends AbstractType

{
private $manager;
{
}
{
$transformer = new IssueToNumberTransformer($this->manager);
$builder->addModelTransformer($transformer);
}
{
'invalid_message' => 'The selected issue does not exist',
));
}
{
return 'text';
}
{
return 'issue_selector';
}
}
Great! This will act and render like a text field (getParent()), but will automatically have the data
transformer and a nice default value for the invalid_message option.
Next, register your type as a service and tag it with form.type so that it's recognized as a custom field
type:
Listing 66-10

1
2
3
4
5
6
7
services:
app.type.issue_selector:
class: AppBundle\Form\IssueSelectorType
arguments: ['@doctrine.orm.entity_manager']
tags:
- { name: form.type, alias: issue_selector }
Now, whenever you need to use your special issue_selector field type, it's quite easy:
Listing 66-11
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
use AppBundle\Form\DataTransformer\IssueToNumberTransformer;
// ...
{
{
$builder
->add('issue', 'issue_selector')
;
}
// ...
}
About Model and View Transformers

In the above example, the transformer was used as a "model" transformer. In fact, there are two different
types of transformers and three different types of underlying data.
In any form, the three different types of data are:

1. Model data - This is the data in the format used in your application (e.g. an Issue object). If you
call Form::getData() or Form::setData(), you're dealing with the "model" data.
2. Norm Data - This is a normalized version of your data and is commonly the same as your
"model" data (though not in our example). It's not commonly used directly.

3. View Data - This is the format that's used to fill in the form fields themselves. It's also the
format in which the user will submit the data. When you call Form::submit($data), the $data is in
the "view" data format.
The two different types of transformers help convert to and from each of these types of data:
Model transformers:
transform:
"model data" => "norm data"

"norm data" => "model data"
reverseTransform:
View transformers:
transform:
"norm data" => "view data"

"view data" => "norm data"
reverseTransform:
Which transformer you need depends on your situation.

To use the view transformer, call addViewTransformer.
So why Use the Model Transformer?

In this example, the field is a text field, and a text field is always expected to be a simple, scalar format
in the "norm" and "view" formats. For this reason, the most appropriate transformer was the "model"
transformer (which converts to/from the norm format - string issue number - to the model format - Issue
object).
The difference between the transformers is subtle and you should always think about what the "norm"
data for a field should really be. For example, the "norm" data for a text field is a string, but is a
DateTime object for a date field.
As a general rule, the normalized data should contain as much information as possible.

Chapter 67
How to Dynamically Modify Forms Using Form

Events
Often times, a form can't be created statically. In this entry, you'll learn how to customize your form
based on three common use-cases:
1. Customizing your Form Based on the Underlying Data
Example: you have a "Product" form and need to modify/add/remove a field
based on the data on the underlying Product being edited.
2. How to dynamically Generate Forms Based on user Data
Example: you create a "Friend Message" form and need to build a drop-down that contains only
users that are friends with the current authenticated user.
3. Dynamic Generation for Submitted Forms
Example: on a registration form, you have a "country" field and a "state" field which should
populate dynamically based on the value in the "country" field.
If you wish to learn more about the basics behind form events, you can take a look at the Form Events
documentation.
Customizing your Form Based on the Underlying Data

Before jumping right into dynamic form generation, hold on and recall what a bare form class looks like:
Listing 67-1
1
2
3
4
5
6
7
8
9
10
11
// src/AppBundle/Form/Type/ProductType.php
{
{

Chapter 67: How to Dynamically Modify Forms Using Form Events | 243
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
$builder->add('name');
$builder->add('price');
}
{
'data_class' => 'AppBundle\Entity\Product'
));
}
{
return 'product';
}
}
If this particular section of code isn't already familiar to you, you probably need to take a step back
and first review the Forms chapter before proceeding.
Assume for a moment that this form utilizes an imaginary "Product" class that has only two properties
("name" and "price"). The form generated from this class will look the exact same regardless if a new
Product is being created or if an existing product is being edited (e.g. a product fetched from the
database).
Suppose now, that you don't want the user to be able to change the name value once the object has been
created. To do this, you can rely on Symfony's EventDispatcher component system to analyze the data on
the object and modify the form based on the Product object's data. In this entry, you'll learn how to add
this level of flexibility to your forms.
Adding an Event Listener to a Form Class

So, instead of directly adding that name widget, the responsibility of creating that particular field is
delegated to an event listener:
Listing 67-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
// ...
use Symfony\Component\Form\FormEvent;
use Symfony\Component\Form\FormEvents;
{
{
$builder->addEventListener(FormEvents::PRE_SET_DATA, function (FormEvent $event) {
// ... adding the name field if needed
});
}
// ...
}
The goal is to create a name field only if the underlying Product object is new (e.g. hasn't been persisted
to the database). Based on that, the event listener might look like the following:
Listing 67-3
1
2
// ...

3
4
5
6
7
8
9
10
11
12
13
14
15
16
// ...
$product = $event->getData();
$form = $event->getForm();
//
//
//
if
check if the Product object is "new"

If no data is passed to the form, the data is "null".
This should be considered a new "Product"
(!$product || null === $product->getId()) {
$form->add('name', 'text');
}
});
}
The FormEvents::PRE_SET_DATA line actually resolves to the string form.pre_set_data.

FormEvents1 serves an organizational purpose. It is a centralized location in which you can find all
of the various form events available. You can view the full list of form events via the FormEvents2
class.
Adding an Event Subscriber to a Form Class

For better reusability or if there is some heavy logic in your event listener, you can also move the logic for
creating the name field to an event subscriber:
Listing 67-4
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
// ...
use AppBundle\Form\EventListener\AddNameFieldSubscriber;
{
{
$builder->addEventSubscriber(new AddNameFieldSubscriber());
}
// ...
}
Now the logic for creating the name field resides in it own subscriber class:
Listing 67-5
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
// src/AppBundle/Form/EventListener/AddNameFieldSubscriber.php
namespace AppBundle\Form\EventListener;
class AddNameFieldSubscriber implements EventSubscriberInterface
{
{
// Tells the dispatcher that you want to listen on the form.pre_set_data
// event and that the preSetData method should be called.
return array(FormEvents::PRE_SET_DATA => 'preSetData');
}
1. http://api.symfony.com/2.7/Symfony/Component/Form/FormEvents.html
2. http://api.symfony.com/2.7/Symfony/Component/Form/FormEvents.html

16
17
18
19
20
21
22
23
24
25
26
public function preSetData(FormEvent $event)

{
$product = $event->getData();
if (!$product || null === $product->getId()) {
$form->add('name', 'text');
}
}
}
How to dynamically Generate Forms Based on user Data

Sometimes you want a form to be generated dynamically based not only on data from the form but also
on something else - like some data from the current user. Suppose you have a social website where a
user can only message people marked as friends on the website. In this case, a "choice list" of whom to
message should only contain users that are the current user's friends.
Creating the Form Type

Using an event listener, your form might look like this:
Listing 67-6
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
// src/AppBundle/Form/Type/FriendMessageFormType.php
use
use
use
use
use
Symfony\Component\Form\AbstractType;
Symfony\Component\Form\FormBuilderInterface;
Symfony\Component\Form\FormEvents;
Symfony\Component\Form\FormEvent;
Symfony\Component\Security\Core\Authentication\Token\Storage\TokenStorageInterface;
class FriendMessageFormType extends AbstractType

{
{
$builder
->add('subject', 'text')
->add('body', 'textarea')
;
// ... add a choice list of friends of the current application user
});
}
{
return 'app_friend_message';
}
}
The problem is now to get the current user and create a choice field that contains only this user's friends.
Luckily it is pretty easy to inject a service inside of the form. This can be done in the constructor:
Listing 67-7
1
2
3
4
5
6
private $tokenStorage;
public function __construct(TokenStorageInterface $tokenStorage)
{
$this->tokenStorage = $tokenStorage;
}

You might wonder, now that you have access to the User (through the token storage), why not just
use it directly in buildForm and omit the event listener? This is because doing so in the buildForm
method would result in the whole form type being modified and not just this one form instance. This
may not usually be a problem, but technically a single form type could be used on a single request to
create many forms or fields.
Customizing the Form Type

Now that you have all the basics in place you can take advantage of the TokenStorageInterface and
fill in the listener logic:
Listing 67-8
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
// src/AppBundle/FormType/FriendMessageFormType.php
use Symfony\Component\Security\Core\Authentication\Token\Storage\TokenStorageInterface;
use Doctrine\ORM\EntityRepository;
// ...
class FriendMessageFormType extends AbstractType
{
private $tokenStorage;
public function __construct(TokenStorageInterface $tokenStorage)
{
}
{
$builder
->add('subject', 'text')
->add('body', 'textarea')
;
// grab the user, do a quick sanity check that one exists

$user = $this->tokenStorage->getToken()->getUser();
if (!$user) {
throw new \LogicException(
'The FriendMessageFormType cannot be used without an authenticated user!'
);
}
$builder->addEventListener(
FormEvents::PRE_SET_DATA,
function (FormEvent $event) use ($user) {
$formOptions = array(
'class' => 'AppBundle\Entity\User',
'property' => 'fullName',
'query_builder' => function (EntityRepository $er) use ($user) {
// build a custom query
// return $er->createQueryBuilder('u')->addOrderBy('fullName', 'DESC');
// or call a method on your repository that returns the query builder

// the $er is an instance of your UserRepository
// return $er->createOrderByFullNameQueryBuilder();
},
);
// create the field, this is similar the $builder->add()

// field name, field type, data, options
$form->add('friend', 'entity', $formOptions);
}
);
}

56
57
// ...
}
New in version 2.6: The TokenStorageInterface3 was introduced in Symfony 2.6. Prior, you had to
use the getToken() method of SecurityContextInterface4.
The multiple and expanded form options will default to false because the type of the friend field
is entity.
Using the Form

Our form is now ready to use and there are two possible ways to use it inside of a controller:
1. create it manually and remember to pass the token storage to it;
or
2. define it as a service.
a) Creating the Form manually

This is very simple, and is probably the better approach unless you're using your new form type in many
places or embedding it into other forms:
Listing 67-9
1
2
3
4
5
6
7
8
9
10
11
12
class FriendMessageController extends Controller

{
{
$tokenStorage = $this->container->get('security.token_storage');
$form = $this->createForm(
new FriendMessageFormType($tokenStorage)
);
// ...
}
}
b) Defining the Form as a Service

To define your form as a service, just create a normal service and then tag it with form.type.
Listing 67-10
1
2
3
4
5
6
7
services:
app.form.friend_message:
class: AppBundle\Form\Type\FriendMessageFormType
arguments: ['@security.token_storage']
tags:
- { name: form.type, alias: app_friend_message }
If you wish to create it from within a service that has access to the form factory, you then use:
Listing 67-11
$form = $formFactory->create('friend_message');
In a controller that extends the Controller5 class, you can simply call:
Listing 67-12
3. http://api.symfony.com/2.7/Symfony/Component/Security/Core/Authentication/Token/Storage/TokenStorageInterface.html
4. http://api.symfony.com/2.7/Symfony/Component/Security/Core/SecurityContextInterface.html
5. http://api.symfony.com/2.7/Symfony/Bundle/FrameworkBundle/Controller/Controller.html

1
2
3
4
5
6
7
8
9
10
11
class FriendMessageController extends Controller
{
{
$form = $this->createForm('app_friend_message');
// ...
}
}
You can also easily embed the form type into another form:
Listing 67-13
1
2
3
4
5
// inside some other "form type" class

{
$builder->add('message', 'app_friend_message');
}
Dynamic Generation for Submitted Forms

Another case that can appear is that you want to customize the form specific to the data that was
submitted by the user. For example, imagine you have a registration form for sports gatherings. Some
events will allow you to specify your preferred position on the field. This would be a choice field for
example. However the possible choices will depend on each sport. Football will have attack, defense,
goalkeeper etc... Baseball will have a pitcher but will not have a goalkeeper. You will need the correct
options in order for validation to pass.
The meetup is passed as an entity field to the form. So we can access each sport like this:
Listing 67-14
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
// src/AppBundle/Form/Type/SportMeetupType.php
// ...
class SportMeetupType extends AbstractType
{
{
$builder
->add('sport', 'entity', array(
'class'
=> 'AppBundle:Sport',
'placeholder' => '',
))
;
function (FormEvent $event) {
// this would be your entity, i.e. SportMeetup

$data = $event->getData();
$sport = $data->getSport();
$positions = null === $sport ? array() : $sport->getAvailablePositions();
$form->add('position', 'entity', array(
'class'
=> 'AppBundle:Position',

34
35
36
37
38
39
40
41
42

'choices'
=> $positions,
));
}
);
}
// ...
}
New in version 2.6: The placeholder option was introduced in Symfony 2.6 and replaces
empty_value, which is available prior to 2.6.
When you're building this form to display to the user for the first time, then this example works perfectly.
However, things get more difficult when you handle the form submission. This is because the
PRE_SET_DATA event tells us the data that you're starting with (e.g. an empty SportMeetup object),
not the submitted data.
On a form, we can usually listen to the following events:
PRE_SET_DATA
POST_SET_DATA
PRE_SUBMIT
SUBMIT
POST_SUBMIT
New in version 2.3: The events PRE_SUBMIT, SUBMIT and POST_SUBMIT were introduced in Symfony
2.3. Before, they were named PRE_BIND, BIND and POST_BIND.
The key is to add a POST_SUBMIT listener to the field that your new field depends on. If you add a
POST_SUBMIT listener to a form child (e.g. sport), and add new children to the parent form, the Form
component will detect the new field automatically and map it to the submitted client data.
The type would now look like:
Listing 67-15
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
// src/AppBundle/Form/Type/SportMeetupType.php
// ...
use Symfony\Component\Form\FormInterface;
use AppBundle\Entity\Sport;
class SportMeetupType extends AbstractType
{
{
$builder
->add('sport', 'entity', array(
'class'
=> 'AppBundle:Sport',
));
;
$formModifier = function (FormInterface $form, Sport $sport = null) {
$positions = null === $sport ? array() : $sport->getAvailablePositions();
$form->add('position', 'entity', array(
'class'
=> 'AppBundle:Position',
'choices'
=> $positions,
));
};

31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
function (FormEvent $event) use ($formModifier) {

// this would be your entity, i.e. SportMeetup
$data = $event->getData();
$formModifier($event->getForm(), $data->getSport());
}
);
$builder->get('sport')->addEventListener(
FormEvents::POST_SUBMIT,
function (FormEvent $event) use ($formModifier) {
// It's important here to fetch $event->getForm()->getData(), as
// $event->getData() will get you the client data (that is, the ID)
$sport = $event->getForm()->getData();
// since we've added the listener to the child, we'll have to pass on
// the parent to the callback functions!
$formModifier($event->getForm()->getParent(), $sport);
}
);
}
// ...
}
You can see that you need to listen on these two events and have different callbacks only because in two
different scenarios, the data that you can use is available in different events. Other than that, the listeners
always perform exactly the same things on a given form.
One piece that is still missing is the client-side updating of your form after the sport is selected. This
should be handled by making an AJAX call back to your application. Assume that you have a sport
meetup creation controller:
Listing 67-16
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
// src/AppBundle/Controller/MeetupController.php
use Symfony\Component\HttpFoundation\Request;
use AppBundle\Entity\SportMeetup;
use AppBundle\Form\Type\SportMeetupType;
// ...
class MeetupController extends Controller
{
public function createAction(Request $request)
{
$meetup = new SportMeetup();
$form = $this->createForm(new SportMeetupType(), $meetup);
// ... save the meetup, redirect etc.
}
'AppBundle:Meetup:create.html.twig',
array('form' => $form->createView())
);
}
// ...
}
The associated template uses some JavaScript to update the position form field according to the
current selection in the sport field:
Listing 67-17
1
2
{# app/Resources/views/Meetup/create.html.twig #}

3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
{{ form_row(form.sport) }}
{# <select id="meetup_sport" ... #}
{{ form_row(form.position) }} {# <select id="meetup_position" ... #}
{# ... #}
<script>
var $sport = $('#meetup_sport');
// When sport gets selected ...
$sport.change(function() {
// ... retrieve the corresponding form.
var $form = $(this).closest('form');
// Simulate form data, but only include the selected sport value.
var data = {};
data[$sport.attr('name')] = $sport.val();
// Submit data via AJAX to the form's action path.
$.ajax({
url : $form.attr('action'),
type: $form.attr('method'),
data : data,
success: function(html) {
// Replace current position field ...
$('#meetup_position').replaceWith(
// ... with the returned one from the AJAX response.
$(html).find('#meetup_position')
);
// Position field now displays the appropriate positions.
}
});
});
</script>
The major benefit of submitting the whole form to just extract the updated position field is that no
additional server-side code is needed; all the code from above to generate the submitted form can be
reused.
Suppressing Form Validation

To
suppress
form
validation you can

ValidationListener6 from being called.
use
the
POST_SUBMIT event and prevent the
The reason for needing to do this is that even if you set validation_groups to false there are still
some integrity checks executed. For example an uploaded file will still be checked to see if it is too large
and the form will still check to see if non-existing fields were submitted. To disable all of this, use a
listener:
Listing 67-18
1
2
3
4
5
6
7
8
9
10
11
12
{
$builder->addEventListener(FormEvents::POST_SUBMIT, function (FormEvent $event) {
$event->stopPropagation();
}, 900); // Always set a higher priority than ValidationListener
// ...
}
6. http://api.symfony.com/2.7/Symfony/Component/Form/Extension/Validator/EventListener/ValidationListener.html

By doing this, you may accidentally disable something more than just form validation, since the
POST_SUBMIT event may have other listeners.

Chapter 68
How to Embed a Collection of Forms

In this entry, you'll learn how to create a form that embeds a collection of many other forms. This could
be useful, for example, if you had a Task class and you wanted to edit/create/remove many Tag objects
related to that Task, right inside the same form.
In this entry, it's loosely assumed that you're using Doctrine as your database store. But if you're not
using Doctrine (e.g. Propel or just a database connection), it's all very similar. There are only a few
parts of this tutorial that really care about "persistence".
If you are using Doctrine, you'll need to add the Doctrine metadata, including the ManyToMany
association mapping definition on the Task's tags property.
First, suppose that each Task belongs to multiple Tag objects. Start by creating a simple Task class:
Listing 68-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
// src/AppBundle/Entity/Task.php
use Doctrine\Common\Collections\ArrayCollection;
class Task
{
protected $description;
protected $tags;
public function __construct()
{
$this->tags = new ArrayCollection();
}
public function getDescription()
{
return $this->description;
}
public function setDescription($description)
{
$this->description = $description;
}
public function getTags()

Chapter 68: How to Embed a Collection of Forms | 254
28
29
30
31
{
return $this->tags;
}
}
The ArrayCollection is specific to Doctrine and is basically the same as using an array (but it
must be an ArrayCollection if you're using Doctrine).
Now, create a Tag class. As you saw above, a Task can have many Tag objects:
Listing 68-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
// src/AppBundle/Entity/Tag.php
class Tag
{
private $name;
{
return $this->name;
}
public function setName($name)
{
$this->name = $name;
}
}
Then, create a form class so that a Tag object can be modified by the user:
Listing 68-3
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
// src/AppBundle/Form/Type/TagType.php
class TagType extends AbstractType
{
{
$builder->add('name');
}
{
'data_class' => 'AppBundle\Entity\Tag',
));
}
{
return 'tag';
}
}
With this, you have enough to render a tag form by itself. But since the end goal is to allow the tags of a
Task to be modified right inside the task form itself, create a form for the Task class.
Notice that you embed a collection of TagType forms using the collection field type:
Listing 68-4
1
2
// src/AppBundle/Form/Type/TaskType.php

3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
{
{
$builder->add('description');
$builder->add('tags', 'collection', array('type' => new TagType()));
}
{
'data_class' => 'AppBundle\Entity\Task',
));
}
{
return 'task';
}
}
In your controller, you'll now initialize a new instance of TaskType:

Listing 68-5
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
// src/AppBundle/Controller/TaskController.php
use
use
use
use
use
AppBundle\Entity\Task;
AppBundle\Entity\Tag;
AppBundle\Form\Type\TaskType;
class TaskController extends Controller

{
{
$task = new Task();
// dummy code - this is here just so that the Task has some tags
// otherwise, this isn't an interesting example
$tag1 = new Tag();
$tag1->setName('tag1');
$task->getTags()->add($tag1);
$tag2 = new Tag();
$tag2->setName('tag2');
$task->getTags()->add($tag2);
// end dummy code
$form = $this->createForm(new TaskType(), $task);
// ... maybe do some form processing, like saving the Task and Tag objects
}
return $this->render('AppBundle:Task:new.html.twig', array(
));
}
}

The corresponding template is now able to render both the description field for the task form as
well as all the TagType forms for any tags that are already related to this Task. In the above controller,
I added some dummy code so that you can see this in action (since a Task has zero tags when first
created).
Listing 68-6
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
{# src/AppBundle/Resources/views/Task/new.html.twig #}
{# ... #}
{# render the task's only field: description #}
{{ form_row(form.description) }}
<h3>Tags</h3>
<ul class="tags">
{# iterate over each existing tag and render its only field: name #}
{% for tag in form.tags %}
<li>{{ form_row(tag.name) }}</li>
{% endfor %}
</ul>
{# ... #}
When the user submits the form, the submitted data for the tags field are used to construct an
ArrayCollection of Tag objects, which is then set on the tag field of the Task instance.
The tags collection is accessible naturally via $task->getTags() and can be persisted to the database
or used however you need.
So far, this works great, but this doesn't allow you to dynamically add new tags or delete existing tags.
So, while editing existing tags will work great, your user can't actually add any new tags yet.
In this entry, you embed only one collection, but you are not limited to this. You can also embed
nested collection as many levels down as you like. But if you use Xdebug in your development setup,
you may receive a Maximum function nesting level of '100' reached, aborting!
error. This is due to the xdebug.max_nesting_level PHP setting, which defaults to 100.
This directive limits recursion to 100 calls which may not be enough for rendering the form in the
template if you render the whole form at once (e.g form_widget(form)). To fix this you can set
this directive to a higher value (either via a php.ini file or via ini_set1, for example in app/
autoload.php) or render each form field by hand using form_row.
Allowing "new" Tags with the "Prototype"

Allowing the user to dynamically add new tags means that you'll need to use some JavaScript. Previously
you added two tags to your form in the controller. Now let the user add as many tag forms as they need
directly in the browser. This will be done through a bit of JavaScript.
The first thing you need to do is to let the form collection know that it will receive an unknown number
of tags. So far you've added two tags and the form type expects to receive exactly two, otherwise an error
will be thrown: This form should not contain extra fields. To make this flexible, add the
allow_add option to your collection field:
Listing 68-7
1
2
3
// ...
1. http://php.net/manual/en/function.ini-set.php

4
5
6
7
8
9
10
11
12
13
14
{
$builder->add('description');
$builder->add('tags', 'collection', array(
'type'
=> new TagType(),
'allow_add'
=> true,
));
}
In addition to telling the field to accept any number of submitted objects, the allow_add also makes a
"prototype" variable available to you. This "prototype" is a little "template" that contains all the HTML
to be able to render any new "tag" forms. To render it, make the following change to your template:
Listing 68-8
1
2
3
<ul class="tags" data-prototype="{{ form_widget(form.tags.vars.prototype)|e('html_attr') }}">

...
</ul>
If you render your whole "tags" sub-form at once (e.g. form_row(form.tags)), then the
prototype is automatically available on the outer div as the data-prototype attribute, similar to
what you see above.
The form.tags.vars.prototype is a form element that looks and feels just like the individual
form_widget(tag) elements inside your for loop. This means that you can call form_widget,
form_row or form_label on it. You could even choose to render only one of its fields (e.g. the
name field):
Listing 68-9
{{ form_widget(form.tags.vars.prototype.name)|e }}
On the rendered page, the result will look something like this:
Listing 68-10
<ul class="tags" data-prototype="<div><label class=" required">__name__</

label><div id="task_tags___name__"><div><label
for="task_tags___name___name" class=" required">Name</label><input
type="text" id="task_tags___name___name" name="task[tags][__name__][name]"
required="required" maxlength="255" /></div></div></div>">
The goal of this section will be to use JavaScript to read this attribute and dynamically add new tag forms
when the user clicks a "Add a tag" link. To make things simple, this example uses jQuery and assumes
you have it included somewhere on your page.
Add a script tag somewhere on your page so you can start writing some JavaScript.
First, add a link to the bottom of the "tags" list via JavaScript. Second, bind to the "click" event of that
link so you can add a new tag form (addTagForm will be show next):
Listing 68-11
1
2
3
4
5
6
7
8
9
10
11
var $collectionHolder;
// setup an "add a tag" link

var $addTagLink = $('<a href="#" class="add_tag_link">Add a tag</a>');
var $newLinkLi = $('<li></li>').append($addTagLink);
jQuery(document).ready(function() {
// Get the ul that holds the collection of tags
$collectionHolder = $('ul.tags');
// add the "add a tag" anchor and li to the tags ul

12
13
14
15
16
17
18
19
20
21
22
23
24
25
$collectionHolder.append($newLinkLi);
// count the current form inputs we have (e.g. 2), use that as the new
// index when inserting a new item (e.g. 2)
$collectionHolder.data('index', $collectionHolder.find(':input').length);
$addTagLink.on('click', function(e) {
// prevent the link from creating a "#" on the URL
e.preventDefault();
// add a new tag form (see next code block)

addTagForm($collectionHolder, $newLinkLi);
});
});
The addTagForm function's job will be to use the data-prototype attribute to dynamically add
a new form when this link is clicked. The data-prototype HTML contains the tag text input
element with a name of task[tags][__name__][name] and id of task_tags___name___name.
The __name__ is a little "placeholder", which you'll replace with a unique, incrementing number (e.g.
task[tags][3][name]).
The actual code needed to make this all work can vary quite a bit, but here's one example:
Listing 68-12
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
function addTagForm($collectionHolder, $newLinkLi) {

// Get the data-prototype explained earlier
var prototype = $collectionHolder.data('prototype');
// get the new index

var index = $collectionHolder.data('index');
// Replace '__name__' in the prototype's HTML to
// instead be a number based on how many items we have
var newForm = prototype.replace(/__name__/g, index);
// increase the index with one for the next item
$collectionHolder.data('index', index + 1);
// Display the form in the page in an li, before the "Add a tag" link li
var $newFormLi = $('<li></li>').append(newForm);
$newLinkLi.before($newFormLi);
}
It is better to separate your JavaScript in real JavaScript files than to write it inside the HTML as is
done here.
Now, each time a user clicks the Add a tag link, a new sub form will appear on the page. When the
form is submitted, any new tag forms will be converted into new Tag objects and added to the tags
property of the Task object.
You can find a working example in this JSFiddle2.
If you want to customize the HTML code in the prototype, read How to Customize a Collection Prototype.
To make handling these new tags easier, add an "adder" and a "remover" method for the tags in the Task
class:
Listing 68-13
1
2
2. http://jsfiddle.net/847Kf/4/

3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
// ...
class Task
{
// ...
public function addTag(Tag $tag)
{
$this->tags->add($tag);
}
public function removeTag(Tag $tag)
{
// ...
}
}
Next, add a by_reference option to the tags field and set it to false:
Listing 68-14
1
2
3
4
5
6
7
8
9
10
11
12
// ...
{
// ...
// ...
'by_reference' => false,
));
}
With these two changes, when the form is submitted, each new Tag object is added to the Task class
by calling the addTag method. Before this change, they were added internally by the form by calling
$task->getTags()->add($tag). That was just fine, but forcing the use of the "adder" method
makes handling these new Tag objects easier (especially if you're using Doctrine, which you will learn
about next!).
You have to create both addTag and removeTag methods, otherwise the form will still use
setTag even if by_reference is false. You'll learn more about the removeTag method later in
this article.

Doctrine: Cascading Relations and saving the "Inverse" side

To save the new tags with Doctrine, you need to consider a couple more things. First, unless you
iterate over all of the new Tag objects and call $em->persist($tag) on each, you'll receive an
error from Doctrine:
A new entity was found through the relationship
configured to cascade persist operations for entity...
AppBundle\Entity\Task#tags
that was not
To fix this, you may choose to "cascade" the persist operation automatically from the Task object to
any related tags. To do this, add the cascade option to your ManyToMany metadata:
Listing 68-15
1
2
3
4
5
6
7
8
// ...
/**
* @ORM\ManyToMany(targetEntity="Tag", cascade={"persist"})
*/
protected $tags;
A second potential issue deals with the Owning Side and Inverse Side3 of Doctrine relationships. In
this example, if the "owning" side of the relationship is "Task", then persistence will work fine as the
tags are properly added to the Task. However, if the owning side is on "Tag", then you'll need to do
a little bit more work to ensure that the correct side of the relationship is modified.
The trick is to make sure that the single "Task" is set on each "Tag". One easy way to do this is to
add some extra logic to addTag(), which is called by the form type since by_reference is set to
false:
Listing 68-16
1
2
3
4
5
6
7
8
9
// ...
public function addTag(Tag $tag)
{
$tag->addTask($this);
$this->tags->add($tag);
}
Inside Tag, just make sure you have an addTask method:

Listing 68-17
1
2
3
4
5
6
7
8
9
// src/AppBundle/Entity/Tag.php
// ...
public function addTask(Task $task)
{
if (!$this->tasks->contains($task)) {
$this->tasks->add($task);
}
}
If you have a one-to-many relationship, then the workaround is similar, except that you can simply
call setTask from inside addTag.
3. http://docs.doctrine-project.org/en/latest/reference/unitofwork-associations.html

Allowing Tags to be Removed

The next step is to allow the deletion of a particular item in the collection. The solution is similar to
allowing tags to be added.
Start by adding the allow_delete option in the form Type:
Listing 68-18
1
2
3
4
5
6
7
8
9
10
11
12
// ...
{
// ...
// ...
'allow_delete' => true,
));
}
Now, you need to put some code into the removeTag method of Task:
Listing 68-19
1
2
3
4
5
6
7
8
9
10
11
12
// ...
class Task
{
// ...
public function removeTag(Tag $tag)
{
$this->tags->removeElement($tag);
}
}
Template Modifications
The allow_delete option has one consequence: if an item of a collection isn't sent on submission, the
related data is removed from the collection on the server. The solution is thus to remove the form element
from the DOM.
First, add a "delete this tag" link to each tag form:
Listing 68-20
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
jQuery(document).ready(function() {
// Get the ul that holds the collection of tags
$collectionHolder = $('ul.tags');
// add a delete link to all of the existing tag form li elements

$collectionHolder.find('li').each(function() {
addTagFormDeleteLink($(this));
});
// ... the rest of the block from above
});
function addTagForm() {
// ...
// add a delete link to the new form

addTagFormDeleteLink($newFormLi);
}
The addTagFormDeleteLink function will look something like this:

Listing 68-21

1
2
3
4
5
6
7
8
9
10
11
12
function addTagFormDeleteLink($tagFormLi) {
var $removeFormA = $('<a href="#">delete this tag</a>');
$tagFormLi.append($removeFormA);
$removeFormA.on('click', function(e) {
// prevent the link from creating a "#" on the URL
e.preventDefault();
// remove the li for the tag form

$tagFormLi.remove();
});
}
When a tag form is removed from the DOM and submitted, the removed Tag object will not be included
in the collection passed to setTags. Depending on your persistence layer, this may or may not be
enough to actually remove the relationship between the removed Tag and Task object.

Doctrine: Ensuring the database persistence

When removing objects in this way, you may need to do a little bit more work to ensure that the
relationship between the Task and the removed Tag is properly removed.
In Doctrine, you have two sides of the relationship: the owning side and the inverse side. Normally
in this case you'll have a many-to-many relationship and the deleted tags will disappear and persist
correctly (adding new tags also works effortlessly).
But if you have a one-to-many relationship or a many-to-many relationship with a mappedBy on the
Task entity (meaning Task is the "inverse" side), you'll need to do more work for the removed tags
to persist correctly.
In this case, you can modify the controller to remove the relationship on the removed tag. This
assumes that you have some editAction which is handling the "update" of your Task:
Listing 68-22
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
// src/AppBundle/Controller/TaskController.php
use Doctrine\Common\Collections\ArrayCollection;
// ...
public function editAction($id, Request $request)
{
$em = $this->getDoctrine()->getManager();
$task = $em->getRepository('AppBundle:Task')->find($id);
if (!$task) {
throw $this->createNotFoundException('No task found for id '.$id);
}
$originalTags = new ArrayCollection();
// Create an ArrayCollection of the current Tag objects in the database

foreach ($task->getTags() as $tag) {
$originalTags->add($tag);
}
$editForm = $this->createForm(new TaskType(), $task);
$editForm->handleRequest($request);
if ($editForm->isValid()) {
// remove the relationship between the tag and the Task

foreach ($originalTags as $tag) {
if (false === $task->getTags()->contains($tag)) {
// remove the Task from the Tag
$tag->getTasks()->removeElement($task);
// if it was a many-to-one relationship, remove the relationship like this
// $tag->setTask(null);
$em->persist($tag);
// if you wanted to delete the Tag entirely, you can also do that
// $em->remove($tag);
}
}
$em->persist($task);
$em->flush();
// redirect back to some edit page

return $this->redirectToRoute('task_edit', array('id' => $id));
}
// render some form template

}
As you can see, adding and removing the elements correctly can be tricky. Unless you have a manyto-many relationship where Task is the "owning" side, you'll need to do extra work to make sure that

the relationship is properly updated (whether you're adding new tags or removing existing tags) on
each Tag object itself.

Chapter 69
How to Create a Custom Form Field Type

Symfony comes with a bunch of core field types available for building forms. However there are situations
where you may want to create a custom form field type for a specific purpose. This recipe assumes
you need a field definition that holds a person's gender, based on the existing choice field. This section
explains how the field is defined, how you can customize its layout and finally, how you can register it
for use in your application.
Defining the Field Type

In order to create the custom field type, first you have to create the class representing the field. In this
situation the class holding the field type will be called GenderType and the file will be stored in the
default location for form fields, which is <BundleName>\Form\Type. Make sure the field extends
AbstractType1:
Listing 69-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
// src/AppBundle/Form/Type/GenderType.php
class GenderType extends AbstractType
{
{
'choices' => array(
'm' => 'Male',
'f' => 'Female',
)
));
}
{
return 'choice';
}
1. http://api.symfony.com/2.7/Symfony/Component/Form/AbstractType.html

Chapter 69: How to Create a Custom Form Field Type | 266
23
24
25
26
27
28

{
return 'app_gender';
}
}
The location of this file is not important - the Form\Type directory is just a convention.
Here, the return value of the getParent function indicates that you're extending the choice field type.
This means that, by default, you inherit all of the logic and rendering of that field type. To see some of
the logic, check out the ChoiceType2 class. There are three methods that are particularly important:
buildForm()
Each field type has a buildForm method, which is where you configure and build any field(s). Notice
that this is the same method you use to setup your forms, and it works the same here.
buildView()
This method is used to set any extra variables you'll need when rendering your field in a template.
For example, in ChoiceType3, a multiple variable is set and used in the template to set (or not set) the
multiple attribute on the select field. See Creating a Template for the Field for more details.
New in version 2.7: The configureOptions() method was introduced in Symfony 2.7. Previously, the
method was called setDefaultOptions().
configureOptions()
This defines options for your form type that can be used in buildForm() and buildView(). There are a lot
of options common to all fields (see form Field Type), but you can create any others that you need
here.
If you're creating a field that consists of many fields, then be sure to set your "parent" type as form
or something that extends form. Also, if you need to modify the "view" of any of your child types
from your parent type, use the finishView() method.
The getName() method returns an identifier which should be unique in your application. This is used
in various places, such as when customizing how your form type will be rendered.
The goal of this field was to extend the choice type to enable selection of a gender. This is achieved by
fixing the choices to a list of possible genders.
Creating a Template for the Field

Each field type is rendered by a template fragment, which is determined in part by the value of your
getName() method. For more information, see What are Form Themes?.
In this case, since the parent field is choice, you don't need to do any work as the custom field type will
automatically be rendered like a choice type. But for the sake of this example, suppose that when your
field is "expanded" (i.e. radio buttons or checkboxes, instead of a select field), you want to always render
it in a ul element. In your form theme template (see above link for details), create a gender_widget
block to handle this:
2. https://github.com/symfony/symfony/blob/master/src/Symfony/Component/Form/Extension/Core/Type/ChoiceType.php
3. https://github.com/symfony/symfony/blob/master/src/Symfony/Component/Form/Extension/Core/Type/ChoiceType.php

Listing 69-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
{# app/Resources/views/form/fields.html.twig #}
{% block gender_widget %}
{% spaceless %}
{% if expanded %}
<ul {{ block('widget_container_attributes') }}>
{% for child in form %}
<li>
{{ form_widget(child) }}
{{ form_label(child) }}
</li>
{% endfor %}
</ul>
{% else %}
{# just let the choice widget render the select tag #}
{{ block('choice_widget') }}
{% endif %}
{% endspaceless %}
{% endblock %}
Make sure the correct widget prefix is used. In this example the name should be gender_widget,
according to the value returned by getName. Further, the main config file should point to the
custom form template so that it's used when rendering all forms.
When using Twig this is:
Listing 69-3
1
2
3
4
twig:
form_themes:
- 'form/fields.html.twig'
For the PHP templating engine, your configuration should look like this:
Listing 69-4
1
2
3
4
5
6
framework:
templating:
form:
resources:
- ':form:fields.html.php'
Using the Field Type

You can now use your custom field type immediately, simply by creating a new instance of the type in
one of your forms:
Listing 69-5
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
// src/AppBundle/Form/Type/AuthorType.php
class AuthorType extends AbstractType
{
{
$builder->add('gender_code', new GenderType(), array(
'placeholder' => 'Choose a gender',
));
}
}

But this only works because the GenderType() is very simple. What if the gender codes were stored
in configuration or in a database? The next section explains how more complex field types solve this
problem.
New in version 2.6: The placeholder option was introduced in Symfony 2.6 and replaces
empty_value, which is available prior to 2.6.
Creating your Field Type as a Service

So far, this entry has assumed that you have a very simple custom field type. But if you need access to
configuration, a database connection, or some other service, then you'll want to register your custom type
as a service. For example, suppose that you're storing the gender parameters in configuration:
Listing 69-6
1
2
3
4
5
parameters:
genders:
m: Male
f: Female
To use the parameter, define your custom field type as a service, injecting the genders parameter value
as the first argument to its to-be-created __construct function:
Listing 69-7
1
2
3
4
5
6
7
8
# src/AppBundle/Resources/config/services.yml
services:
app.form.type.gender:
class: AppBundle\Form\Type\GenderType
arguments:
- '%genders%'
tags:
- { name: form.type, alias: app_gender }
Make sure the services file is being imported. See Importing Configuration with imports for details.
Be sure that the alias attribute of the tag corresponds with the value returned by the getName method
defined earlier. You'll see the importance of this in a moment when you use the custom field type. But
first, add a __construct method to GenderType, which receives the gender configuration:
Listing 69-8
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
// src/AppBundle/Form/Type/GenderType.php
// ...
// ...
class GenderType extends AbstractType
{
private $genderChoices;
public function __construct(array $genderChoices)
{
$this->genderChoices = $genderChoices;
}
{
'choices' => $this->genderChoices,

22
23
24
25
26
));
}
// ...
}
Great! The GenderType is now fueled by the configuration parameters and registered as a service.
Additionally, because you used the form.type alias in its configuration, using the field is now much
easier:
Listing 69-9
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
// src/AppBundle/Form/Type/AuthorType.php
// ...
class AuthorType extends AbstractType
{
{
$builder->add('gender_code', 'gender', array(
'placeholder' => 'Choose a gender',
));
}
}
Notice that instead of instantiating a new instance, you can just refer to it by the alias used in your service
configuration, gender. Have fun!

Chapter 70
How to Create a Form Type Extension

Custom form field types are great when you need field types with a specific purpose, such as a gender
selector, or a VAT number input.
But sometimes, you don't really need to add new field types - you want to add features on top of existing
types. This is where form type extensions come in.
Form type extensions have 2 main use-cases:
1. You want to add a specific feature to a single type (such as adding a "download" feature to
the "file" field type);
2. You want to add a generic feature to several types (such as adding a "help" text to every
"input text"-like type).
It might be possible to achieve your goal with custom form rendering or custom form field types. But
using form type extensions can be cleaner (by limiting the amount of business logic in templates) and
more flexible (you can add several type extensions to a single form type).
Form type extensions can achieve most of what custom field types can do, but instead of being field types
of their own, they plug into existing types.
Imagine that you manage a Media entity, and that each media is associated to a file. Your Media form
uses a file type, but when editing the entity, you would like to see its image automatically rendered next
to the file input.
You could of course do this by customizing how this field is rendered in a template. But field type
extensions allow you to do this in a nice DRY fashion.
Defining the Form Type Extension

Your first task will be to create the form type extension class (called ImageTypeExtension in this
article). By standard, form extensions usually live in the Form\Extension directory of one of your
bundles.

Chapter 70: How to Create a Form Type Extension | 271
When creating a form type extension, you can either implement the FormTypeExtensionInterface1
interface or extend the AbstractTypeExtension2 class. In most cases, it's easier to extend the abstract
class:
Listing 70-1
// src/AppBundle/Form/Extension/ImageTypeExtension.php
namespace AppBundle\Form\Extension;
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
use Symfony\Component\Form\AbstractTypeExtension;
class ImageTypeExtension extends AbstractTypeExtension
{
/**
* Returns the name of the type being extended.
*
* @return string The name of the type being extended
*/
public function getExtendedType()
{
return 'file';
}
}
The only method you must implement is the getExtendedType function. It is used to indicate the
name of the form type that will be extended by your extension.
The value you return in the getExtendedType method corresponds to the value returned by the
getName method in the form type class you wish to extend.
In addition to the getExtendedType function, you will probably want to override one of the following
methods:
buildForm()
buildView()
configureOptions()
finishView()
For more information on what those methods do, you can refer to the Creating Custom Field Types
cookbook article.
Registering your Form Type Extension as a Service

The next step is to make Symfony aware of your extension. All you need to do is to declare it as a service
by using the form.type_extension tag:
Listing 70-2
1
2
3
4
5
services:
app.image_type_extension:
class: AppBundle\Form\Extension\ImageTypeExtension
tags:
- { name: form.type_extension, alias: file }
The alias key of the tag is the type of field that this extension should be applied to. In your case, as you
want to extend the file field type, you will use file as an alias.
1. http://api.symfony.com/2.7/Symfony/Component/Form/FormTypeExtensionInterface.html
2. http://api.symfony.com/2.7/Symfony/Component/Form/AbstractTypeExtension.html

Adding the extension Business Logic

The goal of your extension is to display nice images next to file inputs (when the underlying model
contains images). For that purpose, suppose that you use an approach similar to the one described in
How to handle File Uploads with Doctrine: you have a Media model with a path property, corresponding
to the image path in the database:
Listing 70-3
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
// src/AppBundle/Entity/Media.php
class Media
{
// ...
/**
* @var string The path - typically stored in the database
*/
private $path;
// ...
/**
* Get the image URL
*
* @return null|string
*/
public function getWebPath()
{
// ... $webPath being the full image URL, to be used in templates
return $webPath;
}
}
Your form type extension class will need to do two things in order to extend the file form type:
1. Override the configureOptions method in order to add an image_path option;
2. Override the buildView methods in order to pass the image URL to the view.
The logic is the following: when adding a form field of type file, you will be able to specify a new
option: image_path. This option will tell the file field how to get the actual image URL in order to
display it in the view:
Listing 70-4
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
// src/AppBundle/Form/Extension/ImageTypeExtension.php
namespace AppBundle\Form\Extension;
use
use
use
use
use
Symfony\Component\Form\AbstractTypeExtension;
Symfony\Component\Form\FormView;
Symfony\Component\Form\FormInterface;
Symfony\Component\PropertyAccess\PropertyAccess;
Symfony\Component\OptionsResolver\OptionsResolver;
class ImageTypeExtension extends AbstractTypeExtension

{
/**
* Returns the name of the type being extended.
*
* @return string The name of the type being extended
*/
public function getExtendedType()
{
return 'file';
}
/**
* Add the image_path option

24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
*
* @param OptionsResolver $resolver
*/
{
$resolver->setDefined(array('image_path'));
}
/**
* Pass the image URL to the view
*
* @param FormView $view
* @param FormInterface $form
* @param array $options
*/
public function buildView(FormView $view, FormInterface $form, array $options)
{
if (isset($options['image_path'])) {
$parentData = $form->getParent()->getData();
$imageUrl = null;
if (null !== $parentData) {
$accessor = PropertyAccess::createPropertyAccessor();
$imageUrl = $accessor->getValue($parentData, $options['image_path']);
}
// set an "image_url" variable that will be available when rendering this field
$view->vars['image_url'] = $imageUrl;
}
}
}
Override the File Widget Template Fragment

Each field type is rendered by a template fragment. Those template fragments can be overridden in order
to customize form rendering. For more information, you can refer to the What are Form Themes? article.
In your extension class, you have added a new variable (image_url), but you still need to take
advantage of this new variable in your templates. Specifically, you need to override the file_widget
block:
Listing 70-5
1
2
3
4
5
6
7
8
9
10
11
12
13
{# src/AppBundle/Resources/views/Form/fields.html.twig #}
{% block file_widget %}
{% spaceless %}
{{ block('form_widget') }}
{% if image_url is not null %}
<img src="{{ asset(image_url) }}"/>
{% endif %}
{% endspaceless %}
{% endblock %}
You will need to change your config file or explicitly specify how you want your form to be themed in
order for Symfony to use your overridden block. See What are Form Themes? for more information.

Using the Form Type Extension

From now on, when adding a field of type file in your form, you can specify an image_path option
that will be used to display an image next to the file field. For example:
Listing 70-6
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
// src/AppBundle/Form/Type/MediaType.php
class MediaType extends AbstractType
{
{
$builder
->add('name', 'text')
->add('file', 'file', array('image_path' => 'webPath'));
}
{
return 'app_media';
}
}
When displaying the form, if the underlying model has already been associated with an image, you will
see it displayed next to the file input.
Generic Form Type Extensions

You can modify several form types at once by specifying their common parent (Form Types Reference).
For example, several form types natively available in Symfony inherit from the text form type (such as
email, search, url, etc.). A form type extension applying to text (i.e. whose getExtendedType
method returns text) would apply to all of these form types.
In the same way, since most form types natively available in Symfony inherit from the form form type,
a form type extension applying to form would apply to all of these. A notable exception are the button
form types. Also keep in mind that a custom form type which extends neither the form nor the button
type could always be created.

Chapter 71
How to Reduce Code Duplication with

"inherit_data"
New in version 2.3: This inherit_data option was introduced in Symfony 2.3. Before, it was known
as virtual.
The inherit_data form field option can be very useful when you have some duplicated fields in
different entities. For example, imagine you have two entities, a Company and a Customer:
Listing 71-1
Listing 71-2
1
2
3
4
5
6
7
8
9
10
11
12
13
// src/AppBundle/Entity/Company.php
1
2
3
4
5
6
7
8
9
10
11
12
13
// src/AppBundle/Entity/Customer.php
class Company
{
private $name;
private $website;
private
private
private
private
$address;
$zipcode;
$city;
$country;
class Customer
{
private $firstName;
private $lastName;
private
private
private
private
$address;
$zipcode;
$city;
$country;
As you can see, each entity shares a few of the same fields: address, zipcode, city, country.
Start with building two forms for these entities, CompanyType and CustomerType:
Chapter 71: How to Reduce Code Duplication with "inherit_data" | 276
Listing 71-3
Listing 71-4
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
// src/AppBundle/Form/Type/CompanyType.php
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
// src/AppBundle/Form/Type/CustomerType.php
class CompanyType extends AbstractType
{
{
$builder
->add('name', 'text')
->add('website', 'text');
}
}
class CustomerType extends AbstractType
{
{
$builder
->add('firstName', 'text')
->add('lastName', 'text');
}
}
Instead of including the duplicated fields address, zipcode, city and country in both of these
forms, create a third form called LocationType for that:
Listing 71-5
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
// src/AppBundle/Form/Type/LocationType.php
class LocationType extends AbstractType
{
{
$builder
->add('address', 'textarea')
->add('zipcode', 'text')
->add('city', 'text')
->add('country', 'text');
}
{
'inherit_data' => true
));
}
{
return 'location';
}
}
The location form has an interesting option set, namely inherit_data. This option lets the form
inherit its data from its parent form. If embedded in the company form, the fields of the location form
will access the properties of the Company instance. If embedded in the customer form, the fields will
access the properties of the Customer instance instead. Easy, eh?
Instead of setting the inherit_data option inside LocationType, you can also (just like with
any option) pass it in the third argument of $builder->add().
Finally, make this work by adding the location form to your two original forms:
Listing 71-6
Listing 71-7
1
2
3
4
5
6
7
8
9
// src/AppBundle/Form/Type/CompanyType.php
{
// ...
1
2
3
4
5
6
7
8
9
// src/AppBundle/Form/Type/CustomerType.php
{
// ...
$builder->add('foo', new LocationType(), array(

'data_class' => 'AppBundle\Entity\Company'
));
}
$builder->add('bar', new LocationType(), array(

'data_class' => 'AppBundle\Entity\Customer'
));
}
That's it! You have extracted duplicated field definitions to a separate location form that you can reuse
wherever you need it.
Forms with the inherit_data option set cannot have *_SET_DATA event listeners.

Chapter 72
How to Unit Test your Forms

The Form component consists of 3 core objects: a form type (implementing
the Form2 and the FormView3.
FormTypeInterface1),
The only class that is usually manipulated by programmers is the form type class which serves as a form
blueprint. It is used to generate the Form and the FormView. You could test it directly by mocking its
interactions with the factory but it would be complex. It is better to pass it to FormFactory like it is done
in a real application. It is simple to bootstrap and you can trust the Symfony components enough to use
them as a testing base.
There is already a class that you can benefit from for simple FormTypes testing: TypeTestCase4. It is
used to test the core types and you can use it to test your types too.
New in version 2.3: The TypeTestCase has moved to the Symfony\Component\Form\Test
namespace
in
2.3.
Previously,
the
class
was
located
in
Symfony\Component\Form\Tests\Extension\Core\Type.
Depending on the way you installed your Symfony or Symfony Form component the tests may not
be downloaded. Use the --prefer-source option with Composer if this is the case.
The Basics
The simplest TypeTestCase implementation looks like the following:
Listing 72-1
1
2
3
4
5
// src/AppBundle/Tests/Form/Type/TestedTypeTest.php
namespace AppBundle\Tests\Form\Type;
use AppBundle\Form\Type\TestedType;
use AppBundle\Model\TestObject;
1. http://api.symfony.com/2.7/Symfony/Component/Form/FormTypeInterface.html
2. http://api.symfony.com/2.7/Symfony/Component/Form/Form.html
3. http://api.symfony.com/2.7/Symfony/Component/Form/FormView.html
4. http://api.symfony.com/2.7/Symfony/Component/Form/Test/TypeTestCase.html

Chapter 72: How to Unit Test your Forms | 279
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
use Symfony\Component\Form\Test\TypeTestCase;
class TestedTypeTest extends TypeTestCase
{
public function testSubmitValidData()
{
$formData = array(
'test' => 'test',
'test2' => 'test2',
);
$type = new TestedType();
$form = $this->factory->create($type);
$object = TestObject::fromArray($formData);
// submit the data to the form directly

$form->submit($formData);
$this->assertTrue($form->isSynchronized());
$this->assertEquals($object, $form->getData());
$view = $form->createView();
$children = $view->children;
foreach (array_keys($formData) as $key) {
$this->assertArrayHasKey($key, $children);
}
}
}
So, what does it test? Here comes a detailed explanation.

First you verify if the FormType compiles. This includes basic class inheritance, the buildForm
function and options resolution. This should be the first test you write:
Listing 72-2

This test checks that none of your data transformers used by the form failed. The isSynchronized()5
method is only set to false if a data transformer throws an exception:
Listing 72-3
$form->submit($formData);
$this->assertTrue($form->isSynchronized());
Don't test the validation: it is applied by a listener that is not active in the test case and it relies on
validation configuration. Instead, unit test your custom constraints directly.
Next, verify the submission and mapping of the form. The test below checks if all the fields are correctly
specified:
Listing 72-4
$this->assertEquals($object, $form->getData());
Finally, check the creation of the FormView. You should check if all widgets you want to display are
available in the children property:
Listing 72-5
1
2
3
4
5
6
$view = $form->createView();
$children = $view->children;
foreach (array_keys($formData) as $key) {
$this->assertArrayHasKey($key, $children);
}
5. http://api.symfony.com/2.7/Symfony/Component/Form/FormInterface.html#method_isSynchronized

Adding a Type your Form Depends on

Your form may depend on other types that are defined as services. It might look like this:
Listing 72-6
// src/AppBundle/Form/Type/TestedType.php
// ... the buildForm method
$builder->add('app_test_child_type');
To create your form correctly, you need to make the type available to the form factory in your test. The
easiest way is to register it manually before creating the parent form using the PreloadedExtension
class:
Listing 72-7
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
// src/AppBundle/Tests/Form/Type/TestedTypeTests.php
use
use
use
use
AppBundle\Form\Type\TestedType;
AppBundle\Model\TestObject;
Symfony\Component\Form\Test\TypeTestCase;
Symfony\Component\Form\PreloadedExtension;

{
protected function getExtensions()
{
$childType = new TestChildType();
return array(new PreloadedExtension(array(
$childType->getName() => $childType,
), array()));
}
public function testSubmitValidData()
{
// ... your test

}
}
Make sure the child type you add is well tested. Otherwise you may be getting errors that are not
related to the form you are currently testing but to its children.
Adding Custom Extensions

It often happens that you use some options that are added by form extensions. One of the cases may be
the ValidatorExtension with its invalid_message option. The TypeTestCase only loads the
core form extension, which means an InvalidOptionsException6 will be raised if you try to test a
class that depends on other extensions. The getExtensions()7 method allows you to return a list of
extensions to register:
Listing 72-8
1
2
3
4
6. http://api.symfony.com/2.7/Symfony/Component/OptionsResolver/Exception/InvalidOptionsException.html
7. http://api.symfony.com/2.7/Symfony/Component/Form/Test/TypeTestCase.html#method_getExtensions

5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
use
use
use
use
use
use
AppBundle\Model\TestObject;
Symfony\Component\Form\Extension\Validator\ValidatorExtension;
Symfony\Component\Form\Forms;
Symfony\Component\Form\FormBuilder;
Symfony\Component\Form\Test\TypeTestCase;
Symfony\Component\Validator\ConstraintViolationList;

{
protected function getExtensions()
{
$validator = $this->getMock('\Symfony\Component\Validator\Validator\ValidatorInterface');
$validator->method('validate')->will($this->returnValue(new ConstraintViolationList()));
return array(
new ValidatorExtension($validator),
);
}
// ... your tests

}
Testing against Different Sets of Data

If you are not familiar yet with PHPUnit's data providers8, this might be a good opportunity to use them:
Listing 72-9
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
use AppBundle\Model\TestObject;
use Symfony\Component\Form\Test\TypeTestCase;
{
/**
* @dataProvider getValidTestData
*/
public function testForm($data)
{
// ... your test
}
public function getValidTestData()
{
return array(
array(
'data' => array(
'test' => 'test',
'test2' => 'test2',
),
),
array(
'data' => array(),
),
array(
'data' => array(
'test' => null,
'test2' => null,
),
),
);
8. https://phpunit.de/manual/current/en/writing-tests-for-phpunit.html#writing-tests-for-phpunit.data-providers

38
39
}
}
The code above will run your test three times with 3 different sets of data. This allows for decoupling the
test fixtures from the tests and easily testing against multiple sets of data.
You can also pass another argument, such as a boolean if the form has to be synchronized with the given
set of data or not etc.

Chapter 73
How to Configure empty Data for a Form Class

The empty_data option allows you to specify an empty data set for your form class. This empty data set
would be used if you submit your form, but haven't called setData() on your form or passed in data
when you created your form. For example:
Listing 73-1
1
2
3
4
5
6
7
8
9
10
11
12

{
$blog = ...;
// $blog is passed in as the data, so the empty_data

// option is not needed
$form = $this->createForm(new BlogType(), $blog);
// no data is passed in, so empty_data is
// used to get the "starting data"
$form = $this->createForm(new BlogType());
}
By default, empty_data is set to null. Or, if you have specified a data_class option for your form
class, it will default to a new instance of that class. That instance will be created by calling the constructor
with no arguments.
If you want to override this default behavior, there are two ways to do this.
Option 1: Instantiate a new Class

One reason you might use this option is if you want to use a constructor that takes arguments.
Remember, the default data_class option calls that constructor with no arguments:
Listing 73-2
1
2
3
4
5
6
7
8
9
// src/AppBundle/Form/Type/BlogType.php
// ...
use AppBundle\Entity\Blog;
class BlogType extends AbstractType
{

Chapter 73: How to Configure empty Data for a Form Class | 284
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
private $someDependency;
public function __construct($someDependency)
{
$this->someDependency = $someDependency;
}
// ...
{
'empty_data' => new Blog($this->someDependency),
));
}
}
You can instantiate your class however you want. In this example, we pass some dependency into the
BlogType when we instantiate it, then use that to instantiate the Blog class. The point is, you can set
empty_data to the exact "new" object that you want to use.
Option 2: Provide a Closure

Using a closure is the preferred method, since it will only create the object if it is needed.
The closure must accept a FormInterface instance as the first argument:
Listing 73-3
1
2
3
4
5
6
7
8
9
10
11
12
// ...
{
'empty_data' => function (FormInterface $form) {
return new Blog($form->get('title')->getData());
},
));
}

Chapter 73: How to Configure empty Data for a Form Class | 285
Chapter 74
How to Use the submit() Function to Handle

Form Submissions
New in version 2.3: The handleRequest()1 method was introduced in Symfony 2.3.
With the handleRequest() method, it is really easy to handle form submissions:
Listing 74-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
// ...
{
$form = $this->createFormBuilder()
// ...
->getForm();
// perform some action...
return $this->redirectToRoute('task_success');
}
return $this->render('AppBundle:Default:new.html.twig', array(
));
}
To see more about this method, read Handling Form Submissions.
1. http://api.symfony.com/2.7/Symfony/Component/Form/FormInterface.html#method_handleRequest

Chapter 74: How to Use the submit() Function to Handle Form Submissions | 286
Calling Form::submit() manually

New in version 2.3: Before Symfony 2.3, the submit() method was known as bind().
In some cases, you want better control over when exactly your form is submitted and what data is passed
to it. Instead of using the handleRequest()2 method, pass the submitted data directly to submit()3:
Listing 74-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
// ...
{
// ...
->getForm();
if ($request->isMethod('POST')) {
$form->submit($request->request->get($form->getName()));
}
}
));
}
Forms consisting of nested fields expect an array in

fields by calling submit()5 directly on the field:
Listing 74-3
submit()4. You can also submit individual
$form->get('firstName')->submit('Fabien');
When submitting a form via a "PATCH" request, you may want to update only a few submitted
fields. To achieve this, you may pass an optional second boolean parameter to submit(). Passing
false will remove any missing fields within the form object. Otherwise, the mising fields will be set
to null.
Passing a Request to Form::submit() (Deprecated)

New in version 2.3: Before Symfony 2.3, the submit method was known as bind.
Before Symfony 2.3, the submit()6 method accepted a Request7 object as a convenient shortcut to the
previous example:
Listing 74-4
1
2
// ...
3. http://api.symfony.com/2.7/Symfony/Component/Form/FormInterface.html#method_submit
7. http://api.symfony.com/2.7/Symfony/Component/HttpFoundation/Request.html

3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23

{
// ...
->getForm();
if ($request->isMethod('POST')) {
$form->submit($request);
}
}
));
}
Passing the Request8 directly to submit()9 still works, but is deprecated and will be removed in
Symfony 3.0. You should use the method handleRequest()10 instead.

Chapter 75
How to Use the virtual Form Field Option

As of Symfony 2.3, the virtual option is renamed to inherit_data. You can read everything about
the new option in "How to Reduce Code Duplication with "inherit_data"".

Chapter 75: How to Use the virtual Form Field Option | 289
Chapter 76
Using Bower with Symfony

Symfony and all its packages are perfectly managed by Composer. Bower is a dependency management
tool for front-end dependencies, like Bootstrap or jQuery. As Symfony is purely a back-end framework,
it can't help you much with Bower. Fortunately, it is very easy to use!
Installing Bower
Bower1 is built on top of Node.js2. Make sure you have that installed and then run:
Listing 76-1
$ npm install -g bower
After this command has finished, run bower in your terminal to find out if it's installed correctly.
If you don't want to have NodeJS on your computer, you can also use BowerPHP3 (an unofficial PHP
port of Bower). Beware that this is currently in beta status. If you're using BowerPHP, use bowerphp
instead of bower in the examples.
Configuring Bower in your Project

Normally, Bower downloads everything into a bower_components/ directory. In Symfony, only files
in the web/ directory are publicly accessible, so you need to configure Bower to download things there
instead. To do that, just create a .bowerrc file with a new destination (like web/assets/vendor):
Listing 76-2
1
2
3
{
"directory": "web/assets/vendor/"
}
1. http://bower.io
2. https://nodejs.org
3. http://bowerphp.org/

Chapter 76: Using Bower with Symfony | 290
If you're using a front-end build system like Gulp4 or Grunt5, then you can set the directory to
whatever you want. Typically, you'll use these tools to ultimately move all assets into the web/
directory.
An Example: Installing Bootstrap

Believe it or not, but you're now ready to use Bower in your Symfony application. As an example, you'll
now install Bootstrap in your project and include it in your layout.
Installing the Dependency

To create a bower.json file, just run bower init. Now you're ready to start adding things to your
project. For example, to add Bootstrap6 to your bower.json and download it, just run:
Listing 76-3
$ bower install --save bootstrap
This will install Bootstrap and its dependencies in web/assets/vendor/ (or whatever directory you
configured in .bowerrc).
For more details on how to use Bower, check out Bower documentation7.
Including the Dependency in your Template

Now that the dependencies are installed, you can include bootstrap in your template like normal CSS/JS:
Listing 76-4
1
2
3
4
5
6
7
8
9
10
11
12
{# app/Resources/views/layout.html.twig #}
<!doctype html>
<html>
<head>
{# ... #}
<link rel="stylesheet"
href="{{ asset('assets/vendor/bootstrap/dist/css/bootstrap.min.css') }}">
</head>
{# ... #}
</html>
Great job! Your site is now using Bootstrap. You can now easily upgrade bootstrap to the latest version
and manage other front-end dependencies too.
Should I Git Ignore or Commit Bower Assets?

Currently, you should probably commit the assets downloaded by Bower instead of adding the directory
(e.g. web/assets/vendor) to your .gitignore file:
Listing 76-5
$ git add web/assets/vendor
4. http://gulpjs.com/
5. http://gruntjs.com/
6. http://getbootstrap.com/
7. http://bower.io/

Why? Unlike Composer, Bower currently does not have a "lock" feature, which means that there's no
guarantee that running bower install on a different server will give you the exact assets that you have
on other machines. For more details, read the article Checking in front-end dependencies8.
But, it's very possible that Bower will add a lock feature in the future (e.g. bower/bower#17489).
If you don't care too much about having exact the same versions, you can only commit the bower.json
file. Running bower install will give you the latest versions within the specified version range of each
package in bower.json. Using strict version constraints (e.g. 1.10.*) is often enough to ensure only
bringing in compatible versions.
8. http://addyosmani.com/blog/checking-in-front-end-dependencies/
9. https://github.com/bower/bower/pull/1748

Chapter 77
How to Install or Upgrade to the Latest,

Unreleased Symfony Version
In this article, you'll learn how to install and use new Symfony versions before they are released as stable
versions.
Creating a New Project Based on an Unstable Symfony Version

Suppose that Symfony 2.7 version hasn't been released yet and you want to create a new project to test
its features. First, install the Composer package manager. Then, open a command console, enter your
project's directory and execute the following command:
Listing 77-1
$ composer create-project symfony/framework-standard-edition my_project "2.7.*" --stability=dev
Once the command finishes its execution, you'll have a new Symfony project created in the
my_project/ directory and based on the most recent code found in the 2.7 branch.
If you want to test a beta version, use beta as the value of the stability option:
Listing 77-2
$ composer create-project symfony/framework-standard-edition my_project "2.7.*" --stability=beta
Upgrading your Project to an Unstable Symfony Version

Suppose again that Symfony 2.7 hasn't been released yet and you want to upgrade an existing application
to test that your project works with it.
First, open the composer.json file located in the root directory of your project. Then, edit the value of
the version defined for the symfony/symfony dependency as follows:
Listing 77-3
1
2
{
"require": {

Chapter 77: How to Install or Upgrade to the Latest, Unreleased Symfony Version | 293
3
4
5
"symfony/symfony" : "2.7.*@dev"
}
}
Finally, open a command console, enter your project directory and execute the following command to
update your project dependencies:
Listing 77-4
$ composer update symfony/symfony
If you prefer to test a Symfony beta version, replace the "2.7.*@dev" constraint by "2.7.0-beta1"
to install a specific beta number or 2.7.*@beta to get the most recent beta version.
After upgrading the Symfony version, read the Symfony Upgrading Guide to learn how you should
proceed to update your application's code in case the new Symfony version has deprecated some of its
features.
If you use Git to manage the project's code, it's a good practice to create a new branch to test the
new Symfony version. This solution avoids introducing any issue in your application and allows you
to test the new version with total confidence:
Listing 77-5
1
2
3
4
5
6
7
8
$
$
#
$
cd projects/my_project/
git checkout -b testing_new_symfony
... update composer.json configuration
composer update symfony/symfony
# ... after testing the new Symfony version

$ git checkout master
$ git branch -D testing_new_symfony

Chapter 77: How to Install or Upgrade to the Latest, Unreleased Symfony Version | 294
Chapter 78
How to Use Monolog to Write Logs

Monolog1 is a logging library for PHP used by Symfony. It is inspired by the Python LogBook library.
Usage
To log a message simply get the logger service from the container in your controller:
Listing 78-1
1
2
3
4
5
6
7
8

{
$logger = $this->get('logger');
$logger->info('I just got the logger');
$logger->error('An error occurred');
// ...
}
The logger service has different methods for different logging levels. See LoggerInterface2 for details on
which methods are available.
Handlers and Channels: Writing Logs to different Locations

In Monolog each logger defines a logging channel, which organizes your log messages into different
"categories". Then, each channel has a stack of handlers to write the logs (the handlers can be shared).
When injecting the logger in a service you can use a custom channel control which "channel" the
logger will log to.
The basic handler is the StreamHandler which writes logs in a stream (by default in the app/logs/
prod.log in the prod environment and app/logs/dev.log in the dev environment).
1. https://github.com/Seldaek/monolog
2. https://github.com/php-fig/log/blob/master/Psr/Log/LoggerInterface.php

Chapter 78: How to Use Monolog to Write Logs | 295
Monolog comes also with a powerful built-in handler for the logging in prod environment:
FingersCrossedHandler. It allows you to store the messages in a buffer and to log them only if a
message reaches the action level (error in the configuration provided in the Symfony Standard Edition)
by forwarding the messages to another handler.
Using several Handlers

The logger uses a stack of handlers which are called successively. This allows you to log the messages in
several ways easily.
Listing 78-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
monolog:
handlers:
applog:
type: stream
path: /var/log/symfony.log
level: error
main:
type: fingers_crossed
action_level: warning
handler: file
file:
type: stream
level: debug
syslog:
type: syslog
level: error
The above configuration defines a stack of handlers which will be called in the order they are defined.
The handler named "file" will not be included in the stack itself as it is used as a nested handler of
the fingers_crossed handler.
If you want to change the config of MonologBundle in another config file you need to redefine the
whole stack. It cannot be merged because the order matters and a merge does not allow to control
the order.
Changing the Formatter

The handler uses a Formatter to format the record before logging it. All Monolog handlers use an
instance of Monolog\Formatter\LineFormatter by default but you can replace it easily. Your
formatter must implement Monolog\Formatter\FormatterInterface.
Listing 78-3
1
2
3
4
5
6
7
8
9
10
services:
my_formatter:
class: Monolog\Formatter\JsonFormatter
monolog:
handlers:
file:
type: stream
level: debug
formatter: my_formatter

How to Rotate your Log Files

Over time, log files can grow to be huge, both while developing and on production. One best-practice
solution is to use a tool like the logrotate3 Linux command to rotate log files before they become too
large.
Another option is to have Monolog rotate the files for you by using the rotating_file handler. This
handler creates a new log file every day and can also remove old files automatically. To use it, just set the
type option of your handler to rotating_file:
Listing 78-4
1
2
3
4
5
6
7
8
9
10
monolog:
handlers:
main:
type: rotating_file
path: '%kernel.logs_dir%/%kernel.environment%.log'
level: debug
# max number of log files to keep
# defaults to zero, which means infinite files
max_files: 10
How to Disable Microseconds Precision

New in version 2.11: The use_microseconds option was introduced in MonologBundle 2.11.
Setting the parameter use_microseconds to false forces the logger to reduce the precision in
the datetime field of the log messages from microsecond to second, avoiding a call to the
microtime(true) function and the subsequent parsing. Disabling the use of microseconds can
provide a small performance gain speeding up the log generation. This is recommended for systems that
generate a large number of log events.
Listing 78-5
1
2
3
4
5
6
7
8
monolog:
use_microseconds: false
handlers:
applog:
type: stream
path: /var/log/symfony.log
level: error
Adding some extra Data in the Log Messages

Monolog allows you to process the record before logging it to add some extra data. A processor can be
applied for the whole handler stack or only for a specific handler.
A processor is simply a callable receiving the record as its first argument. Processors are configured using
the monolog.processor DIC tag. See the reference about it.
Adding a Session/Request Token

Sometimes it is hard to tell which entries in the log belong to which session and/or request. The following
example will add a unique token for each request using a processor.
Listing 78-6
3. https://fedorahosted.org/logrotate/

Listing 78-7
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
services:
monolog.formatter.session_request:
class: Monolog\Formatter\LineFormatter
arguments:
- "[%%datetime%%] [%%extra.token%%] %%channel%%.%%level_name%%: %%message%% %%context%%
%%extra%%\n"
use Symfony\Component\HttpFoundation\Session\Session;
class SessionRequestProcessor
{
private $session;
private $token;
public function __construct(Session $session)
{
$this->session = $session;
}
public function processRecord(array $record)
{
if (null === $this->token) {
try {
$this->token = substr($this->session->getId(), 0, 8);
} catch (\RuntimeException $e) {
$this->token = '????????';
}
$this->token .= '-' . substr(uniqid(), -8);
}
$record['extra']['token'] = $this->token;
return $record;
}
}
monolog.processor.session_request:
class: AppBundle\SessionRequestProcessor
arguments: ['@session']
tags:
- { name: monolog.processor, method: processRecord }
monolog:
handlers:
main:
type: stream
level: debug
formatter: monolog.formatter.session_request
If you use several handlers, you can also register a processor at the handler level or at the channel
level instead of registering it globally (see the following sections).
Registering Processors per Handler

You can register a processor per handler using the handler option of the monolog.processor tag:
Listing 78-8
1
2
3
services:

4
5
6
7
tags:
- { name: monolog.processor, method: processRecord, handler: main }
Registering Processors per Channel

You can register a processor per channel using the channel option of the monolog.processor tag:
Listing 78-9
1
2
3
4
5
6
7
services:
tags:
- { name: monolog.processor, method: processRecord, channel: main }

Chapter 79
How to Configure Monolog to Email Errors

Monolog1 can be configured to send an email when an error occurs with an application. The configuration
for this requires a few nested handlers in order to avoid receiving too many emails. This configuration
looks complicated at first but each handler is fairly straightforward when it is broken down.
Listing 79-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
monolog:
handlers:
mail:
type:
fingers_crossed
# 500 errors are logged at the critical level
action_level: critical
# to also log 400 level errors (but not 404's):
# action_level: error
# excluded_404s:
#
- ^/
handler:
deduplicated
deduplicated:
type:
deduplication
handler: swift
swift:
type:
swift_mailer
from_email: '[email protected]'
to_email:
'[email protected]'
# or list of recipients
# to_email: ['[email protected]', '[email protected]', ...]
subject:
'An Error Occurred! %%message%%'
level:
debug
formatter: monolog.formatter.html
content_type: text/html
The mail handler is a fingers_crossed handler which means that it is only triggered when the action
level, in this case critical is reached. The critical level is only triggered for 5xx HTTP code errors.
If this level is reached once, the fingers_crossed handler will log all messages regardless of their level.
The handler setting means that the output is then passed onto the deduplicated handler.
1. https://github.com/Seldaek/monolog

Chapter 79: How to Configure Monolog to Email Errors | 300
If you want both 400 level and 500 level errors to trigger an email, set the action_level to error
instead of critical. See the code above for an example.
The deduplicated handler simply keeps all the messages for a request and then passes them onto the
nested handler in one go, but only if the records are unique over a given period of time (60 seconds by
default). If the records are duplicates they are simply discarded. Adding this handler reduces the amount
of notifications to a manageable level, specially in critical failure scenarios.
The messages are then passed to the swift handler. This is the handler that actually deals with emailing
you the error. The settings for this are straightforward, the to and from addresses, the formatter, the
content type and the subject.
You can combine these handlers with other handlers so that the errors still get logged on the server as
well as the emails being sent:
Listing 79-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
monolog:
handlers:
main:
type:
fingers_crossed
action_level: critical
handler:
grouped
grouped:
type:
group
members: [streamed, deduplicated]
streamed:
type: stream
level: debug
deduplicated:
type:
deduplication
handler: swift
swift:
type:
swift_mailer
from_email: '[email protected]'
to_email:
'[email protected]'
subject:
'An Error Occurred! %%message%%'
level:
debug
formatter: monolog.formatter.html
content_type: text/html
This uses the group handler to send the messages to the two group members, the deduplicated and
the stream handlers. The messages will now be both written to the log file and emailed.

Chapter 79: How to Configure Monolog to Email Errors | 301
Chapter 80
How to Configure Monolog to Display Console

Messages
It is possible to use the console to print messages for certain verbosity levels using the
OutputInterface1 instance that is passed when a command gets executed.
Alternatively, you can use the standalone PSR-3 logger provided with the console component.
When a lot of logging has to happen, it's cumbersome to print information depending on the verbosity
settings (-v, -vv, -vvv) because the calls need to be wrapped in conditions. The code quickly gets
verbose or dirty. For example:
Listing 80-1
1
2
3
4
5
6
7
8
9
10
11
12
13
{
if ($output->getVerbosity() >= OutputInterface::VERBOSITY_DEBUG) {
$output->writeln('Some info');
}
if ($output->getVerbosity() >= OutputInterface::VERBOSITY_VERBOSE) {
$output->writeln('Some more info');
}
}
Instead of using these semantic methods to test for each of the verbosity levels, the MonologBridge2
provides a ConsoleHandler3 that listens to console events and writes log messages to the console output
depending on the current log level and the console verbosity.
The example above could then be rewritten as:
Listing 80-2
1
2
1. http://api.symfony.com/2.7/Symfony/Component/Console/Output/OutputInterface.html
2. https://github.com/symfony/MonologBridge
3. https://github.com/symfony/MonologBridge/blob/master/Handler/ConsoleHandler.php

Chapter 80: How to Configure Monolog to Display Console Messages | 302
3
4
5
6
7
8
9
10
11

{
// assuming the Command extends ContainerAwareCommand...
$logger->debug('Some info');
$logger->notice('Some more info');
}
Depending on the verbosity level that the command is run in and the user's configuration (see below),
these messages may or may not be displayed to the console. If they are displayed, they are timestamped
and colored appropriately. Additionally, error logs are written to the error output (php://stderr). There
is no need to conditionally handle the verbosity settings anymore.
The Monolog console handler is enabled in the Monolog configuration. This is the default in Symfony
Standard Edition 2.4 too.
Listing 80-3
1
2
3
4
5
monolog:
handlers:
console:
type: console
With the verbosity_levels option you can adapt the mapping between verbosity and log level.
In the given example it will also show notices in normal verbosity mode (instead of warnings only).
Additionally, it will only use messages logged with the custom my_channel channel and it changes the
display style via a custom formatter (see the MonologBundle reference for more information):
Listing 80-4
1
2
3
4
5
6
7
8
9
monolog:
handlers:
console:
type:
console
verbosity_levels:
VERBOSITY_NORMAL: NOTICE
channels: my_channel
formatter: my_formatter
Listing 80-5
1
2
3
4
5
6
services:
my_formatter:
class: Symfony\Bridge\Monolog\Formatter\ConsoleFormatter
arguments:
- "[%%datetime%%] %%start_tag%%%%message%%%%end_tag%% (%%level_name%%) %%context%% %%extra%%\n"

Chapter 80: How to Configure Monolog to Display Console Messages | 303
Chapter 81
How to Configure Monolog to Exclude 404

Errors from the Log
Sometimes your logs become flooded with unwanted 404 HTTP errors, for example, when an attacker
scans your app for some well-known application paths (e.g. /phpmyadmin). When using a
fingers_crossed handler, you can exclude logging these 404 errors based on a regular expression in
the MonologBundle configuration:
Listing 81-1
1
2
3
4
5
6
7
8
9
monolog:
handlers:
main:
# ...
type: fingers_crossed
handler: ...
excluded_404s:
- ^/phpmyadmin

Chapter 81: How to Configure Monolog to Exclude 404 Errors from the Log | 304
Chapter 82
How to Log Messages to different Files

The Symfony Framework organizes log messages into channels. By default, there are several channels,
including doctrine, event, security, request and more. The channel is printed in the log message
and can also be used to direct different channels to different places/files.
By default, Symfony logs every message into a single file (regardless of the channel).
Each channel corresponds to a logger service (monolog.logger.XXX) in the container (use the
debug:container command to see a full list) and those are injected into different services.
Switching a Channel to a different Handler

Now, suppose you want to log the security channel to a different file. To do this, just create a new
handler and configure it to log only messages from the security channel. You might add this in
config.yml to log in all environments, or just config_prod.yml to happen only in prod:
Listing 82-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
monolog:
handlers:
security:
# log all messages (since debug is the lowest level)
level:
debug
type:
stream
path:
'%kernel.logs_dir%/security.log'
channels: [security]
# an example of *not* logging security channel messages for this handler
main:
# ...
# channels: ['!security']
The channels configuration only works for top level handlers. Handlers that are nested inside a
group, buffer, filter, fingers crossed or other such handler will ignore this configuration and will
process every message passed to them.

Chapter 82: How to Log Messages to different Files | 305
YAML Specification
You can specify the configuration by many forms:
Listing 82-2
1
2
3
4
5
6
7
channels: ~
# Include all the channels
channels: foo # Include only channel 'foo'

channels: '!foo' # Include all channels, except 'foo'
channels: [foo, bar] # Include only channels 'foo' and 'bar'
channels: ['!foo', '!bar'] # Include all channels, except 'foo' and 'bar'
Creating your own Channel

You can change the channel monolog logs to one service at a time. This is done either via the
configuration below or by tagging your service with monolog.logger and specifying which channel the
service should log to. With the tag, the logger that is injected into that service is preconfigured to use the
channel you've specified.
Configure Additional Channels without Tagged Services

With MonologBundle 2.4 you can configure additional channels without the need to tag your services:
Listing 82-3
1
2
3
monolog:
channels: ['foo', 'bar']
With this, you can now send log messages to the foo channel by using the automatically registered logger
service monolog.logger.foo.
Learn more from the Cookbook

How to Use Monolog to Write Logs

Chapter 82: How to Log Messages to different Files | 306
Chapter 83
How to Create a custom Data Collector

The Symfony Profiler delegates data collection to some special classes called data collectors. Symfony
comes bundled with a few of them, but you can easily create your own.
Creating a custom Data Collector

Creating a custom data collector is as simple as implementing the DataCollectorInterface1:
Listing 83-1
1
2
3
4
5
interface DataCollectorInterface
{
function collect(Request $request, Response $response, \Exception $exception = null);
function getName();
}
The getName()2 method returns the name of the data collector and must be unique in the application.
This value is also used to access the information later on (see How to Use the Profiler in a Functional Test
for instance).
The collect()3 method is responsible for storing the collected data in local properties.
Most of the time, it is convenient to extend DataCollector4 and populate the $this->data property
(it takes care of serializing the $this->data property). Imagine you create a new data collector that
collects the method and accepted content types from the request:
Listing 83-2
1
2
3
4
5
6
7
8
// src/AppBundle/DataCollector/RequestCollector.php
namespace AppBundle\DataCollector;
use Symfony\Component\HttpKernel\DataCollector\DataCollector;
class RequestCollector extends DataCollector
{
public function collect(Request $request, Response $response, \Exception $exception = null)
1. http://api.symfony.com/2.7/Symfony/Component/HttpKernel/DataCollector/DataCollectorInterface.html
2. http://api.symfony.com/2.7/Symfony/Component/HttpKernel/DataCollector/DataCollectorInterface.html#method_getName
3. http://api.symfony.com/2.7/Symfony/Component/HttpKernel/DataCollector/DataCollectorInterface.html#method_collect
4. http://api.symfony.com/2.7/Symfony/Component/HttpKernel/DataCollector/DataCollector.html

Chapter 83: How to Create a custom Data Collector | 307
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
{
$this->data = array(
'method' => $request->getMethod(),
'acceptable_content_types' => $request->getAcceptableContentTypes(),
);
}
public function getMethod()
{
return $this->data['method'];
}
public function getAcceptableContentTypes()
{
return $this->data['acceptable_content_types'];
}
{
return 'app.request_collector';
}
}
The getters are added to give the template access to the collected information.
As the profiler serializes data collector instances, you should not store objects that cannot be
serialized (like PDO objects) or you need to provide your own serialize() method.
Enabling Custom Data Collectors

To enable a data collector, define it as a regular service and tag it as data_collector:
Listing 83-3
services:
app.request_collector:
class: AppBundle\DataCollector\RequestCollector
public: false
tags:
- { name: data_collector }
1
2
3
4
5
6
7
Adding Web Profiler Templates

The information collected by your data collector can be displayed both in the web debug toolbar and in
the web profiler. To do so, you need to create a Twig template that includes some specific blocks.
In the simplest case, you just want to display the information in the toolbar without providing a profiler
panel. This requires to define the toolbar block and set the value of two variables called icon and
text:
Listing 83-4
1
2
3
4
5
6
7
8
9
{% extends 'WebProfilerBundle:Profiler:layout.html.twig' %}
{% block toolbar %}
{% set icon %}
{# this is the content displayed as a panel in the toolbar #}
<span class="icon"><img src="..." alt=""/></span>
<span class="sf-toolbar-status">Request</span>
{% endset %}

10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
{% set text %}
{# this is the content displayed when hovering the mouse over
the toolbar panel #}
<div class="sf-toolbar-info-piece">
<b>Method</b>
<span>{{ collector.method }}</span>
</div>
<b>Accepted content type</b>
<span>{{ collector.acceptableContentTypes|join(', ') }}</span>
</div>
{% endset %}
{# the 'link' value set to 'false' means that this panel doesn't
show a section in the web profiler #}
{{ include('@WebProfiler/Profiler/toolbar_item.html.twig', { link: false }) }}
{% endblock %}
Built-in collector templates define all their images as embedded base64-encoded images. This makes
them work everywhere without having to mess with web assets links:
Listing 83-5
<img src="data:image/png;base64,..." />
Another solution is to define the images as SVG files. In addition to being resolution-independent,
these images can be easily embedded in the Twig template or included from an external file to reuse
them in several templates:
Listing 83-6
{{ include('@App/data_collector/icon.svg') }}
You are encouraged to use the latter technique for your own toolbar panels.
If the toolbar panel includes extended web profiler information, the Twig template must also define
additional blocks:
Listing 83-7
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
{% extends '@WebProfiler/Profiler/layout.html.twig' %}
{% block toolbar %}
{% set icon %}
<span class="sf-toolbar-status">Request</span>
{% endset %}
{% set text %}
{# ... #}
</div>
{% endset %}
{{ include('@WebProfiler/Profiler/toolbar_item.html.twig', { 'link': true }) }}
{% endblock %}
{% block head %}
{# Optional. Here you can link to or define your own CSS and JS contents. #}
{# Use {{ parent() }} to extend the default styles instead of overriding them. #}
{% endblock %}
{% block menu %}
{# This left-hand menu appears when using the full-screen profiler. #}
<span class="label">
<strong>Request</strong>
</span>
{% endblock %}

31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
{% block panel %}
{# Optional, for showing the most details. #}
<h2>Acceptable Content Types</h2>
<table>
<tr>
<th>Content Type</th>
</tr>
{% for type in collector.acceptableContentTypes %}
<tr>
<td>{{ type }}</td>
</tr>
{% endfor %}
</table>
{% endblock %}
The menu and panel blocks are the only required blocks to define the contents displayed in the web
profiler panel associated with this data collector. All blocks have access to the collector object.
Finally, to enable the data collector template, add a template attribute to the data_collector tag in
your service configuration:
Listing 83-8
1
2
3
4
5
6
7
8
9
10
services:
tags:
name:
data_collector
template: 'data_collector/template.html.twig'
id:
'app.request_collector'
public: false
The id attribute must match the value returned by the getName() method.
The position of each panel in the toolbar is determined by the priority defined by each collector. Most
built-in collectors use 255 as their priority. If you want your collector to be displayed before them, use a
higher value:
Listing 83-9
1
2
3
4
5
6
services:
tags:
- { name: data_collector, template: '...', id: '...', priority: 300 }

Chapter 84
How to Use Matchers to Enable the Profiler

Conditionally
The Symfony profiler is only activated in the development environment to not hurt your application
performance. However, sometimes it may be useful to conditionally enable the profiler in the production
environment to assist you in debugging issues. This behavior is implemented with the Request
Matchers.
Using the built-in Matcher

A request matcher is a class that checks whether a given Request instance matches a set of conditions.
Symfony provides a built-in matcher1 which matches paths and IPs. For example, if you want
to only show the profiler when accessing the page with the 168.0.0.1 IP, then you can use this
configuration:
Listing 84-1
1
2
3
4
5
6
framework:
# ...
profiler:
matcher:
ip: 168.0.0.1
You can also set a path option to define the path on which the profiler should be enabled. For instance,
setting it to ^/admin/ will enable the profiler only for the URLs which start with /admin/.
Creating a Custom Matcher

Leveraging the concept of Request Matchers you can define a custom matcher to enable the profiler
conditionally in your application. To do so, create a class which implements
RequestMatcherInterface2. This interface requires one method: matches()3. This method
1. http://api.symfony.com/2.7/Symfony/Component/HttpFoundation/RequestMatcher.html

Chapter 84: How to Use Matchers to Enable the Profiler Conditionally | 311
returns false when the request doesn't match the conditions and true otherwise. Therefore, the
custom matcher must return false to disable the profiler and true to enable it.
Suppose that the profiler must be enabled whenever a user with a ROLE_SUPER_ADMIN is logged in.
This is the only code needed for that custom matcher:
Listing 84-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
// src/AppBundle/Profiler/SuperAdminMatcher.php
namespace AppBundle\Profiler;
use Symfony\Component\Security\Core\Authorization\AuthorizationCheckerInterface;
use Symfony\Component\HttpFoundation\RequestMatcherInterface;
class SuperAdminMatcher implements RequestMatcherInterface
{
protected $authorizationChecker;
public function __construct(AuthorizationCheckerInterface $authorizationChecker)
{
$this->authorizationChecker = $authorizationChecker;
}
public function matches(Request $request)
{
return $this->authorizationChecker->isGranted('ROLE_SUPER_ADMIN');
}
}
New in version 2.6: The AuthorizationCheckerInterface4 was introduced in Symfony 2.6. Prior,
you had to use the isGranted method of SecurityContextInterface5.
Then, configure a new service and set it as private because the application won't use it directly:
Listing 84-3
1
2
3
4
5
6
services:
app.super_admin_matcher:
class: AppBundle\Profiler\SuperAdminMatcher
arguments: ['@security.authorization_checker']
public: false
New in version 2.6: The security.authorization_checker service was introduced in Symfony

2.6. Prior to Symfony 2.6, you had to use the isGranted() method of the security.context
service.
Once the service is registered, the only thing left to do is configure the profiler to use this service as the
matcher:
Listing 84-4
1
2
3
4
5
6
framework:
# ...
profiler:
matcher:
service: app.super_admin_matcher
2. http://api.symfony.com/2.7/Symfony/Component/HttpFoundation/RequestMatcherInterface.html
3. http://api.symfony.com/2.7/Symfony/Component/HttpFoundation/RequestMatcherInterface.html#method_matches
4. http://api.symfony.com/2.7/Symfony/Component/Security/Core/Authorization/AuthorizationCheckerInterface.html
5. http://api.symfony.com/2.7/Symfony/Component/Security/Core/SecurityContextInterface.html

Chapter 84: How to Use Matchers to Enable the Profiler Conditionally | 312
Chapter 85
Switching the Profiler Storage

By default the profile stores the collected data in files in the %kernel.cache_dir%/profiler/
directory. You can control the storage being used through the dsn, username, password and
lifetime options. For example, the following configuration uses MySQL as the storage for the profiler
with a lifetime of one hour:
Listing 85-1
1
2
3
4
5
6
7
framework:
profiler:
dsn:
'mysql:host=localhost;dbname=%database_name%'
username: '%database_user%'
password: '%database_password%'
lifetime: 3600
The HttpKernel component currently supports the following profiler storage drivers:
file
sqlite
mysql
mongodb
memcache
memcached
redis

Chapter 85: Switching the Profiler Storage | 313
Chapter 86
How to Access Profiling Data Programmatically

Most of the times, the profiler information is accessed and analyzed using its web-based visualizer.
However, you can also retrieve profiling information programmatically thanks to the methods provided
by the profiler service.
When the response object is available, use the
its associated profile:
Listing 86-1
loadProfileFromResponse()1 method to access to
// ... $profiler is the 'profiler' service

$profile = $profiler->loadProfileFromResponse($response);
When the profiler stores data about a request, it also associates a token with it; this token is available in
the X-Debug-Token HTTP header of the response. Using this token, you can access the profile of any
past response thanks to the loadProfile()2 method:
Listing 86-2
$token = $response->headers->get('X-Debug-Token');
$profile = $container->get('profiler')->loadProfile($token);
When the profiler is enabled but not the web debug toolbar, inspect the page with your browser's
developer tools to get the value of the X-Debug-Token HTTP header.
The profiler service also provides the find()3 method to look for tokens based on some criteria:
Listing 86-3
1
2
3
4
5
6
7
8
9
10
// get the latest 10 tokens

$tokens = $container->get('profiler')->find('', '', 10, '', '', '');
// get the latest 10 tokens for all URL containing /admin/
$tokens = $container->get('profiler')->find('', '/admin/', 10, '', '', '');
// get the latest 10 tokens for local POST requests
$tokens = $container->get('profiler')->find('127.0.0.1', '', 10, 'POST', '', '');
// get the latest 10 tokens for requests that happened between 2 and 4 days ago
1. http://api.symfony.com/2.7/Symfony/Component/HttpKernel/Profiler/Profiler.html#method_loadProfileFromResponse
2. http://api.symfony.com/2.7/Symfony/Component/HttpKernel/Profiler/Profiler.html#method_loadProfile
3. http://api.symfony.com/2.7/Symfony/Component/HttpKernel/Profiler/Profiler.html#method_find

Chapter 86: How to Access Profiling Data Programmatically | 314
11
12
$tokens = $container->get('profiler')
->find('', '', 10, '', '4 days ago', '2 days ago');
Lastly, if you want to manipulate profiling data on a different machine than the one where the
information was generated, use the profiler:export and profiler:import commands:
Listing 86-4
1
2
3
4
5
6
7
8
# on the production machine

$ php app/console profiler:export > profile.data
# on the development machine
$ php app/console profiler:import /path/to/profile.data
# you can also pipe from the STDIN
$ cat /path/to/profile.data | php app/console profiler:import

Chapter 86: How to Access Profiling Data Programmatically | 315
Chapter 87
The PSR-7 Bridge
The PSR-7 bridge converts HttpFoundation objects from and to objects implementing HTTP
message interfaces defined by the PSR-71.
Installation
You can install the component in 2 different ways:
Install it via Composer (symfony/psr-http-message-bridge on Packagist2);
Use the official Git repository (https://github.com/symfony/psr-http-message-bridge).
The bridge also needs a PSR-7 implementation to allow converting HttpFoundation objects to PSR-7
objects. It provides native support for Zend Diactoros3. Use Composer (zendframework/zend-diactoros on
Packagist4) or refer to the project documentation to install it.
Usage
Converting from HttpFoundation Objects to PSR-7
The bridge provides an interface of a factory called HttpMessageFactoryInterface5 that builds
objects implementing PSR-7 interfaces from HttpFoundation objects. It also provide a default
implementation using Zend Diactoros internally.
2. https://packagist.org/packages/symfony/psr-http-message-bridge
3. https://github.com/zendframework/zend-diactoros
4. https://packagist.org/packages/zendframework/zend-diactoros
5. http://api.symfony.com/2.7/Symfony/Bridge/PsrHttpMessage/HttpMessageFactoryInterface.html

Chapter 87: The PSR-7 Bridge | 316
The following code snippet explain how to convert a Request6 to a Zend Diactoros ServerRequest7
implementing the ServerRequestInterface8 interface:
Listing 87-1
1
2
3
4
5
6
7
8
use Symfony\Bridge\PsrHttpMessage\Factory\DiactorosFactory;
$symfonyRequest = new Request(array(), array(), array(), array(), array(), array('HTTP_HOST' => 'dunglas.fr'),
'Content');
// The HTTP_HOST server key must be set to avoid an unexpected error
$psr7Factory = new DiactorosFactory();
$psrRequest = $psr7Factory->createRequest($symfonyRequest);
And
now
from
Response9 to a Zend Diactoros Response10 implementing the
a
11
ResponseInterface interface:
Listing 87-2
1
2
3
4
5
6
7
use Symfony\Bridge\PsrHttpMessage\Factory\DiactorosFactory;
$symfonyResponse = new Response('Content');
$psr7Factory = new DiactorosFactory();
$psrResponse = $psr7Factory->createResponse($symfonyResponse);
Converting Objects implementing PSR-7 Interfaces to HttpFoundation

On
the
other
hand,
the
bridge
provide
factory
interface
called
HttpFoundationFactoryInterface12 that builds HttpFoundation objects from objects

implementing PSR-7 interfaces.
The next snippet explain how to convert an object implementing the
interface to a Request14 instance:
Listing 87-3
1
2
3
4
5
6
ServerRequestInterface13
use Symfony\Bridge\PsrHttpMessage\Factory\HttpFoundationFactory;
// $psrRequest is an instance of Psr\Http\Message\ServerRequestInterface

$httpFoundationFactory = new HttpFoundationFactory();
$symfonyRequest = $httpFoundationFactory->createRequest($psrRequest);
From an object implementing the ResponseInterface15 to a Response16 instance:

Listing 87-4
1
2
3
4
5
6
use Symfony\Bridge\PsrHttpMessage\Factory\HttpFoundationFactory;
// $psrResponse is an instance of Psr\Http\Message\ResponseInterface

$httpFoundationFactory = new HttpFoundationFactory();
$symfonyResponse = $httpFoundationFactory->createResponse($psrResponse);
7. http://api.symfony.com/2.7/Zend/Diactoros/ServerRequest.html
8. http://api.symfony.com/2.7/Psr/Http/Message/ServerRequestInterface.html
9. http://api.symfony.com/2.7/Symfony/Component/HttpFoundation/Response.html
10. http://api.symfony.com/2.7/Zend/Diactoros/Response.html
11. http://api.symfony.com/2.7/Psr/Http/Message/ResponseInterface.html
12. http://api.symfony.com/2.7/Symfony/Bridge/PsrHttpMessage/HttpFoundationFactoryInterface.html
13. http://api.symfony.com/2.7/Psr/Http/Message/ServerRequestInterface.html
15. http://api.symfony.com/2.7/Psr/Http/Message/ResponseInterface.html
16. http://api.symfony.com/2.7/Symfony/Component/HttpFoundation/Response.html

Chapter 87: The PSR-7 Bridge | 317
Chapter 88
How to Configure Symfony to Work behind a

Load Balancer or a Reverse Proxy
When you deploy your application, you may be behind a load balancer (e.g. an AWS Elastic Load
Balancer) or a reverse proxy (e.g. Varnish for caching).
For the most part, this doesn't cause any problems with Symfony. But, when a request passes through a
proxy, certain request information is sent using either the standard Forwarded header or non-standard
special X-Forwarded-* headers. For example, instead of reading the REMOTE_ADDR header (which will
now be the IP address of your reverse proxy), the user's true IP will be stored in a standard Forwarded:
for="..." header or a non standard X-Forwarded-For header.
New in version 2.7: Forwarded header support was introduced in Symfony 2.7.
If you don't configure Symfony to look for these headers, you'll get incorrect information about the
client's IP address, whether or not the client is connecting via HTTPS, the client's port and the hostname
being requested.
Solution: trusted_proxies
This is no problem, but you do need to tell Symfony what is happening and which reverse proxy IP
addresses will be doing this type of thing:
Listing 88-1
1
2
3
4
# ...
framework:
trusted_proxies: [192.0.0.1, 10.0.0.0/8]
In this example, you're saying that your reverse proxy (or proxies) has the IP address 192.0.0.1 or
matches the range of IP addresses that use the CIDR notation 10.0.0.0/8. For more details, see the
framework.trusted_proxies option.
You are also saying that you trust that the proxy does not send conflicting headers, e.g. sending both XForwarded-For and Forwarded in the same request.

Chapter 88: How to Configure Symfony to Work behind a Load Balancer or a Reverse Proxy | 318
That's it! Symfony will now look for the correct headers to get information like the client's IP address,
host, port and whether the request is using HTTPS.
But what if the IP of my Reverse Proxy Changes Constantly!

Some reverse proxies (like Amazon's Elastic Load Balancers) don't have a static IP address or even a range
that you can target with the CIDR notation. In this case, you'll need to - very carefully - trust all proxies.
1. Configure your web server(s) to not respond to traffic from any clients other than your load
balancers. For AWS, this can be done with security groups1.
2. Once you've guaranteed that traffic will only come from your trusted reverse proxies, configure
Symfony to always trust incoming request. This is done inside of your front controller:
Listing 88-2
1
2
3
4
5
6
7
// web/app.php
// ...
Request::setTrustedProxies(array('127.0.0.1', $request->server->get('REMOTE_ADDR')));
$response = $kernel->handle($request);
// ...
3. Ensure that the trusted_proxies setting in your app/config/config.yml is not set or it will
overwrite the setTrustedProxies call above.
That's it! It's critical that you prevent traffic from all non-trusted sources. If you allow outside traffic, they
could "spoof" their true IP address and other information.
My Reverse Proxy Sends X-Forwarded-For but Does not Filter the Forwarded
Header
Many popular proxy implementations do not yet support the Forwarded header and do not filter it by
default. Ideally, you would configure this in your proxy. If this is not possible, you can tell Symfony to
distrust the Forwarded header, while still trusting your proxy's X-Forwarded-For header.
This is done inside of your front controller:
Listing 88-3
1
2
3
4
5
6
7
// web/app.php
// ...
Request::setTrustedHeaderName(Request::HEADER_FORWARDED, null);
$response = $kernel->handle($request);
// ...
Configuring the proxy server trust is very important, as not doing so will allow malicious users to "spoof"
their IP address.
My Reverse Proxy Uses Non-Standard (not X-Forwarded) Headers

Although RFC 72392 recently defined a standard Forwarded header to disclose all proxy information,
most reverse proxies store information in non-standard X-Forwarded-* headers.
1. http://docs.aws.amazon.com/ElasticLoadBalancing/latest/DeveloperGuide/using-elb-security-groups.html
2. http://tools.ietf.org/html/rfc7239

But if your reverse proxy uses other non-standard header names, you can configure these (see "Trusting
Proxies").
The code for doing this will need to live in your front controller (e.g. web/app.php).

Chapter 89
How to Register a new Request Format and

Mime Type
Every Request has a "format" (e.g. html, json), which is used to determine what type of content to
return in the Response. In fact, the request format, accessible via getRequestFormat()1, is used to
set the MIME type of the Content-Type header on the Response object. Internally, Symfony contains
a map of the most common formats (e.g. html, json) and their associated MIME types (e.g. text/
html, application/json). Of course, additional format-MIME type entries can easily be added. This
document will show how you can add the jsonp format and corresponding MIME type.
Configure your New Format

The FrameworkBundle registers a subscriber that will add formats to incoming requests.
All you have to do is to configure the jsonp format:
Listing 89-1
1
2
3
4
5
framework:
request:
formats:
jsonp: 'application/javascript'
You can also associate multiple mime types to a format, but please note that the preferred one must
be the first as it will be used as the content type:
Listing 89-2
1
2
3
4
5
framework:
request:
formats:
csv: ['text/csv', 'text/plain']
1. http://api.symfony.com/2.7/Symfony/Component/HttpFoundation/Request.html#method_getRequestFormat

Chapter 89: How to Register a new Request Format and Mime Type | 321
Chapter 90
How to Force Routes to always Use HTTPS or

HTTP
Sometimes, you want to secure some routes and be sure that they are always accessed via the HTTPS
protocol. The Routing component allows you to enforce the URI scheme via schemes:
Listing 90-1
1
2
3
4
secure:
path:
/secure
defaults: { _controller: AppBundle:Main:secure }
schemes: [https]
The above configuration forces the secure route to always use HTTPS.
When generating the secure URL, and if the current scheme is HTTP, Symfony will automatically
generate an absolute URL with HTTPS as the scheme:
Listing 90-2
1
2
3
4
5
6
7
{# If the current scheme is HTTPS #}

{{ path('secure') }}
{# generates /secure #}
{# If the current scheme is HTTP #}
{{ path('secure') }}
{# generates https://example.com/secure #}
The requirement is also enforced for incoming requests. If you try to access the /secure path with
HTTP, you will automatically be redirected to the same URL, but with the HTTPS scheme.
The above example uses https for the scheme, but you can also force a URL to always use http.
The Security component provides another way to enforce HTTP or HTTPS via the
requires_channel setting. This alternative method is better suited to secure an "area" of your
website (all URLs under /admin) or when you want to secure URLs defined in a third party bundle
(see How to Force HTTPS or HTTP for different URLs for more details).

Chapter 90: How to Force Routes to always Use HTTPS or HTTP | 322
Chapter 91
How to Allow a "/" Character in a Route

Parameter
Sometimes, you need to compose URLs with parameters that can contain a slash /. For example, take
the classic /hello/{username} route. By default, /hello/Fabien will match this route but not
/hello/Fabien/Kris. This is because Symfony uses this character as separator between route parts.
This guide covers how you can modify a route so that /hello/Fabien/Kris matches the /hello/
{username} route, where {username} equals Fabien/Kris.
Configure the Route

By default, the Symfony Routing component requires that the parameters match the following regex
path: [^/]+. This means that all characters are allowed except /.
You must explicitly allow / to be part of your parameter by specifying a more permissive regex path.
Listing 91-1
1
2
3
4
5
6
7
8
9
10
11
12
use Sensio\Bundle\FrameworkExtraBundle\Configuration\Route;
class DemoController
{
/**
* @Route("/hello/{username}", name="_hello", requirements={"username"=".+"})
*/
public function helloAction($username)
{
// ...
}
}
That's it! Now, the {username} parameter can contain the / character.

Chapter 91: How to Allow a "/" Character in a Route Parameter | 323
Chapter 92
How to Configure a Redirect without a custom

Controller
Sometimes, a URL needs to redirect to another URL. You can do that by creating a new controller action
whose only task is to redirect, but using the RedirectController1 of the FrameworkBundle is even
easier.
You can redirect to a specific path (e.g. /about) or to a specific route using its name (e.g. homepage).
Redirecting Using a Path

Assume there is no default controller for the / path of your application and you want to redirect these
requests to /app. You will need to use the urlRedirectAction()2 action to redirect to this new url:
Listing 92-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
# load some routes - one should ultimately have the path "/app"
AppBundle:
resource: '@AppBundle/Controller/'
type:
annotation
prefix: /app
# redirecting the root
root:
path: /
defaults:
_controller: FrameworkBundle:Redirect:urlRedirect
path: /app
permanent: true
In this example, you configured a route for the / path and let the RedirectController redirect it to
/app. The permanent switch tells the action to issue a 301 HTTP status code instead of the default
302 HTTP status code.
1. http://api.symfony.com/2.7/Symfony/Bundle/FrameworkBundle/Controller/RedirectController.html
2. http://api.symfony.com/2.7/Symfony/Bundle/FrameworkBundle/Controller/RedirectController.html#method_urlRedirectAction

Chapter 92: How to Configure a Redirect without a custom Controller | 324
Redirecting Using a Route

Assume you are migrating your website from WordPress to Symfony, you want to redirect /wp-admin
to the route sonata_admin_dashboard. You don't know the path, only the route name. This can be
achieved using the redirectAction()3 action:
Listing 92-2
1
2
3
4
5
6
7
8
9
10
11
# ...
# redirecting the admin home
root:
path: /wp-admin
defaults:
_controller: FrameworkBundle:Redirect:redirect
route: sonata_admin_dashboard
permanent: true
Because you are redirecting to a route instead of a path, the required option is called route in the
redirect action, instead of path in the urlRedirect action.
3. http://api.symfony.com/2.7/Symfony/Bundle/FrameworkBundle/Controller/RedirectController.html#method_redirectAction

Chapter 92: How to Configure a Redirect without a custom Controller | 325
Chapter 93
How to Use HTTP Methods beyond GET and

POST in Routes
The HTTP method of a request is one of the requirements that can be checked when seeing if it matches
a route. This is introduced in the routing chapter of the book "Routing" with examples using GET and
POST. You can also use other HTTP verbs in this way. For example, if you have a blog post entry then
you could use the same URL path to show it, make changes to it and delete it by matching on GET, PUT
and DELETE.
Listing 93-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
blog_show:
path:
/blog/{slug}
defaults: { _controller: AppBundle:Blog:show }
methods: [GET]
blog_update:
path:
/blog/{slug}
defaults: { _controller: AppBundle:Blog:update }
methods: [PUT]
blog_delete:
path:
/blog/{slug}
defaults: { _controller: AppBundle:Blog:delete }
methods: [DELETE]
Faking the Method with _method

Unfortunately, life isn't quite this simple, since most browsers do not support sending PUT and DELETE
requests via the method attribute in an HTML form. Fortunately, Symfony provides you with a simple
way of working around this limitation. By including a _method parameter in the query string or
parameters of an HTTP request, Symfony will use this as the method when matching routes. Forms
automatically include a hidden field for this parameter if their submission method is not GET or POST.
See the related chapter in the forms documentation for more information.

Chapter 93: How to Use HTTP Methods beyond GET and POST in Routes | 326
You can disable the _method functionality shown here using the http_method_override option.

Chapter 93: How to Use HTTP Methods beyond GET and POST in Routes | 327
Chapter 94
How to Use Service Container Parameters in

your Routes
Sometimes you may find it useful to make some parts of your routes globally configurable. For instance,
if you build an internationalized site, you'll probably start with one or two locales. Surely you'll add a
requirement to your routes to prevent a user from matching a locale other than the locales you support.
You could hardcode your _locale requirement in all your routes, but a better solution is to use a
configurable service container parameter right inside your routing configuration:
Listing 94-1
1
2
3
4
5
6
contact:
path:
/{_locale}/contact
defaults: { _controller: AppBundle:Main:contact }
requirements:
_locale: '%app.locales%'
You can now control and set the app.locales parameter somewhere in your container:
Listing 94-2
1
2
3
parameters:
app.locales: en|es
You can also use a parameter to define your route path (or part of your path):
Listing 94-3
1
2
3
4
some_route:
path:
/%app.route_prefix%/contact
defaults: { _controller: AppBundle:Main:contact }
Just like in normal service container configuration files, if you actually need a % in your route,
you can escape the percent sign by doubling it, e.g. /score-50%%, which would resolve to
/score-50%.
However, as the % characters included in any URL are automatically encoded, the resulting URL of
this example would be /score-50%25 (%25 is the result of encoding the % character).

Chapter 94: How to Use Service Container Parameters in your Routes | 328
For parameter handling within a Dependency Injection Class see Using Parameters within a Dependency
Injection Class.

Chapter 94: How to Use Service Container Parameters in your Routes | 329
Chapter 95
How to Create a custom Route Loader
What is a Custom Route Loader

A custom route loader enables you to generate routes based on some conventions or patterns. A great
example for this use-case is the FOSRestBundle1 where routes are generated based on the names of the
action methods in a controller.
You still need to modify your routing configuration (e.g. app/config/routing.yml) manually, even
when using a custom route loader.
There are many bundles out there that use their own route loaders to accomplish cases like those
described above, for instance FOSRestBundle2, JMSI18nRoutingBundle3, KnpRadBundle4 and
SonataAdminBundle5.
Loading Routes
The routes in a Symfony application are loaded by the DelegatingLoader6. This loader uses several
other loaders (delegates) to load resources of different types, for instance YAML files or @Route and
@Method annotations in controller files. The specialized loaders implement LoaderInterface7 and
therefore have two important methods: supports()8 and load()9.
Take these lines from the routing.yml in the Symfony Standard Edition:
1.
2.
3.
4.
5.
https://github.com/FriendsOfSymfony/FOSRestBundle
https://github.com/FriendsOfSymfony/FOSRestBundle
https://github.com/schmittjoh/JMSI18nRoutingBundle
https://github.com/KnpLabs/KnpRadBundle
https://github.com/sonata-project/SonataAdminBundle
6. http://api.symfony.com/2.7/Symfony/Bundle/FrameworkBundle/Routing/DelegatingLoader.html
7. http://api.symfony.com/2.7/Symfony/Component/Config/Loader/LoaderInterface.html
8. http://api.symfony.com/2.7/Symfony/Component/Config/Loader/LoaderInterface.html#method_supports
9. http://api.symfony.com/2.7/Symfony/Component/Config/Loader/LoaderInterface.html#method_load

Chapter 95: How to Create a custom Route Loader | 330
Listing 95-1
1
2
3
4
app:
resource: '@AppBundle/Controller/'
type:
annotation
When the main loader parses this, it tries all registered delegate loaders and calls their supports()10
method with the given resource (@AppBundle/Controller/) and type (annotation) as arguments.
When one of the loader returns true, its load()11 method will be called, which should return a
RouteCollection12 containing Route13 objects.
Routes loaded this way will be cached by the Router the same way as when they are defined in one
of the default formats (e.g. XML, YML, PHP file).
Creating a custom Loader

To load routes from some custom source (i.e. from something other than annotations, YAML or XML
files), you need to create a custom route loader. This loader has to implement LoaderInterface14.
In most cases it is easier to extend from
yourself.
Loader15 instead of implementing LoaderInterface16
The sample loader below supports loading routing resources with a type of extra. The type name
should not clash with other loaders that might support the same type of resource. Just make up a name
specific to what you do. The resource name itself is not actually used in the example:
Listing 95-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
// src/AppBundle/Routing/ExtraLoader.php
namespace AppBundle\Routing;
use Symfony\Component\Config\Loader\Loader;
use Symfony\Component\Routing\Route;
use Symfony\Component\Routing\RouteCollection;
class ExtraLoader extends Loader
{
private $loaded = false;
public function load($resource, $type = null)
{
if (true === $this->loaded) {
throw new \RuntimeException('Do not add the "extra" loader twice');
}
$routes = new RouteCollection();
// prepare a new route

$path = '/extra/{parameter}';
$defaults = array(
'_controller' => 'AppBundle:Extra:extra',
);
$requirements = array(
'parameter' => '\d+',
12. http://api.symfony.com/2.7/Symfony/Component/Routing/RouteCollection.html
13. http://api.symfony.com/2.7/Symfony/Component/Routing/Route.html
15. http://api.symfony.com/2.7/Symfony/Component/Config/Loader/Loader.html

27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
);
$route = new Route($path, $defaults, $requirements);
// add the new route to the route collection

$routeName = 'extraRoute';
$routes->add($routeName, $route);
$this->loaded = true;
return $routes;
}
public function supports($resource, $type = null)
{
return 'extra' === $type;
}
}
Make sure the controller you specify really exists. In this case you have to create an extraAction
method in the ExtraController of the AppBundle:
Listing 95-3
1
2
3
4
5
6
7
8
9
10
11
12
13
// src/AppBundle/Controller/ExtraController.php
class ExtraController extends Controller
{
public function extraAction($parameter)
{
return new Response($parameter);
}
}
Now define a service for the ExtraLoader:

Listing 95-4
1
2
3
4
5
6
services:
app.routing_loader:
class: AppBundle\Routing\ExtraLoader
tags:
- { name: routing.loader }
Notice the tag routing.loader. All services with this tag will be marked as potential route loaders
and added as specialized route loaders to the routing.loader service, which is an instance of
DelegatingLoader17.
Using the custom Loader

If you did nothing else, your custom routing loader would not be called. What remains to do is adding a
few lines to the routing configuration:
Listing 95-5
1
2
3
4
app_extra:
resource: .
type: extra
17. http://api.symfony.com/2.7/Symfony/Bundle/FrameworkBundle/Routing/DelegatingLoader.html

The important part here is the type key. Its value should be "extra" as this is the type which the
ExtraLoader supports and this will make sure its load() method gets called. The resource key is
insignificant for the ExtraLoader, so it is set to ".".
The routes defined using custom route loaders will be automatically cached by the framework. So
whenever you change something in the loader class itself, don't forget to clear the cache.
More advanced Loaders

If your custom route loader extends from Loader18 as shown above, you can also make use of the
provided resolver, an instance of LoaderResolver19, to load secondary routing resources.
Of course you still need to implement supports()20 and load()21. Whenever you want to load
another resource - for instance a YAML routing configuration file - you can call the import()22 method:
Listing 95-6
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
// src/AppBundle/Routing/AdvancedLoader.php
namespace AppBundle\Routing;
use Symfony\Component\Config\Loader\Loader;
use Symfony\Component\Routing\RouteCollection;
class AdvancedLoader extends Loader
{
public function load($resource, $type = null)
{
$collection = new RouteCollection();
$resource = '@AppBundle/Resources/config/import_routing.yml';
$type = 'yaml';
$importedRoutes = $this->import($resource, $type);
$collection->addCollection($importedRoutes);
return $collection;
}
public function supports($resource, $type = null)
{
return 'advanced_extra' === $type;
}
}
The resource name and type of the imported routing configuration can be anything that would
normally be supported by the routing configuration loader (YAML, XML, PHP, annotation, etc.).
18. http://api.symfony.com/2.7/Symfony/Component/Config/Loader/Loader.html
19. http://api.symfony.com/2.7/Symfony/Component/Config/Loader/LoaderResolver.html
22. http://api.symfony.com/2.7/Symfony/Component/Config/Loader/Loader.html#method_import

Chapter 96
Redirect URLs with a Trailing Slash

The goal of this cookbook is to demonstrate how to redirect URLs with a trailing slash to the same URL
without a trailing slash (for example /en/blog/ to /en/blog).
Create a controller that will match any URL with a trailing slash, remove the trailing slash (keeping query
parameters if any) and redirect to the new URL with a 301 response status code:
Listing 96-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
// src/AppBundle/Controller/RedirectingController.php
class RedirectingController extends Controller
{
public function removeTrailingSlashAction(Request $request)
{
$pathInfo = $request->getPathInfo();
$requestUri = $request->getRequestUri();
$url = str_replace($pathInfo, rtrim($pathInfo, ' /'), $requestUri);
return $this->redirect($url, 301);
}
}
After that, create a route to this controller that's matched whenever a URL with a trailing slash is
requested. Be sure to put this route last in your system, as explained below:
Listing 96-2
1
2
3
4
5
6
7
8
9
10
11
12
13
// src/AppBundle/Controller/RedirectingController.php
class RedirectingController extends Controller
{
/**
* @Route("/{url}", name="remove_trailing_slash",
*
requirements={"url" = ".*\/$"}, methods={"GET"})
*/
public function removeTrailingSlashAction(Request $request)

Chapter 96: Redirect URLs with a Trailing Slash | 334
14
15
16
17
// ...
}
}
Redirecting a POST request does not work well in old browsers. A 302 on a POST request would
send a GET request after the redirection for legacy reasons. For that reason, the route here only
matches GET requests.
Make sure to include this route in your routing configuration at the very end of your route listing.
Otherwise, you risk redirecting real routes (including Symfony core routes) that actually do have a
trailing slash in their path.

Chapter 96: Redirect URLs with a Trailing Slash | 335
Chapter 97
How to Pass Extra Information from a Route to

a Controller
Parameters inside the defaults collection don't necessarily have to match a placeholder in the route
path. In fact, you can use the defaults array to specify extra parameters that will then be accessible as
arguments to your controller, and as attributes of the Request object:
Listing 97-1
1
2
3
4
5
6
7
blog:
path:
/blog/{page}
defaults:
_controller: AppBundle:Blog:index
page:
1
title:
"Hello world!"
Now, you can access this extra parameter in your controller, as an argument to the controller method:
Listing 97-2
public function indexAction($page, $title)

{
// ...
}
Alternatively, the title could be accessed through the Request object:

Listing 97-3
1
2
3
4
5
6
7
8
public function indexAction(Request $request, $page)
{
$title = $request->attributes->get('title');
// ...
}
As you can see, the $title variable was never defined inside the route path, but you can still access
its value from inside your controller, through the method's argument, or from the Request object's
attributes bag.

Chapter 97: How to Pass Extra Information from a Route to a Controller | 336
Chapter 98
Looking up Routes from a Database: Symfony

CMF DynamicRouter
The core Symfony Routing System is excellent at handling complex sets of routes. A highly optimized
routing cache is dumped during deployments.
However, when working with large amounts of data that each need a nice readable URL (e.g. for search
engine optimization purposes), the routing can get slowed down. Additionally, if routes need to be edited
by users, the route cache would need to be rebuilt frequently.
For these cases, the DynamicRouter offers an alternative approach:
Routes are stored in a database;
There is a database index on the path field, the lookup scales to huge numbers of different routes;
Writes only affect the index of the database, which is very efficient.
When all routes are known during deploy time and the number is not too high, using a custom route
loader is the preferred way to add more routes. When working with just one type of objects, a slug
parameter on the object and the @ParamConverter annotation work fine (see FrameworkExtraBundle1)
.
The DynamicRouter is useful when you need Route objects with the full feature set of Symfony. Each
route can define a specific controller so you can decouple the URL structure from your application logic.
The DynamicRouter comes with built-in support for Doctrine ORM and Doctrine PHPCR-ODM but
offers the ContentRepositoryInterface to write a custom loader, e.g. for another database type or
a REST API or anything else.
The DynamicRouter is explained in the Symfony CMF documentation2.
1. https://symfony.com/doc/current/bundles/SensioFrameworkExtraBundle/annotations/converters.html
2. http://symfony.com/doc/master/cmf/book/routing.html

Chapter 98: Looking up Routes from a Database: Symfony CMF DynamicRouter | 337
Chapter 99
How to Build a Traditional Login Form
If you need a login form and are storing users in some sort of a database, then you should consider
using FOSUserBundle1, which helps you build your User object and gives you many routes and
controllers for common tasks like login, registration and forgot password.
In this entry, you'll build a traditional login form. Of course, when the user logs in, you can load your
users from anywhere - like the database. See B) Configuring how Users are Loaded for details.
First, enable form login under your firewall:
Listing 99-1
1
2
3
4
5
6
7
8
9
10
security:
# ...
firewalls:
main:
anonymous: ~
form_login:
login_path: login
check_path: login
The login_path and check_path can also be route names (but cannot have mandatory wildcards
- e.g. /login/{foo} where foo has no default value).
Now, when the security system initiates the authentication process, it will redirect the user to the
login form /login. Implementing this login form visually is your job. First, create a new
SecurityController inside a bundle:
Listing 99-2
1
2
3
4
5
6
// src/AppBundle/Controller/SecurityController.php
class SecurityController extends Controller

Chapter 99: How to Build a Traditional Login Form | 338
7
8
{
}
Next, configure the route that you earlier used under your form_login configuration (login):
Listing 99-3
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
// ...
use Sensio\Bundle\FrameworkExtraBundle\Configuration\Route;
class SecurityController extends Controller
{
/**
* @Route("/login", name="login")
*/
public function loginAction(Request $request)
{
}
}
Great! Next, add the logic to loginAction that will display the login form:
Listing 99-4
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
public function loginAction(Request $request)
{
$authenticationUtils = $this->get('security.authentication_utils');
// get the login error if there is one

$error = $authenticationUtils->getLastAuthenticationError();
// last username entered by the user
$lastUsername = $authenticationUtils->getLastUsername();
'security/login.html.twig',
array(
// last username entered by the user
'last_username' => $lastUsername,
'error'
=> $error,
)
);
}
in
version
2.6:
The
security.authentication_utils
AuthenticationUtils2 class were introduced in Symfony 2.6.
New
service
and
the
Don't let this controller confuse you. As you'll see in a moment, when the user submits the form, the
security system automatically handles the form submission for you. If the user had submitted an invalid
username or password, this controller reads the form submission error from the security system so that it
can be displayed back to the user.
In other words, your job is to display the login form and any login errors that may have occurred, but the
security system itself takes care of checking the submitted username and password and authenticating
the user.
Finally, create the template:
Listing 99-5
1
2
3
4
{# app/Resources/views/security/login.html.twig #}
{# ... you will probably extends your base template, like base.html.twig #}
{% if error %}
2. http://api.symfony.com/2.7/Symfony/Component/Security/Http/Authentication/AuthenticationUtils.html

5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
<div>{{ error.messageKey|trans(error.messageData, 'security') }}</div>

{% endif %}
<form action="{{ path('login') }}" method="post">
<label for="username">Username:</label>
<input type="text" id="username" name="_username" value="{{ last_username }}" />
<label for="password">Password:</label>
<input type="password" id="password" name="_password" />
{#
If you want to control the URL the user
is redirected to on success (more details below)
<input type="hidden" name="_target_path" value="/account" />
#}
<button type="submit">login</button>
</form>
The error variable passed into the template is an instance of AuthenticationException3. It

may contain more information - or even sensitive information - about the authentication failure, so
use it wisely!
The form can look like anything, but has a few requirements:
The form must POST to the login route, since that's what you configured under the form_login key in
security.yml.
The username must have the name _username and the password must have the name _password.
Actually, all of this can be configured under the form_login key. See Form Login Configuration
for more details.
This login form is currently not protected against CSRF attacks. Read Using CSRF Protection in the
Login Form on how to protect your login form.
And that's it! When you submit the form, the security system will automatically check the user's
credentials and either authenticate the user or send the user back to the login form where the error can
be displayed.
To review the whole process:
1. The user tries to access a resource that is protected;
2. The firewall initiates the authentication process by redirecting the user to the login form (/login);
3. The /login page renders login form via the route and controller created in this example;
4. The user submits the login form to /login;
5. The security system intercepts the request, checks the user's submitted credentials,
authenticates the user if they are correct, and sends the user back to the login form if they are
not.
Redirecting after Success

If the submitted credentials are correct, the user will be redirected to the original page that was requested
(e.g. /admin/foo). If the user originally went straight to the login page, they'll be redirected to the
homepage. This can all be customized, allowing you to, for example, redirect the user to a specific URL.
3. http://api.symfony.com/2.7/Symfony/Component/Security/Core/Exception/AuthenticationException.html

For more details on this and how to customize the form login process in general, see How to Customize
your Form Login.
Avoid Common Pitfalls

When setting up your login form, watch out for a few common pitfalls.
1. Create the Correct Routes

First, be sure that you've defined the /login route correctly and that it corresponds to the login_path
and check_path config values. A misconfiguration here can mean that you're redirected to a 404 page
instead of the login page, or that submitting the login form does nothing (you just see the login form over
and over again).
2. Be Sure the Login Page Isn't Secure (Redirect Loop!)

Also, be sure that the login page is accessible by anonymous users. For example, the following
configuration - which requires the ROLE_ADMIN role for all URLs (including the /login URL), will
cause a redirect loop:
Listing 99-6
1
2
3
4
5
# ...
access_control:
- { path: ^/, roles: ROLE_ADMIN }
Adding an access control that matches /login/* and requires no authentication fixes the problem:
Listing 99-7
1
2
3
4
5
6
# ...
access_control:
- { path: ^/login, roles: IS_AUTHENTICATED_ANONYMOUSLY }
- { path: ^/, roles: ROLE_ADMIN }
Also, if your firewall does not allow for anonymous users (no anonymous key), you'll need to create a
special firewall that allows anonymous users for the login page:
Listing 99-8
1
2
3
4
5
6
7
8
9
10
11
# ...
firewalls:
# order matters! This must be before the ^/ firewall
login_firewall:
pattern:
^/login$
anonymous: ~
secured_area:
pattern:
^/
form_login: ~
3. Be Sure check_path Is Behind a Firewall

Next, make sure that your check_path URL (e.g. /login) is behind the firewall you're using for your
form login (in this example, the single firewall matches all URLs, including /login). If /login doesn't
match any firewall, you'll receive a Unable to find the controller for path "/login"
exception.

4. Multiple Firewalls Don't Share the Same Security Context

If you're using multiple firewalls and you authenticate against one firewall, you will not be authenticated
against any other firewalls automatically. Different firewalls are like different security systems. To do
this you have to explicitly specify the same Firewall Context for different firewalls. But usually for most
applications, having one main firewall is enough.
5. Routing Error Pages Are not Covered by Firewalls

As routing is done before security, 404 error pages are not covered by any firewall. This means you can't
check for security or even access the user object on these pages. See How to Customize Error Pages for
more details.

Chapter 100
How to Load Security Users from the Database

(the Entity Provider)
Symfony's security system can load security users from anywhere - like a database, via Active Directory
or an OAuth server. This article will show you how to load your users from the database via a Doctrine
entity.
Introduction
Before you start, you should check out FOSUserBundle1. This external bundle allows you to load
users from the database (like you'll learn here) and gives you built-in routes & controllers for things
like login, registration and forgot password. But, if you need to heavily customize your user system
or if you want to learn how things work, this tutorial is even better.
Loading users via a Doctrine entity has 2 basic steps:

1. Create your User entity
2. Configure security.yml to load from your entity
Afterwards, you can learn more about forbidding inactive users, using a custom query and user
serialization to the session
1) Create your User Entity

For this entry, suppose that you already have a User entity inside an AppBundle with the following
fields: id, username, password, email and isActive:
Listing 100-1
1
2
3

Chapter 100: How to Load Security Users from the Database (the Entity Provider) | 343
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74

use Symfony\Component\Security\Core\User\UserInterface;
/**
* @ORM\Table(name="app_users")
* @ORM\Entity(repositoryClass="AppBundle\Entity\UserRepository")
*/
class User implements UserInterface, \Serializable
{
/**
* @ORM\Column(type="integer")
* @ORM\Id
* @ORM\GeneratedValue(strategy="AUTO")
*/
private $id;
/**
*/
private $username;
/**
* @ORM\Column(type="string", length=64)
*/
private $password;
/**
*/
private $email;
/**
* @ORM\Column(name="is_active", type="boolean")
*/
private $isActive;
public function __construct()
{
$this->isActive = true;
// may not be needed, see section on salt below
// $this->salt = md5(uniqid(null, true));
}
{
}
{
// you *may* need a real salt depending on your encoder
// see section on salt below
return null;
}
public function getPassword()
{
return $this->password;
}
public function getRoles()
{
return array('ROLE_USER');
}
public function eraseCredentials()
{
}
/** @see \Serializable::serialize() */

public function serialize()

75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
{
return serialize(array(
$this->id,
$this->username,
$this->password,
// $this->salt,
));
}
/** @see \Serializable::unserialize() */

public function unserialize($serialized)
{
list (
$this->id,
$this->username,
$this->password,
// $this->salt
) = unserialize($serialized);
}
}
To make things shorter, some of the getter and setter methods aren't shown. But you can generate these
by running:
Listing 100-2
$ php app/console doctrine:generate:entities AppBundle/Entity/User
Next, make sure to create the database table:

Listing 100-3
What's this UserInterface?

So far, this is just a normal entity. But to use this class in the security system, it must implement
UserInterface2. This forces the class to have the five following methods:
getRoles()3
getPassword()4
getSalt()5
getUsername()6
eraseCredentials()7
To learn more about each of these, see UserInterface8.
What do the serialize and unserialize Methods do?

At the end of each request, the User object is serialized to the session. On the next request, it's
unserialized. To help PHP do this correctly, you need to implement Serializable. But you don't need
to serialize everything: you only need a few fields (the ones shown above plus a few extra if you decide
to implement AdvancedUserInterface). On each request, the id is used to query for a fresh User object
from the database.
Want to know more? See Understanding serialize and how a User is Saved in the Session.
2.
3.
4.
5.
6.
7.
http://api.symfony.com/2.7/Symfony/Component/Security/Core/User/UserInterface.html
http://api.symfony.com/2.7/Symfony/Component/Security/Core/User/UserInterface.html#method_getRoles
http://api.symfony.com/2.7/Symfony/Component/Security/Core/User/UserInterface.html#method_getPassword
http://api.symfony.com/2.7/Symfony/Component/Security/Core/User/UserInterface.html#method_getSalt
http://api.symfony.com/2.7/Symfony/Component/Security/Core/User/UserInterface.html#method_getUsername
http://api.symfony.com/2.7/Symfony/Component/Security/Core/User/UserInterface.html#method_eraseCredentials

2) Configure Security to load from your Entity

Now that you have a User entity that implements UserInterface, you just need to tell Symfony's
security system about it in security.yml.
In this example, the user will enter their username and password via HTTP basic authentication. Symfony
will query for a User entity matching the username and then check the password (more on passwords in
a moment):
Listing 100-4
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
security:
encoders:
AppBundle\Entity\User:
algorithm: bcrypt
# ...
providers:
our_db_provider:
entity:
class: AppBundle:User
property: username
# if you're using multiple entity managers
# manager_name: customer
firewalls:
main:
pattern:
^/
http_basic: ~
provider: our_db_provider
# ...
First, the encoders section tells Symfony to expect that the passwords in the database will be encoded
using bcrypt. Second, the providers section creates a "user provider" called our_db_provider
that knows to query from your AppBundle:User entity by the username property. The name
our_db_provider isn't important: it just needs to match the value of the provider key under
your firewall. Or, if you don't set the provider key under your firewall, the first "user provider" is
automatically used.
If you're using PHP 5.4 or lower, you'll need to install the ircmaxell/password-compat library
via Composer in order to be able to use the bcrypt encoder:
Listing 100-5
$ composer require ircmaxell/password-compat "~1.0"
Creating your First User

To add users, you can implement a registration form or add some fixtures9. This is just a normal entity,
so there's nothing tricky, except that you need to encode each user's password. But don't worry, Symfony
gives you a service that will do this for you. See Dynamically Encoding a Password for details.
Below is an export of the app_users table from MySQL with user admin and password admin (which
has been encoded).
Listing 100-6
1
2
3
$ mysql> SELECT * FROM app_users;

+----+----------+--------------------------------------------------------------+--------------------+-----------+
| id | username | password
| email
| is_active |
9. https://symfony.com/doc/master/bundles/DoctrineFixturesBundle/index.html

4
5
6
+----+----------+--------------------------------------------------------------+--------------------+-----------+
| 1 | admin
| $2a$08$jHZj/wJfcVKlIwr5AvR78euJxYK7Ku5kURNhNx.7.CSIJ3Pq6LEPC | [email protected] |
1 |
+----+----------+--------------------------------------------------------------+--------------------+-----------+
Do you need to use a Salt property?

If you use bcrypt, no. Otherwise, yes. All passwords must be hashed with a salt, but bcrypt does
this internally. Since this tutorial does use bcrypt, the getSalt() method in User can just return
null (it's not used). If you use a different algorithm, you'll need to uncomment the salt lines in the
User entity and add a persisted salt property.
Forbid Inactive Users (AdvancedUserInterface)

If a User's isActive property is set to false (i.e. is_active is 0 in the database), the user will still be
able to login to the site normally. This is easily fixable.
To exclude inactive users, change your User class to implement
extends UserInterface11, so you only need the new interface:
Listing 100-7
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
AdvancedUserInterface10. This
use Symfony\Component\Security\Core\User\AdvancedUserInterface;
// ...
class User implements AdvancedUserInterface, \Serializable
{
// ...
public function isAccountNonExpired()
{
return true;
}
public function isAccountNonLocked()
{
return true;
}
public function isCredentialsNonExpired()
{
return true;
}
public function isEnabled()
{
return $this->isActive;
}
// serialize and unserialize must be updated - see below

public function serialize()
{
return serialize(array(
// ...
$this->isActive
));
}
public function unserialize($serialized)
{
10. http://api.symfony.com/2.7/Symfony/Component/Security/Core/User/AdvancedUserInterface.html

40
41
42
43
44
45
list (
// ...
$this->isActive
) = unserialize($serialized);
}
}
The AdvancedUserInterface12 interface adds four extra methods to validate the account status:
checks whether the user's account has expired;

checks whether the user is locked;
isCredentialsNonExpired()15 checks whether the user's credentials (password) has expired;
isEnabled()16 checks whether the user is enabled.
isAccountNonExpired()13
isAccountNonLocked()14
If any of these return false, the user won't be allowed to login. You can choose to have persisted
properties for all of these, or whatever you need (in this example, only isActive pulls from the
database).
So what's the difference between the methods? Each returns a slightly different error message (and these
can be translated when you render them in your login template to customize them further).
If you use AdvancedUserInterface, you also need to add any of the properties used by these
methods (like isActive) to the serialize() and unserialize() methods. If you don't do
this, your user may not be deserialized correctly from the session on each request.
Congrats! Your database-loading security system is all setup! Next, add a true login form instead of HTTP
Basic or keep reading for other topics.
Using a Custom Query to Load the User

It would be great if a user could login with their username or email, as both are unique in the database.
Unfortunately, the native entity provider is only able to handle querying via a single property on the user.
To do this, make your UserRepository implement a special UserProviderInterface17. This
interface
requires
three
methods:
loadUserByUsername($username),
refreshUser(UserInterface $user), and supportsClass($class):
Listing 100-8
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
12.
13.
14.
15.
16.
// src/AppBundle/Entity/UserRepository.php
use
use
use
use
Symfony\Component\Security\Core\User\UserProviderInterface;
Symfony\Component\Security\Core\Exception\UnsupportedUserException;
Doctrine\ORM\EntityRepository;
class UserRepository extends EntityRepository implements UserProviderInterface

{
public function loadUserByUsername($username)
{
return $this->createQueryBuilder('u')
->where('u.username = :username OR u.email = :email')
->setParameter('username', $username)
->setParameter('email', $username)
http://api.symfony.com/2.7/Symfony/Component/Security/Core/User/AdvancedUserInterface.html
http://api.symfony.com/2.7/Symfony/Component/Security/Core/User/AdvancedUserInterface.html#method_isAccountNonExpired
http://api.symfony.com/2.7/Symfony/Component/Security/Core/User/AdvancedUserInterface.html#method_isAccountNonLocked
http://api.symfony.com/2.7/Symfony/Component/Security/Core/User/AdvancedUserInterface.html#method_isCredentialsNonExpired
http://api.symfony.com/2.7/Symfony/Component/Security/Core/User/AdvancedUserInterface.html#method_isEnabled
17. http://api.symfony.com/2.7/Symfony/Component/Security/Core/User/UserProviderInterface.html

17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
->getQuery()
->getOneOrNullResult();
}
public function refreshUser(UserInterface $user)
{
$class = get_class($user);
if (!$this->supportsClass($class)) {
throw new UnsupportedUserException(
sprintf(
'Instances of "%s" are not supported.',
$class
)
);
}
return $this->find($user->getId());
}
public function supportsClass($class)
{
return $this->getEntityName() === $class
|| is_subclass_of($class, $this->getEntityName());
}
}
For more details on these methods, see UserProviderInterface18.

Don't forget to add the repository class to the mapping definition of your entity.
To finish this, just remove the property key from the user provider in security.yml:
Listing 100-9
1
2
3
4
5
6
7
8
security:
# ...
providers:
our_db_provider:
entity:
class: AppBundle:User
This tells Symfony to not query automatically for the User. Instead, when someone logs in, the
loadUserByUsername() method on UserRepository will be called.
Understanding serialize and how a User is Saved in the Session

If you're curious about the importance of the serialize() method inside the User class or how the
User object is serialized or deserialized, then this section is for you. If not, feel free to skip this.
Once the user is logged in, the entire User object is serialized into the session. On the next request, the
User object is deserialized. Then, the value of the id property is used to re-query for a fresh User object
from the database. Finally, the fresh User object is compared to the deserialized User object to make sure
that they represent the same user. For example, if the username on the 2 User objects doesn't match for
some reason, then the user will be logged out for security reasons.
Even though this all happens automatically, there are a few important side-effects.

First, the Serializable19 interface and its serialize and unserialize methods have been added
to allow the User class to be serialized to the session. This may or may not be needed depending on
your setup, but it's probably a good idea. In theory, only the id needs to be serialized, because the
refreshUser()20 method refreshes the user on each request by using the id (as explained above). This
gives us a "fresh" User object.
But Symfony also uses the username, salt, and password to verify that the User has not changed
between requests (it also calls your AdvancedUserInterface methods if you implement it). Failing
to serialize these may cause you to be logged out on each request. If your User implements the
EquatableInterface21, then instead of these properties being checked, your isEqualTo method
is simply called, and you can check whatever properties you want. Unless you understand this, you
probably won't need to implement this interface or worry about it.
19. http://php.net/manual/en/class.serializable.php
20. http://api.symfony.com/2.7/Symfony/Bridge/Doctrine/Security/User/EntityUserProvider.html#method_refreshUser
21. http://api.symfony.com/2.7/Symfony/Component/Security/Core/User/EquatableInterface.html

Chapter 101
How to Add "Remember Me" Login

Functionality
Once a user is authenticated, their credentials are typically stored in the session. This means that when
the session ends they will be logged out and have to provide their login details again next time they wish
to access the application. You can allow users to choose to stay logged in for longer than the session lasts
using a cookie with the remember_me firewall option:
Listing 101-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
security:
# ...
firewalls:
main:
# ...
remember_me:
key:
'%secret%'
lifetime: 604800 # 1 week in seconds
path:
/
# by default, the feature is enabled by checking a
# checkbox in the login form (see below), uncomment the
# following line to always enable it.
#always_remember_me: true
The remember_me firewall defines the following configuration options:

key (required)
The value used to encrypt the cookie's content. It's common to use the
app/config/parameters.yml file.
secret
value defined in the
name (default
value: REMEMBERME)
The name of the cookie used to keep the user logged in. If you enable the remember_me feature in
several firewalls of the same application, make sure to choose a different name for the cookie of each
firewall. Otherwise, you'll face lots of security related problems.
lifetime (default
value: 31536000)
The number of seconds during which the user will remain logged in. By default users are logged in
for one year.

Chapter 101: How to Add "Remember Me" Login Functionality | 351
path (default
value: /)
The path where the cookie associated with this feature is used. By default the cookie will be applied
to the entire website but you can restrict to a specific section (e.g. /forum, /admin).
domain (default
value: null)
The domain where the cookie associated with this feature is used. By default cookies use the current
domain obtained from $_SERVER.
secure (default
value: false)
If true, the cookie associated with this feature is sent to the user through an HTTPS secure
connection.
httponly (default
value: true)
If true, the cookie associated with this feature is accessible only through the HTTP protocol. This
means that the cookie won't be accessible by scripting languages, such as JavaScript.
remember_me_parameter (default
value: _remember_me)
The name of the form field checked to decide if the "Remember Me" feature should be enabled or
not. Keep reading this article to know how to enable this feature conditionally.
always_remember_me (default
value: false)
If true, the value of the remember_me_parameter is ignored and the "Remember Me" feature is always
enabled, regardless of the desire of the end user.
token_provider (default
value: null)
Defines the service id of a token provider to use. By default, tokens are stored in a cookie.
For example, you might want to store the token in a database, to not have a (hashed) version
of
the
password
in
a
cookie.
The
DoctrineBridge
comes
with
a
Symfony\Bridge\Doctrine\Security\RememberMe\DoctrineTokenProvider that you can use.
Forcing the User to Opt-Out of the Remember Me Feature

It's a good idea to provide the user with the option to use or not use the remember me functionality,
as it will not always be appropriate. The usual way of doing this is to add a checkbox to the login
form. By giving the checkbox the name _remember_me (or the name you configured using
remember_me_parameter), the cookie will automatically be set when the checkbox is checked and
the user successfully logs in. So, your specific login form might ultimately look like this:
Listing 101-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
{# app/Resources/views/security/login.html.twig #}
{% if error %}
<div>{{ error.message }}</div>
{% endif %}
<input type="checkbox" id="remember_me" name="_remember_me" checked />
<label for="remember_me">Keep me logged in</label>
<input type="submit" name="login" />
</form>
The user will then automatically be logged in on subsequent visits while the cookie remains valid.

Forcing the User to Re-Authenticate before Accessing certain Resources

When the user returns to your site, they are authenticated automatically based on the information stored
in the remember me cookie. This allows the user to access protected resources as if the user had actually
authenticated upon visiting the site.
In some cases, however, you may want to force the user to actually re-authenticate before accessing
certain resources. For example, you might allow "remember me" users to see basic account information,
but then require them to actually re-authenticate before modifying that information.
The Security component provides an easy way to do this. In addition to roles explicitly assigned to them,
users are automatically given one of the following roles depending on how they are authenticated:
IS_AUTHENTICATED_ANONYMOUSLY
Automatically assigned to a user who is in a firewall protected part of the site but who has not
actually logged in. This is only possible if anonymous access has been allowed.
IS_AUTHENTICATED_REMEMBERED
Automatically assigned to a user who was authenticated via a remember me cookie.

IS_AUTHENTICATED_FULLY
Automatically assigned to a user that has provided their login details during the current session.
You can use these to control access beyond the explicitly assigned roles.
you have the IS_AUTHENTICATED_REMEMBERED role, then you also have the
IS_AUTHENTICATED_ANONYMOUSLY role. If you have the IS_AUTHENTICATED_FULLY role,
If
then you also have the other two roles. In other words, these roles represent three levels of increasing
"strength" of authentication.
You can use these additional roles for finer grained control over access to parts of a site. For example,
you may want your user to be able to view their account at /account when authenticated by cookie but
to have to provide their login details to be able to edit the account details. You can do this by securing
specific controller actions using these roles. The edit action in the controller could be secured using the
service context.
In the following example, the action is only allowed if the user has the IS_AUTHENTICATED_FULLY
role.
Listing 101-3
1
2
3
4
5
6
7
8
9
10
// ...
use Symfony\Component\Security\Core\Exception\AccessDeniedException
// ...
public function editAction()
{
$this->denyAccessUnlessGranted('IS_AUTHENTICATED_FULLY');
// ...
}
If your application is based on the Symfony Standard Edition, you can also secure your controller using
annotations:
Listing 101-4
1
2
3
4
5
6
7
use Sensio\Bundle\FrameworkExtraBundle\Configuration\Security;
/**
* @Security("has_role('IS_AUTHENTICATED_FULLY')")
*/
public function editAction($name)
{

8
9
// ...
}
If you also had an access control in your security configuration that required the user to have a
ROLE_USER role in order to access any of the account area, then you'd have the following situation:
If a non-authenticated (or anonymously authenticated user) tries to access the account area,
the user will be asked to authenticate.
Once the user has entered their username and password, assuming the user receives the
ROLE_USER role per your configuration, the user will have the IS_AUTHENTICATED_FULLY role and be able
to access any page in the account section, including the editAction controller.
If the user's session ends, when the user returns to the site, they will be able to access every
account page - except for the edit page - without being forced to re-authenticate. However,
when they try to access the editAction controller, they will be forced to re-authenticate, since
they are not, yet, fully authenticated.
For more information on securing services or methods in this way, see How to Secure any Service or
Method in your Application.

Chapter 102
How to Impersonate a User

Sometimes, it's useful to be able to switch from one user to another without having to log out and log
in again (for instance when you are debugging or trying to understand a bug a user sees that you can't
reproduce).
User impersonation is not compatible with pre authenticated firewalls. The reason is that
impersonation requires the authentication state to be maintained server-side, but pre-authenticated
information (SSL_CLIENT_S_DN_Email, REMOTE_USER or other) is sent in each request.
Impersonating the user can be easily done by activating the switch_user firewall listener:
Listing 102-1
1
2
3
4
5
6
7
8
security:
# ...
firewalls:
main:
# ...
switch_user: true
To switch to another user, just add a query string with the _switch_user parameter and the username
as the value to the current URL:
Listing 102-2
http://example.com/somewhere?_switch_user=thomas
To switch back to the original user, use the special _exit username:
Listing 102-3
http://example.com/somewhere?_switch_user=_exit
During impersonation, the user is provided with a special role called ROLE_PREVIOUS_ADMIN. In a
template, for instance, this role can be used to show a link to exit impersonation:
Listing 102-4
1
2
3
{% if is_granted('ROLE_PREVIOUS_ADMIN') %}
<a href="{{ path('homepage', {'_switch_user': '_exit'}) }}">Exit impersonation</a>
{% endif %}

Chapter 102: How to Impersonate a User | 355
In some cases you may need to get the object that represents the impersonating user rather than the
impersonated user. Use the following snippet to iterate over the user's roles until you find one that a
SwitchUserRole object:
Listing 102-5
1
2
3
4
5
6
7
8
9
10
11
12
13
use Symfony\Component\Security\Core\Role\SwitchUserRole;
$authChecker = $this->get('security.authorization_checker');
$tokenStorage = $this->get('security.token_storage');
if ($authChecker->isGranted('ROLE_PREVIOUS_ADMIN')) {
foreach ($tokenStorage->getToken()->getRoles() as $role) {
if ($role instanceof SwitchUserRole) {
$impersonatingUser = $role->getSource()->getUser();
break;
}
}
}
Of course, this feature needs to be made available to a small group of users. By default, access is restricted
to users having the ROLE_ALLOWED_TO_SWITCH role. The name of this role can be modified via the
role setting. For extra security, you can also change the query parameter name via the parameter
setting:
Listing 102-6
security:
# ...
1
2
3
4
5
6
7
8
firewalls:
main:
# ...
switch_user: { role: ROLE_ADMIN, parameter: _want_to_be_this_user }
Events
The firewall dispatches the security.switch_user event right after the impersonation is completed.
The SwitchUserEvent1 is passed to the listener, and you can use this to get the user that you are now
impersonating.
The cookbook article about Making the Locale "Sticky" during a User's Session does not update the locale
when you impersonate a user. The following code sample will show how to change the sticky locale:
Listing 102-7
services:
app.switch_user_listener:
class: AppBundle\EventListener\SwitchUserListener
tags:
- { name: kernel.event_listener, event: security.switch_user, method: onSwitchUser }
1
2
3
4
5
6
The listener implementation assumes your User entity has a getLocale() method.
Listing 102-8
1
2
3
4
// src/AppBundle/EventListener/SwitchUserListener.php
use Symfony\Component\Security\Http\Event\SwitchUserEvent;
1. http://api.symfony.com/2.7/Symfony/Component/Security/Http/Event/SwitchUserEvent.html

5
6
7
8
9
10
11
12
13
14
15
class SwitchUserListener
{
public function onSwitchUser(SwitchUserEvent $event)
{
$event->getRequest()->getSession()->set(
'_locale',
$event->getTargetUser()->getLocale()
);
}
}

Chapter 103
How to Customize your Form Login

Using a form login for authentication is a common, and flexible, method for handling authentication in
Symfony. Pretty much every aspect of the form login can be customized. The full, default configuration
is shown in the next section.
Form Login Configuration Reference

To see the full form login configuration reference, see SecurityBundle Configuration ("security"). Some of
the more interesting options are explained below.
Redirecting after Success

You can change where the login form redirects after a successful login using the various config options.
By default the form will redirect to the URL the user requested (i.e. the URL which triggered the
login form being shown). For example, if the user requested http://www.example.com/admin/
post/18/edit, then after they successfully log in, they will eventually be sent back to
http://www.example.com/admin/post/18/edit. This is done by storing the requested URL in
the session. If no URL is present in the session (perhaps the user went directly to the login page), then
the user is redirected to the default page, which is / (i.e. the homepage) by default. You can change this
behavior in several ways.
As mentioned, by default the user is redirected back to the page originally requested. Sometimes, this
can cause problems, like if a background Ajax request "appears" to be the last visited URL, causing
the user to be redirected there. For information on controlling this behavior, see How to Change the
default Target Path Behavior.
Changing the default Page

First, the default page can be set (i.e. the page the user is redirected to if no previous page was stored in
the session). To set it to the default_security_target route use the following config:

Chapter 103: How to Customize your Form Login | 358
Listing 103-1
1
2
3
4
5
6
7
8
9
security:
# ...
firewalls:
main:
form_login:
# ...
default_target_path: default_security_target
Now, when no URL is set in the session, users will be sent to the default_security_target route.
Always Redirect to the default Page

You can make it so that users are always redirected to the default page regardless of what URL they had
requested previously by setting the always_use_default_target_path option to true:
Listing 103-2
1
2
3
4
5
6
7
8
9
security:
# ...
firewalls:
main:
form_login:
# ...
always_use_default_target_path: true
Using the Referring URL

In case no previous URL was stored in the session, you may wish to try using the HTTP_REFERER
instead, as this will often be the same. You can do this by setting use_referer to true (it defaults to
false):
Listing 103-3
1
2
3
4
5
6
7
8
9
10
security:
# ...
firewalls:
main:
# ...
form_login:
# ...
use_referer: true
Control the Redirect URL from inside the Form

You can also override where the user is redirected to via the form itself by including a hidden field with
the name _target_path. For example, to redirect to the URL defined by some account route, use the
following:
Listing 103-4
1
2
3
4
5
6
7
8
9
10
{# src/AppBundle/Resources/views/Security/login.html.twig #}
{% if error %}
<div>{{ error.message }}</div>
{% endif %}

11
12
13
14
15
16

<input type="hidden" name="_target_path" value="account" />
<input type="submit" name="login" />
</form>
Now, the user will be redirected to the value of the hidden form field. The value attribute can be a relative
path, absolute URL, or a route name. You can even change the name of the hidden form field by changing
the target_path_parameter option to another value.
Listing 103-5
1
2
3
4
5
6
7
8
9
security:
# ...
firewalls:
main:
# ...
form_login:
target_path_parameter: redirect_url
Redirecting on Login Failure

In addition to redirecting the user after a successful login, you can also set the URL that the user should
be redirected to after a failed login (e.g. an invalid username or password was submitted). By default, the
user is redirected back to the login form itself. You can set this to a different route (e.g. login_failure)
with the following config:
Listing 103-6
1
2
3
4
5
6
7
8
9
10
security:
# ...
firewalls:
main:
# ...
form_login:
# ...
failure_path: login_failure

Chapter 104
How to Create a custom User Provider

Part of Symfony's standard authentication process depends on "user providers". When a user submits a
username and password, the authentication layer asks the configured user provider to return a user object
for a given username. Symfony then checks whether the password of this user is correct and generates a
security token so the user stays authenticated during the current session. Out of the box, Symfony has
an "in_memory" and an "entity" user provider. In this entry you'll see how you can create your own user
provider, which could be useful if your users are accessed via a custom database, a file, or - as shown in
this example - a web service.
Create a User Class

First, regardless of where your user data is coming from, you'll need to create a User class that
represents that data. The User can look however you want and contain any data. The only requirement
is that the class implements UserInterface1. The methods in this interface should therefore be
defined in the custom user class: getRoles()2, getPassword()3, getSalt()4, getUsername()5,
eraseCredentials()6. It may also be useful to implement the EquatableInterface7 interface,
which defines a method to check if the user is equal to the current user. This interface requires an
isEqualTo()8 method.
This is how your WebserviceUser class looks in action:
Listing 104-1
1
2
3
4
// src/AppBundle/Security/User/WebserviceUser.php
namespace AppBundle\Security\User;
2. http://api.symfony.com/2.7/Symfony/Component/Security/Core/User/UserInterface.html#method_getRoles
3. http://api.symfony.com/2.7/Symfony/Component/Security/Core/User/UserInterface.html#method_getPassword
4. http://api.symfony.com/2.7/Symfony/Component/Security/Core/User/UserInterface.html#method_getSalt
5. http://api.symfony.com/2.7/Symfony/Component/Security/Core/User/UserInterface.html#method_getUsername
6. http://api.symfony.com/2.7/Symfony/Component/Security/Core/User/UserInterface.html#method_eraseCredentials
7. http://api.symfony.com/2.7/Symfony/Component/Security/Core/User/EquatableInterface.html
8. http://api.symfony.com/2.7/Symfony/Component/Security/Core/User/EquatableInterface.html#method_isEqualTo

Chapter 104: How to Create a custom User Provider | 361
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
use Symfony\Component\Security\Core\User\EquatableInterface;
class WebserviceUser implements UserInterface, EquatableInterface
{
private $username;
private $password;
private $salt;
private $roles;
public function __construct($username, $password, $salt, array $roles)
{
$this->username = $username;
$this->password = $password;
$this->salt = $salt;
$this->roles = $roles;
}
public function getRoles()
{
return $this->roles;
}
public function getPassword()
{
return $this->password;
}
{
return $this->salt;
}
{
}
public function eraseCredentials()
{
}
public function isEqualTo(UserInterface $user)
{
if (!$user instanceof WebserviceUser) {
return false;
}
if ($this->password !== $user->getPassword()) {
return false;
}
if ($this->salt !== $user->getSalt()) {
return false;
}
if ($this->username !== $user->getUsername()) {
return false;
}
return true;
}
}
If you have more information about your users - like a "first name" - then you can add a firstName field
to hold that data.

Create a User Provider

Now that you have a User class, you'll create a user provider, which will grab user information from
some web service, create a WebserviceUser object, and populate it with data.
The user provider is just a plain PHP class that has to implement the UserProviderInterface9,
which requires three methods to be defined: loadUserByUsername($username),
refreshUser(UserInterface $user), and supportsClass($class). For more details, see
UserProviderInterface10.
Here's an example of how this might look:
Listing 104-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
// src/AppBundle/Security/User/WebserviceUserProvider.php
namespace AppBundle\Security\User;
use
use
use
use
Symfony\Component\Security\Core\Exception\UsernameNotFoundException;
class WebserviceUserProvider implements UserProviderInterface

{
{
// make a call to your webservice here
$userData = ...
// pretend it returns an array on success, false if there is no user
if ($userData) {
$password = '...';
// ...
return new WebserviceUser($username, $password, $salt, $roles);
}
throw new UsernameNotFoundException(
sprintf('Username "%s" does not exist.', $username)
);
}
{
if (!$user instanceof WebserviceUser) {
throw new UnsupportedUserException(
sprintf('Instances of "%s" are not supported.', get_class($user))
);
}
return $this->loadUserByUsername($user->getUsername());
}
{
return $class === 'AppBundle\Security\User\WebserviceUser';
}
}
Create a Service for the User Provider

Now you make the user provider available as a service:

Listing 104-3
1
2
3
4
services:
app.webservice_user_provider:
class: AppBundle\Security\User\WebserviceUserProvider
The real implementation of the user provider will probably have some dependencies or configuration
options or other services. Add these as arguments in the service definition.
Make sure the services file is being imported. See Importing Configuration with imports for details.
Modify security.yml
Everything comes together in your security configuration. Add the user provider to the list of providers
in the "security" section. Choose a name for the user provider (e.g. "webservice") and mention the id of
the service you just defined.
Listing 104-4
1
2
3
4
5
6
7
security:
# ...
providers:
webservice:
id: app.webservice_user_provider
Symfony also needs to know how to encode passwords that are supplied by website users, e.g. by filling
in a login form. You can do this by adding a line to the "encoders" section in your security configuration:
Listing 104-5
1
2
3
4
5
6
security:
# ...
encoders:
AppBundle\Security\User\WebserviceUser: bcrypt
The value here should correspond with however the passwords were originally encoded when creating
your users (however those users were created). When a user submits their password, it's encoded using
this algorithm and the result is compared to the hashed password returned by your getPassword()
method.

Specifics on how Passwords are Encoded

Symfony uses a specific method to combine the salt and encode the password before comparing it
to your encoded password. If getSalt() returns nothing, then the submitted password is simply
encoded using the algorithm you specify in security.yml. If a salt is specified, then the following
value is created and then hashed via the algorithm:
Listing 104-6
$password.'{'.$salt.'}'
If your external users have their passwords salted via a different method, then you'll need to do
a bit more work so that Symfony properly encodes the password. That is beyond the scope of
this entry, but would include sub-classing MessageDigestPasswordEncoder and overriding the
mergePasswordAndSalt method.
Additionally, you can configure the details of the algorithm used to hash passwords. In this example,
the application sets explicitly the cost of the bcrypt hashing:
Listing 104-7
1
2
3
4
5
6
7
8
security:
# ...
encoders:
AppBundle\Security\User\WebserviceUser:
algorithm: bcrypt
cost: 12

Chapter 105
How to Create a Custom Form Password

Authenticator
Imagine you want to allow access to your website only between 2pm and 4pm UTC. Before Symfony 2.4,
you had to create a custom token, factory, listener and provider. In this entry, you'll learn how to do this
for a login form (i.e. where your user submits their username and password). Before Symfony 2.6, you
had to use the password encoder to authenticate the user password.
The Password Authenticator

New in version 2.6: The UserPasswordEncoderInterface interface was introduced in Symfony 2.6.
First, create a new class that implements SimpleFormAuthenticatorInterface1. Eventually, this
will allow you to create custom logic for authenticating the user:
Listing 105-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
// src/Acme/HelloBundle/Security/TimeAuthenticator.php
namespace Acme\HelloBundle\Security;
use
use
use
use
use
use
use
use
Symfony\Component\Security\Core\Authentication\SimpleFormAuthenticatorInterface;
Symfony\Component\Security\Core\Authentication\Token\TokenInterface;
Symfony\Component\Security\Core\Authentication\Token\UsernamePasswordToken;
Symfony\Component\Security\Core\Encoder\UserPasswordEncoderInterface;
Symfony\Component\Security\Core\Exception\AuthenticationException;
Symfony\Component\Security\Core\Exception\UsernameNotFoundException;
class TimeAuthenticator implements SimpleFormAuthenticatorInterface

{
private $encoder;
public function __construct(UserPasswordEncoderInterface $encoder)
{
$this->encoder = $encoder;
}
1. http://api.symfony.com/2.7/Symfony/Component/Security/Core/Authentication/SimpleFormAuthenticatorInterface.html

Chapter 105: How to Create a Custom Form Password Authenticator | 366
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
public function authenticateToken(TokenInterface $token, UserProviderInterface $userProvider,

$providerKey)
{
try {
$user = $userProvider->loadUserByUsername($token->getUsername());
} catch (UsernameNotFoundException $e) {
throw new AuthenticationException('Invalid username or password');
}
$passwordValid = $this->encoder->isPasswordValid($user, $token->getCredentials());
if ($passwordValid) {
$currentHour = date('G');
if ($currentHour < 14 || $currentHour > 16) {
throw new AuthenticationException(
'You can only log in between 2 and 4!',
100
);
}
return new UsernamePasswordToken(
$user,
$user->getPassword(),
$providerKey,
$user->getRoles()
);
}
throw new AuthenticationException('Invalid username or password');
}
public function supportsToken(TokenInterface $token, $providerKey)
{
return $token instanceof UsernamePasswordToken
&& $token->getProviderKey() === $providerKey;
}
public function createToken(Request $request, $username, $password, $providerKey)
{
return new UsernamePasswordToken($username, $password, $providerKey);
}
}
How it Works
Great! Now you just need to setup some Configuration. But first, you can find out more about what each
method in this class does.
1) createToken
When Symfony begins handling a request, createToken() is called, where you create a
TokenInterface2 object that contains whatever information you need in authenticateToken() to
authenticate the user (e.g. the username and password).
Whatever token object you create here will be passed to you later in authenticateToken().
2) supportsToken
After Symfony calls createToken(), it will then call supportsToken() on your class (and any other
authentication listeners) to figure out who should handle the token. This is just a way to allow several
2. http://api.symfony.com/2.7/Symfony/Component/Security/Core/Authentication/Token/TokenInterface.html

authentication mechanisms to be used for the same firewall (that way, you can for instance first try to
authenticate the user via a certificate or an API key and fall back to a form login).
Mostly, you just need to make sure that this method returns true for a token that has been created by
createToken(). Your logic should probably look exactly like this example.
3) authenticateToken
If supportsToken returns true, Symfony will now call authenticateToken(). Your job here is to
check that the token is allowed to log in by first getting the User object via the user provider and then,
by checking the password and the current time.
The "flow" of how you get the User object and determine whether or not the token is valid (e.g.
checking the password), may vary based on your requirements.
Ultimately, your job is to return a new token object that is "authenticated" (i.e. it has at least 1 role set on
it) and which has the User object inside of it.
Inside this method, the password encoder is needed to check the password's validity:
Listing 105-2
$passwordValid = $this->encoder->isPasswordValid($user, $token->getCredentials());
This is a service that is already available in Symfony and it uses the password algorithm that is configured
in the security configuration (e.g. security.yml) under the encoders key. Below, you'll see how to
inject that into the TimeAuthenticator.
Configuration
Now, configure your TimeAuthenticator as a service:
Listing 105-3
1
2
3
4
5
6
7
services:
# ...
time_authenticator:
class:
Acme\HelloBundle\Security\TimeAuthenticator
arguments: ["@security.password_encoder"]
Then, activate it in the firewalls section of the security configuration using the simple_form key:
Listing 105-4
1
2
3
4
5
6
7
8
9
10
11
12
security:
# ...
firewalls:
secured_area:
pattern: ^/admin
# ...
simple_form:
authenticator: time_authenticator
check_path:
login_check
login_path:
login
The simple_form key has the same options as the normal form_login option, but with the additional
authenticator key that points to the new service. For details, see Form Login Configuration.
If creating a login form in general is new to you or you don't understand the check_path or
login_path options, see How to Customize your Form Login.
Chapter 106
How to Authenticate Users with API Keys

Nowadays, it's quite usual to authenticate the user via an API key (when developing a web service for
instance). The API key is provided for every request and is passed as a query string parameter or via an
HTTP header.
The API Key Authenticator

Authenticating a user based on the Request information should be done via a pre-authentication
mechanism. The SimplePreAuthenticatorInterface1 allows you to implement such a scheme
really easily.
Your exact situation may differ, but in this example, a token is read from an apikey query parameter,
the proper username is loaded from that value and then a User object is created:
Listing 106-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
// src/AppBundle/Security/ApiKeyAuthenticator.php
namespace AppBundle\Security;
use
use
use
use
use
use
use
Symfony\Component\Security\Core\Authentication\SimplePreAuthenticatorInterface;
Symfony\Component\Security\Core\Authentication\Token\PreAuthenticatedToken;
Symfony\Component\Security\Core\Exception\BadCredentialsException;
class ApiKeyAuthenticator implements SimplePreAuthenticatorInterface

{
public function createToken(Request $request, $providerKey)
{
// look for an apikey query parameter
$apiKey = $request->query->get('apikey');
// or if you want to use an "apikey" header, then do something like this:

// $apiKey = $request->headers->get('apikey');
if (!$apiKey) {
throw new BadCredentialsException();
1. http://api.symfony.com/2.7/Symfony/Component/Security/Core/Authentication/SimplePreAuthenticatorInterface.html

Chapter 106: How to Authenticate Users with API Keys | 369
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
// or to just skip api key authentication

// return null;
}
return new PreAuthenticatedToken(
'anon.',
$apiKey,
$providerKey
);
}
public function supportsToken(TokenInterface $token, $providerKey)
{
return $token instanceof PreAuthenticatedToken && $token->getProviderKey() === $providerKey;
}
$providerKey)
{
if (!$userProvider instanceof ApiKeyUserProvider) {
throw new \InvalidArgumentException(
sprintf(
'The user provider must be an instance of ApiKeyUserProvider (%s was given).',
get_class($userProvider)
)
);
}
$apiKey = $token->getCredentials();
$username = $userProvider->getUsernameForApiKey($apiKey);
if (!$username) {
sprintf('API Key "%s" does not exist.', $apiKey)
);
}
$user = $userProvider->loadUserByUsername($username);
$user,
$apiKey,
$providerKey,
$user->getRoles()
);
}
}
Once you've configured everything, you'll be able to authenticate by adding an apikey parameter to
the
query
string,
like
http://example.com/api/
foo?apikey=37b51d194a7513e45b56f6524f2d51f2.
The authentication process has several steps, and your implementation will probably differ:
1. createToken
Early in the request cycle, Symfony calls createToken(). Your job here is to create a token object that
contains all of the information from the request that you need to authenticate the user (e.g. the apikey
query parameter). If that information is missing, throwing a BadCredentialsException2 will cause
authentication to fail. You might want to return null instead to just skip the authentication, so Symfony
can fallback to another authentication method, if any.
2. http://api.symfony.com/2.7/Symfony/Component/Security/Core/Exception/BadCredentialsException.html

In case you return null from your createToken() method, be sure to enable anonymous in you
firewall. This way you'll be able to get an AnonymousToken.
2. supportsToken
After Symfony calls createToken(), it will then call supportsToken() on your class (and any other
authentication listeners) to figure out who should handle the token. This is just a way to allow several
authentication mechanisms to be used for the same firewall (that way, you can for instance first try to
authenticate the user via a certificate or an API key and fall back to a form login).
Mostly, you just need to make sure that this method returns true for a token that has been created by
createToken(). Your logic should probably look exactly like this example.
3. authenticateToken
If supportsToken() returns true, Symfony will now call authenticateToken(). One key part is
the $userProvider, which is an external class that helps you load information about the user. You'll
learn more about this next.
In this specific example, the following things happen in authenticateToken():
1. First, you use the $userProvider to somehow look up the $username that corresponds to the $apiKey;
2. Second, you use the $userProvider again to load or create a User object for the $username;
3. Finally, you create an authenticated token (i.e. a token with at least one role) that has the proper
roles and the User object attached to it.
The goal is ultimately to use the $apiKey to find or create a User object. How you do this (e.g. query a
database) and the exact class for your User object may vary. Those differences will be most obvious in
your user provider.
The User Provider

The $userProvider can be any user provider (see How to Create a custom User Provider). In this
example, the $apiKey is used to somehow find the username for the user. This work is done in a
getUsernameForApiKey() method, which is created entirely custom for this use-case (i.e. this isn't a
method that's used by Symfony's core user provider system).
The $userProvider might look something like this:
Listing 106-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
// src/AppBundle/Security/ApiKeyUserProvider.php
use
use
use
use
Symfony\Component\Security\Core\User\User;
class ApiKeyUserProvider implements UserProviderInterface

{
public function getUsernameForApiKey($apiKey)
{
// Look up the username based on the token in the database, via
// an API call, or do something entirely different
$username = ...;
return $username;
}

21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
{
return new User(
$username,
null,
// the roles for the user - you may choose to determine
// these dynamically somehow based on the user
array('ROLE_API')
);
}
{
// this is used for storing authentication in the session
// but in this example, the token is sent in each request,
// so authentication can be stateless. Throwing this exception
// is proper to make things stateless
throw new UnsupportedUserException();
}
{
return 'Symfony\Component\Security\Core\User\User' === $class;
}
}
Now register your user provider as a service:

Listing 106-3
1
2
3
4
services:
api_key_user_provider:
class: AppBundle\Security\ApiKeyUserProvider
Read the dedicated article to learn how to create a custom user provider.
The logic inside getUsernameForApiKey() is up to you. You may somehow transform the API key
(e.g. 37b51d) into a username (e.g. jondoe) by looking up some information in a "token" database
table.
The same is true for loadUserByUsername(). In this example, Symfony's core User3 class is simply
created. This makes sense if you don't need to store any extra information on your User object (e.g.
firstName). But if you do, you may instead have your own user class which you create and populate
here by querying a database. This would allow you to have custom data on the User object.
Finally, just make sure that supportsClass() returns true for User objects with the same class as
whatever user you return in loadUserByUsername().
If your authentication is stateless like in this example (i.e. you expect the user to send the API key
with every request and so you don't save the login to the session), then you can simply throw the
UnsupportedUserException exception in refreshUser().
If you do want to store authentication data in the session so that the key doesn't need to be sent on
every request, see Storing Authentication in the Session.
3. http://api.symfony.com/2.7/Symfony/Component/Security/Core/User/User.html

Handling Authentication Failure

In order for your ApiKeyAuthenticator to correctly display a 401 http status when either bad
credentials
or
authentication
fails
you
will
need
to
implement
the
4
AuthenticationFailureHandlerInterface on your Authenticator. This will provide a method
onAuthenticationFailure which you can use to create an error Response.
Listing 106-4
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
use
use
use
use
use
Symfony\Component\Security\Core\Authentication\SimplePreAuthenticatorInterface;
Symfony\Component\Security\Http\Authentication\AuthenticationFailureHandlerInterface;
class ApiKeyAuthenticator implements SimplePreAuthenticatorInterface, AuthenticationFailureHandlerInterface

{
// ...
public function onAuthenticationFailure(Request $request, AuthenticationException $exception)
{
return new Response("Authentication Failed.", 401);
}
}
Configuration
Once you have your ApiKeyAuthenticator all setup, you need to register it as a service and use it in
your security configuration (e.g. security.yml). First, register it as a service.
Listing 106-5
1
2
3
4
5
6
7
services:
# ...
apikey_authenticator:
class: AppBundle\Security\ApiKeyAuthenticator
public: false
Now, activate it and your custom user provider (see How to Create a custom User Provider) in the
firewalls section of your security configuration using the simple_preauth and provider keys
respectively:
Listing 106-6
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
security:
# ...
firewalls:
secured_area:
pattern: ^/api
stateless: true
simple_preauth:
authenticator: apikey_authenticator
provider: api_key_user_provider
providers:
id: api_key_user_provider
4. http://api.symfony.com/2.7/Symfony/Component/Security/Http/Authentication/AuthenticationFailureHandlerInterface.html

If you have defined access_control, make sure to add a new entry:

Listing 106-7
1
2
3
4
5
6
security:
# ...
access_control:
- { path: ^/api, roles: ROLE_API }
That's it! Now, your ApiKeyAuthenticator should be called at the beginning of each request and
your authentication process will take place.
The stateless configuration parameter prevents Symfony from trying to store the authentication
information in the session, which isn't necessary since the client will send the apikey on each request. If
you do need to store authentication in the session, keep reading!
Storing Authentication in the Session

So far, this entry has described a situation where some sort of authentication token is sent on every
request. But in some situations (like an OAuth flow), the token may be sent on only one request. In this
case, you will want to authenticate the user and store that authentication in the session so that the user
is automatically logged in for every subsequent request.
To make this work, first remove the stateless key from your firewall configuration or set it to false:
Listing 106-8
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
security:
# ...
firewalls:
secured_area:
pattern: ^/api
stateless: false
simple_preauth:
authenticator: apikey_authenticator
provider: api_key_user_provider
providers:
id: api_key_user_provider
Even though the token is being stored in the session, the credentials - in this case the API key (i.e.
$token->getCredentials()) - are not stored in the session for security reasons. To take advantage
of the session, update ApiKeyAuthenticator to see if the stored token has a valid User object that
can be used:
Listing 106-9
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
// ...
{
// ...
$providerKey)
{
if (!$userProvider instanceof ApiKeyUserProvider) {
throw new \InvalidArgumentException(
sprintf(
'The user provider must be an instance of ApiKeyUserProvider (%s was given).',
get_class($userProvider)
)
);
}

18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
$apiKey = $token->getCredentials();
$username = $userProvider->getUsernameForApiKey($apiKey);
// User is the Entity which represents your user

$user = $token->getUser();
if ($user instanceof User) {
$user,
$apiKey,
$providerKey,
$user->getRoles()
);
}
if (!$username) {
sprintf('API Key "%s" does not exist.', $apiKey)
);
}
$user = $userProvider->loadUserByUsername($username);
$user,
$apiKey,
$providerKey,
$user->getRoles()
);
}
// ...
}
Storing authentication information in the session works like this:

1. At the end of each request, Symfony serializes the token object (returned from
authenticateToken()), which also serializes the User object (since it's set on a property on the token);
2. On the next request the token is deserialized and the deserialized User object is passed to the
refreshUser() function of the user provider.
The second step is the important one: Symfony calls refreshUser() and passes you the user object
that was serialized in the session. If your users are stored in the database, then you may want to re-query
for a fresh version of the user to make sure it's not out-of-date. But regardless of your requirements,
refreshUser() should now return the User object:
Listing 106-10
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
// src/AppBundle/Security/ApiKeyUserProvider.php
// ...
class ApiKeyUserProvider implements UserProviderInterface
{
// ...
{
// $user is the User that you set in the token inside authenticateToken()
// after it has been deserialized from the session
// you might use $user to query the database for a fresh user
// $id = $user->getId();
// use $id to make a query
// if you are *not* reading from a database and are just creating
// a User object (like in this example), you can just return it
return $user;
}
}

You'll also want to make sure that your User object is being serialized correctly. If your User object
has private properties, PHP can't serialize those. In this case, you may get back a User object that has
a null value for each property. For an example, see How to Load Security Users from the Database
(the Entity Provider).
Only Authenticating for Certain URLs

This entry has assumed that you want to look for the apikey authentication on every request. But in
some situations (like an OAuth flow), you only really need to look for authentication information once
the user has reached a certain URL (e.g. the redirect URL in OAuth).
Fortunately, handling this situation is easy: just check to see what the current URL is before creating the
token in createToken():
Listing 106-11
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
// ...
use Symfony\Component\Security\Http\HttpUtils;
{
protected $httpUtils;
public function __construct(HttpUtils $httpUtils)
{
$this->httpUtils = $httpUtils;
}
public function createToken(Request $request, $providerKey)
{
// set the only URL where we should look for auth information
// and only return the token if we're at that URL
$targetUrl = '/login/check';
if (!$this->httpUtils->checkRequestPath($request, $targetUrl)) {
return;
}
// ...
}
}
This uses the handy HttpUtils5 class to check if the current URL matches the URL you're looking for.
In this case, the URL (/login/check) has been hardcoded in the class, but you could also inject it as
the second constructor argument.
Next, just update your service configuration to inject the security.http_utils service:
Listing 106-12 1
2
3
4
5
6
7
8
services:
# ...
apikey_authenticator:
class:
AppBundle\Security\ApiKeyAuthenticator
arguments: ["@security.http_utils"]
public:
false
That's it! Have fun!
5. http://api.symfony.com/2.7/Symfony/Component/Security/Http/HttpUtils.html

Chapter 107
How to Create a custom Authentication

Provider
Creating a custom authentication system is hard, and this entry will walk you through that process.
But depending on your needs, you may be able to solve your problem in a simpler, or via a
community bundle:
How to Create a Custom Form Password Authenticator
How to Authenticate Users with API Keys
To authenticate via OAuth using a third-party service such as Google, Facebook or Twitter, try
using the HWIOAuthBundle1 community bundle.
If you have read the chapter on Security, you understand the distinction Symfony makes between
authentication and authorization in the implementation of security. This chapter discusses the core
classes involved in the authentication process, and how to implement a custom authentication provider.
Because authentication and authorization are separate concepts, this extension will be user-provider
agnostic, and will function with your application's user providers, may they be based in memory, a
database, or wherever else you choose to store them.
Meet WSSE
The following chapter demonstrates how to create a custom authentication provider for WSSE
authentication. The security protocol for WSSE provides several security benefits:
1. Username / Password encryption
2. Safe guarding against replay attacks
3. No web server configuration required
WSSE is very useful for the securing of web services, may they be SOAP or REST.
1. https://github.com/hwi/HWIOAuthBundle

Chapter 107: How to Create a custom Authentication Provider | 377
There is plenty of great documentation on WSSE2, but this article will focus not on the security protocol,
but rather the manner in which a custom protocol can be added to your Symfony application. The basis
of WSSE is that a request header is checked for encrypted credentials, verified using a timestamp and
nonce3, and authenticated for the requested user using a password digest.
WSSE also supports application key validation, which is useful for web services, but is outside the
scope of this chapter.
The Token
The role of the token in the Symfony security context is an important one. A token represents the user
authentication data present in the request. Once a request is authenticated, the token retains the user's
data, and delivers this data across the security context. First, you'll create your token class. This will allow
the passing of all relevant information to your authentication provider.
Listing 107-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
// src/AppBundle/Security/Authentication/Token/WsseUserToken.php
namespace AppBundle\Security\Authentication\Token;
use Symfony\Component\Security\Core\Authentication\Token\AbstractToken;
class WsseUserToken extends AbstractToken
{
public $created;
public $digest;
public $nonce;
public function __construct(array $roles = array())
{
parent::__construct($roles);
// If the user has roles, consider it authenticated

$this->setAuthenticated(count($roles) > 0);
}
public function getCredentials()
{
return '';
}
}
The WsseUserToken class extends the Security component's AbstractToken4 class, which
provides basic token functionality. Implement the TokenInterface5 on any class to use as a token.
The Listener
Next, you need a listener to listen on the firewall. The listener is responsible for fielding requests
to the firewall and calling the authentication provider. A listener must be an instance of
2. http://www.xml.com/pub/a/2003/12/17/dive.html
3. https://en.wikipedia.org/wiki/Cryptographic_nonce
4. http://api.symfony.com/2.7/Symfony/Component/Security/Core/Authentication/Token/AbstractToken.html
5. http://api.symfony.com/2.7/Symfony/Component/Security/Core/Authentication/Token/TokenInterface.html

ListenerInterface6. A security listener should handle the GetResponseEvent7 event, and set an
authenticated token in the token storage if successful.
Listing 107-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
// src/AppBundle/Security/Firewall/WsseListener.php
namespace AppBundle\Security\Firewall;
use
use
use
use
use
use
use
Symfony\Component\HttpKernel\Event\GetResponseEvent;
Symfony\Component\Security\Core\Authentication\AuthenticationManagerInterface;
Symfony\Component\Security\Core\Authentication\Token\Storage\TokenStorageInterface;
Symfony\Component\Security\Http\Firewall\ListenerInterface;
AppBundle\Security\Authentication\Token\WsseUserToken;
class WsseListener implements ListenerInterface

{
protected $tokenStorage;
protected $authenticationManager;
public function __construct(TokenStorageInterface $tokenStorage, AuthenticationManagerInterface
$authenticationManager)
{
$this->authenticationManager = $authenticationManager;
}
public function handle(GetResponseEvent $event)
{
$request = $event->getRequest();
$wsseRegex = '/UsernameToken Username="([^"]+)", PasswordDigest="([^"]+)", Nonce="([a-zA-Z0-9+\/
]+={0,2})", Created="([^"]+)"/';
if (!$request->headers->has('x-wsse') || 1 !== preg_match($wsseRegex,
$request->headers->get('x-wsse'), $matches)) {
return;
}
$token = new WsseUserToken();
$token->setUser($matches[1]);
$token->digest
$token->nonce
$token->created
= $matches[2];
= $matches[3];
= $matches[4];
try {
$authToken = $this->authenticationManager->authenticate($token);
$this->tokenStorage->setToken($authToken);
return;
} catch (AuthenticationException $failed) {
// ... you might log something here
//
//
//
//
//
//
//
To deny the authentication clear the token. This will redirect to the login page.
Make sure to only clear your token, not those of other authentication listeners.
$token = $this->tokenStorage->getToken();
if ($token instanceof WsseUserToken && $this->providerKey === $token->getProviderKey()) {
$this->tokenStorage->setToken(null);
}
return;
// By default deny authorization

$response->setStatusCode(Response::HTTP_FORBIDDEN);
$event->setResponse($response);
}
}
6. http://api.symfony.com/2.7/Symfony/Component/Security/Http/Firewall/ListenerInterface.html
7. http://api.symfony.com/2.7/Symfony/Component/HttpKernel/Event/GetResponseEvent.html

This listener checks the request for the expected X-WSSE header, matches the value returned for the
expected WSSE information, creates a token using that information, and passes the token on to the
authentication manager. If the proper information is not provided, or the authentication manager throws
an AuthenticationException8, a 403 Response is returned.
A class not used above, the AbstractAuthenticationListener9 class, is a very useful base
class which provides commonly needed functionality for security extensions. This includes
maintaining the token in the session, providing success / failure handlers, login form URLs, and
more. As WSSE does not require maintaining authentication sessions or login forms, it won't be used
for this example.
Returning prematurely from the listener is relevant only if you want to chain authentication
providers (for example to allow anonymous users). If you want to forbid access to anonymous users
and have a nice 403 error, you should set the status code of the response before returning.
The Authentication Provider

The authentication provider will do the verification of the WsseUserToken. Namely, the provider will
verify the Created header value is valid within five minutes, the Nonce header value is unique within
five minutes, and the PasswordDigest header value matches with the user's password.
Listing 107-3
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
// src/AppBundle/Security/Authentication/Provider/WsseProvider.php
namespace AppBundle\Security\Authentication\Provider;
use
use
use
use
use
use
use
Symfony\Component\Security\Core\Authentication\Provider\AuthenticationProviderInterface;
Symfony\Component\Security\Core\Exception\NonceExpiredException;
AppBundle\Security\Authentication\Token\WsseUserToken;
Symfony\Component\Security\Core\Util\StringUtils;
class WsseProvider implements AuthenticationProviderInterface

{
private $userProvider;
private $cacheDir;
public function __construct(UserProviderInterface $userProvider, $cacheDir)
{
$this->userProvider = $userProvider;
$this->cacheDir
= $cacheDir;
}
public function authenticate(TokenInterface $token)
{
$user = $this->userProvider->loadUserByUsername($token->getUsername());
if ($user && $this->validateDigest($token->digest, $token->nonce, $token->created,
$user->getPassword())) {
$authenticatedToken = new WsseUserToken($user->getRoles());
$authenticatedToken->setUser($user);
return $authenticatedToken;
}
throw new AuthenticationException('The WSSE authentication failed.');
}
8. http://api.symfony.com/2.7/Symfony/Component/Security/Core/Exception/AuthenticationException.html
9. http://api.symfony.com/2.7/Symfony/Component/Security/Http/Firewall/AbstractAuthenticationListener.html

37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
/**
* This function is specific to Wsse authentication and is only used to help this example
*
* For more information specific to the logic here, see
* https://github.com/symfony/symfony-docs/pull/3134#issuecomment-27699129
*/
protected function validateDigest($digest, $nonce, $created, $secret)
{
// Check created time is not in the future
if (strtotime($created) > time()) {
return false;
}
// Expire timestamp after 5 minutes
if (time() - strtotime($created) > 300) {
return false;
}
// Validate that the nonce is *not* used in the last 5 minutes
// if it has, this could be a replay attack
if (
file_exists($this->cacheDir.'/'.md5($nonce))
&& file_get_contents($this->cacheDir.'/'.md5($nonce)) + 300 > time()
) {
throw new NonceExpiredException('Previously used nonce detected');
}
// If cache directory does not exist we create it
if (!is_dir($this->cacheDir)) {
mkdir($this->cacheDir, 0777, true);
}
file_put_contents($this->cacheDir.'/'.md5($nonce), time());
// Validate Secret
$expected = base64_encode(sha1(base64_decode($nonce).$created.$secret, true));
return StringUtils::equals($expected, $digest);
}
public function supports(TokenInterface $token)
{
return $token instanceof WsseUserToken;
}
}
The AuthenticationProviderInterface10 requires an authenticate method on the user

token, and a supports method, which tells the authentication manager whether or not to use this
provider for the given token. In the case of multiple providers, the authentication manager will then
move to the next provider in the list.
The comparison of the expected and the provided digests uses a constant time comparison provided
by the equals()11 method of the StringUtils class. It is used to mitigate possible timing
attacks12.
The Factory
You have created a custom token, custom listener, and custom provider. Now you need to tie them all
together. How do you make a unique provider available for every firewall? The answer is by using a
10. http://api.symfony.com/2.7/Symfony/Component/Security/Core/Authentication/Provider/AuthenticationProviderInterface.html
11. http://api.symfony.com/2.7/Symfony/Component/Security/Core/Util/StringUtils.html#method_equals
12. https://en.wikipedia.org/wiki/Timing_attack

factory. A factory is where you hook into the Security component, telling it the name of your provider
and any configuration options available for it. First, you must create a class which implements
SecurityFactoryInterface13.
Listing 107-4
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
// src/AppBundle/DependencyInjection/Security/Factory/WsseFactory.php
namespace AppBundle\DependencyInjection\Security\Factory;
use
use
use
use
use
Symfony\Component\DependencyInjection\ContainerBuilder;
Symfony\Component\DependencyInjection\Reference;
Symfony\Component\DependencyInjection\DefinitionDecorator;
Symfony\Component\Config\Definition\Builder\NodeDefinition;
Symfony\Bundle\SecurityBundle\DependencyInjection\Security\Factory\SecurityFactoryInterface;
class WsseFactory implements SecurityFactoryInterface

{
public function create(ContainerBuilder $container, $id, $config, $userProvider, $defaultEntryPoint)
{
$providerId = 'security.authentication.provider.wsse.'.$id;
$container
->setDefinition($providerId, new DefinitionDecorator('wsse.security.authentication.provider'))
->replaceArgument(0, new Reference($userProvider))
;
$listenerId = 'security.authentication.listener.wsse.'.$id;
$listener = $container->setDefinition($listenerId, new
DefinitionDecorator('wsse.security.authentication.listener'));
return array($providerId, $listenerId, $defaultEntryPoint);
}
public function getPosition()
{
return 'pre_auth';
}
public function getKey()
{
return 'wsse';
}
public function addConfiguration(NodeDefinition $node)
{
}
}
The SecurityFactoryInterface14 requires the following methods:

create
Method which adds the listener and authentication provider to the DI container for the appropriate
security context.
getPosition
Returns when the provider should be called. This can be one of pre_auth, form, http or remember_me.
getKey
Method which defines the configuration key used to reference the provider in the firewall
configuration.
addConfiguration
Method which is used to define the configuration options underneath the configuration key in your
security configuration. Setting configuration options are explained later in this chapter.
13. http://api.symfony.com/2.7/Symfony/Bundle/SecurityBundle/DependencyInjection/Security/Factory/SecurityFactoryInterface.html
14. http://api.symfony.com/2.7/Symfony/Bundle/SecurityBundle/DependencyInjection/Security/Factory/SecurityFactoryInterface.html

A class not used in this example, AbstractFactory15, is a very useful base class which provides
commonly needed functionality for security factories. It may be useful when defining an
authentication provider of a different type.
Now that you have created a factory class, the wsse key can be used as a firewall in your security
configuration.
You may be wondering "why do you need a special factory class to add listeners and providers to
the dependency injection container?". This is a very good question. The reason is you can use your
firewall multiple times, to secure multiple parts of your application. Because of this, each time your
firewall is used, a new service is created in the DI container. The factory is what creates these new
services.
Configuration
It's time to see your authentication provider in action. You will need to do a few things in order to
make this work. The first thing is to add the services above to the DI container. Your factory class above
makes reference to service ids that do not exist yet: wsse.security.authentication.provider
and wsse.security.authentication.listener. It's time to define those services.
Listing 107-5
1
2
3
4
5
6
7
8
9
10
11
12
13
services:
wsse.security.authentication.provider:
class: AppBundle\Security\Authentication\Provider\WsseProvider
arguments:
- '' # User Provider
- '%kernel.cache_dir%/security/nonces'
public: false
wsse.security.authentication.listener:
class: AppBundle\Security\Firewall\WsseListener
arguments: ['@security.token_storage', '@security.authentication.manager']
public: false
Now that your services are defined, tell your security context about your factory in your bundle class:
Listing 107-6
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
// src/AppBundle/AppBundle.php
use AppBundle\DependencyInjection\Security\Factory\WsseFactory;
class AppBundle extends Bundle
{
{
$extension = $container->getExtension('security');
$extension->addSecurityListenerFactory(new WsseFactory());
}
}
You are finished! You can now define parts of your app as under WSSE protection.
Listing 107-7
15. http://api.symfony.com/2.7/Symfony/Bundle/SecurityBundle/DependencyInjection/Security/Factory/AbstractFactory.html

1
2
3
4
5
6
7
8
9
security:
# ...
firewalls:
wsse_secured:
pattern: ^/api/
stateless: true
wsse:
true
Congratulations! You have written your very own custom security authentication provider!
A little Extra
How about making your WSSE authentication provider a bit more exciting? The possibilities are endless.
Why don't you start by adding some sparkle to that shine?
Configuration
You can add custom options under the wsse key in your security configuration. For instance, the time
allowed before expiring the Created header item, by default, is 5 minutes. Make this configurable, so
different firewalls can have different timeout lengths.
You will first need to edit WsseFactory and define the new option in the addConfiguration
method.
Listing 107-8
1
2
3
4
5
6
7
8
9
10
11
12

{
// ...
public function addConfiguration(NodeDefinition $node)
{
$node
->children()
->scalarNode('lifetime')->defaultValue(300)
->end();
}
}
Now, in the create method of the factory, the $config argument will contain a lifetime key,
set to 5 minutes (300 seconds) unless otherwise set in the configuration. Pass this argument to your
authentication provider in order to put it to use.
Listing 107-9
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15

{
public function create(ContainerBuilder $container, $id, $config, $userProvider, $defaultEntryPoint)
{
$providerId = 'security.authentication.provider.wsse.'.$id;
$container
->setDefinition($providerId,
new DefinitionDecorator('wsse.security.authentication.provider'))
->replaceArgument(0, new Reference($userProvider))
->replaceArgument(2, $config['lifetime']);
// ...
}
// ...
}

You'll also need to add a third argument to the wsse.security.authentication.provider

service configuration, which can be blank, but will be filled in with the lifetime in the factory. The
WsseProvider class will also now need to accept a third constructor argument - the lifetime which it should use instead of the hard-coded 300 seconds. These two steps are not shown here.
The lifetime of each WSSE request is now configurable, and can be set to any desirable value per firewall.
Listing 107-10 1
2
3
4
5
6
7
8
9
security:
# ...
firewalls:
wsse_secured:
pattern:
^/api/
stateless: true
wsse:
{ lifetime: 30 }
The rest is up to you! Any relevant configuration items can be defined in the factory and consumed or
passed to the other classes in the container.

Chapter 108
Using pre Authenticated Security Firewalls

A lot of authentication modules are already provided by some web servers, including Apache. These
modules generally set some environment variables that can be used to determine which user is accessing
your application. Out of the box, Symfony supports most authentication mechanisms. These requests
are called pre authenticated requests because the user is already authenticated when reaching your
application.
User impersonation is not compatible with pre-authenticated firewalls. The reason is that
impersonation requires the authentication state to be maintained server-side, but pre-authenticated
information (SSL_CLIENT_S_DN_Email, REMOTE_USER or other) is sent in each request.
X.509 Client Certificate Authentication

When using client certificates, your webserver is doing all the authentication process itself. With Apache,
for example, you would use the SSLVerifyClient Require directive.
Enable the x509 authentication for a particular firewall in the security configuration:
Listing 108-1
1
2
3
4
5
6
7
8
9
security:
# ...
firewalls:
secured_area:
pattern: ^/
x509:
provider: your_user_provider
By default, the firewall provides the SSL_CLIENT_S_DN_Email variable to the user provider, and sets
the SSL_CLIENT_S_DN as credentials in the PreAuthenticatedToken1. You can override these by
setting the user and the credentials keys in the x509 firewall configuration respectively.
1. http://api.symfony.com/2.7/Symfony/Component/Security/Core/Authentication/Token/PreAuthenticatedToken.html

Chapter 108: Using pre Authenticated Security Firewalls | 386
An authentication provider will only inform the user provider of the username that made the request.
You will need to create (or use) a "user provider" that is referenced by the provider configuration
parameter (your_user_provider in the configuration example). This provider will turn the
username into a User object of your choice. For more information on creating or configuring a user
provider, see:
How to Create a custom User Provider
How to Load Security Users from the Database (the Entity Provider)
REMOTE_USER Based Authentication

New in version 2.6: REMOTE_USER pre authenticated firewall was introduced in Symfony 2.6.
A lot of authentication modules, like auth_kerb for Apache provide the username using the
REMOTE_USER environment variable. This variable can be trusted by the application since the
authentication happened before the request reached it.
To configure Symfony using the REMOTE_USER environment variable, simply enable the corresponding
firewall in your security configuration:
Listing 108-2
1
2
3
4
5
6
7
security:
firewalls:
secured_area:
pattern: ^/
remote_user:
provider: your_user_provider
The firewall will then provide the REMOTE_USER environment variable to your user provider. You can
change the variable name used by setting the user key in the remote_user firewall configuration.
Just like for X509 authentication, you will need to configure a "user provider". See the previous note
for more information.

Chapter 108: Using pre Authenticated Security Firewalls | 387
Chapter 109
How to Change the default Target Path

Behavior
By default, the Security component retains the information of the last request URI in a session variable
named _security.main.target_path (with main being the name of the firewall, defined in
security.yml). Upon a successful login, the user is redirected to this path, as to help them continue
from the last known page they visited.
In some situations, this is not ideal. For example, when the last request URI was an XMLHttpRequest
which returned a non-HTML or partial HTML response, the user is redirected back to a page which the
browser cannot render.
To get around this behavior, you would simply need to extend the ExceptionListener class and
override the default method named setTargetPath().
First, override the security.exception_listener.class parameter in your configuration file.
This can be done from your main configuration file (in app/config) or from a configuration file being
imported from a bundle:
Listing 109-1
1
2
3
4
parameters:
# ...
security.exception_listener.class: AppBundle\Security\Firewall\ExceptionListener
Next, create your own ExceptionListener:

Listing 109-2
1
2
3
4
5
6
7
8
9
10
11
12
// src/AppBundle/Security/Firewall/ExceptionListener.php
namespace AppBundle\Security\Firewall;
use Symfony\Component\Security\Http\Firewall\ExceptionListener as BaseExceptionListener;
class ExceptionListener extends BaseExceptionListener
{
protected function setTargetPath(Request $request)
{
// Do not save target path for XHR requests
// You can add any more logic here you want

Chapter 109: How to Change the default Target Path Behavior | 388
13
14
15
16
17
18
19
20
// Note that non-GET requests are already ignored

if ($request->isXmlHttpRequest()) {
return;
}
parent::setTargetPath($request);
}
}
Add as much or as little logic here as required for your scenario!

Chapter 109: How to Change the default Target Path Behavior | 389
Chapter 110
Using CSRF Protection in the Login Form

When using a login form, you should make sure that you are protected against CSRF (Cross-site request
forgery1). The Security component already has built-in support for CSRF. In this article you'll learn how
you can use it in your login form.
Login CSRF attacks are a bit less well-known. See Forging Login Requests2 if you're curious about
more details.
Configuring CSRF Protection

First, configure the Security component so it can use CSRF protection. The Security component needs a
CSRF token provider. You can set this to use the default provider available in the Security component:
Listing 110-1
1
2
3
4
5
6
7
8
9
10
security:
# ...
firewalls:
secured_area:
# ...
form_login:
# ...
csrf_provider: security.csrf.token_manager
The Security component can be configured further, but this is all information it needs to be able to use
CSRF in the login form.
1. https://en.wikipedia.org/wiki/Cross-site_request_forgery
2. https://en.wikipedia.org/wiki/Cross-site_request_forgery#Forging_login_requests

Chapter 110: Using CSRF Protection in the Login Form | 390
Rendering the CSRF field

Now that Security component will check for the CSRF token, you have to add a hidden field to the login
form containing the CSRF token. By default, this field is named _csrf_token. That hidden field must
contain the CSRF token, which can be generated by using the csrf_token function. That function
requires a token ID, which must be set to authenticate when using the login form:
Listing 110-2
1
2
3
4
5
6
7
8
9
10
11
12
{# src/AppBundle/Resources/views/Security/login.html.twig #}
{# ... #}
{# ... the login fields #}
<input type="hidden" name="_csrf_token"
value="{{ csrf_token('authenticate') }}"
>
<button type="submit">login</button>
</form>
After this, you have protected your login form against CSRF attacks.
You can change the name of the field by setting csrf_parameter and change the token ID by
setting intention in your configuration:
Listing 110-3
1
2
3
4
5
6
7
8
9
10
11
security:
# ...
firewalls:
secured_area:
# ...
form_login:
# ...
csrf_parameter: _csrf_security_token
intention: a_private_string

Chapter 110: Using CSRF Protection in the Login Form | 391
Chapter 111
How to Choose the Password Encoder

Algorithm Dynamically
Usually, the same password encoder is used for all users by configuring it to apply to all instances of a
specific class:
Listing 111-1
security:
# ...
encoders:
Symfony\Component\Security\Core\User\User: sha512
1
2
3
4
5
Another option is to use a "named" encoder and then select which encoder you want to use dynamically.
In the previous example, you've set the sha512 algorithm for Acme\UserBundle\Entity\User. This
may be secure enough for a regular user, but what if you want your admins to have a stronger algorithm,
for example bcrypt. This can be done with named encoders:
Listing 111-2
security:
# ...
encoders:
harsh:
algorithm: bcrypt
cost: 15
1
2
3
4
5
6
7
This creates an encoder named harsh. In order for a User instance to use it, the class must implement
EncoderAwareInterface1. The interface requires one method - getEncoderName - which should
return the name of the encoder to use:
Listing 111-3
1
2
3
4
5
6
// src/Acme/UserBundle/Entity/User.php
namespace Acme\UserBundle\Entity;
use Symfony\Component\Security\Core\Encoder\EncoderAwareInterface;
1. http://api.symfony.com/2.7/Symfony/Component/Security/Core/Encoder/EncoderAwareInterface.html

Chapter 111: How to Choose the Password Encoder Algorithm Dynamically | 392
7
8
9
10
11
12
13
14
15
16
17
class User implements UserInterface, EncoderAwareInterface

{
public function getEncoderName()
{
if ($this->isAdmin()) {
return 'harsh';
}
return null; // use the default encoder
}
}

Chapter 111: How to Choose the Password Encoder Algorithm Dynamically | 393
Chapter 112
How to Use multiple User Providers

Each authentication mechanism (e.g. HTTP Authentication, form login, etc) uses exactly one user
provider, and will use the first declared user provider by default. But what if you want to specify a few
users via configuration and the rest of your users in the database? This is possible by creating a new
provider that chains the two together:
Listing 112-1
1
2
3
4
5
6
7
8
9
10
11
12
security:
providers:
chain_provider:
chain:
providers: [in_memory, user_db]
in_memory:
memory:
users:
foo: { password: test }
user_db:
entity: { class: AppBundle\Entity\User, property: username }
Now, all authentication mechanisms will use the chain_provider, since it's the first specified. The
chain_provider will, in turn, try to load the user from both the in_memory and user_db providers.
You can also configure the firewall or individual authentication mechanisms to use a specific provider.
Again, unless a provider is specified explicitly, the first provider is always used:
Listing 112-2
1
2
3
4
5
6
7
8
9
10
11
security:
firewalls:
secured_area:
# ...
pattern: ^/
provider: user_db
http_basic:
realm: 'Secured Demo Area'
provider: in_memory
form_login: ~
In this example, if a user tries to log in via HTTP authentication, the authentication system will use the
in_memory user provider. But if the user tries to log in via the form login, the user_db provider will be
used (since it's the default for the firewall as a whole).

Chapter 112: How to Use multiple User Providers | 394
For more information about user provider and firewall configuration, see the SecurityBundle
Configuration ("security").

Chapter 112: How to Use multiple User Providers | 395
Chapter 113
How to Restrict Firewalls to a Specific Request

When using the Security component, you can create firewalls that match certain request options. In most
cases, matching against the URL is sufficient, but in special cases you can further restrict the initialization
of a firewall against other options of the request.
You can use any of these restrictions individually or mix them together to get your desired firewall
configuration.
Restricting by Pattern
This is the default restriction and restricts a firewall to only be initialized if the request URL matches the
configured pattern.
Listing 113-1
1
2
3
4
5
6
7
8
# ...
security:
firewalls:
secured_area:
pattern: ^/admin
# ...
The pattern is a regular expression. In this example, the firewall will only be activated if the URL starts
(due to the ^ regex character) with /admin. If the URL does not match this pattern, the firewall will not
be activated and subsequent firewalls will have the opportunity to be matched for this request.
Restricting by Host
If matching against the pattern only is not enough, the request can also be matched against host.
When the configuration option host is set, the firewall will be restricted to only initialize if the host from
the request matches against the configuration.

Chapter 113: How to Restrict Firewalls to a Specific Request | 396
Listing 113-2
1
2
3
4
5
6
7
8
# ...
security:
firewalls:
secured_area:
host: âdmin\.example\.com$
# ...
The host (like the pattern) is a regular expression. In this example, the firewall will only be activated
if the host is equal exactly (due to the ^ and $ regex characters) to the hostname admin.example.com.
If the hostname does not match this pattern, the firewall will not be activated and subsequent firewalls
will have the opportunity to be matched for this request.
Restricting by HTTP Methods

The configuration option methods restricts the initialization of the firewall to the provided HTTP
methods.
Listing 113-3
1
2
3
4
5
6
7
8
# ...
security:
firewalls:
secured_area:
methods: [GET, POST]
# ...
In this example, the firewall will only be activated if the HTTP method of the request is either GET or
POST. If the method is not in the array of the allowed methods, the firewall will not be activated and
subsequent firewalls will again have the opportunity to be matched for this request.

Chapter 113: How to Restrict Firewalls to a Specific Request | 397
Chapter 114
How to Restrict Firewalls to a Specific Host

As of Symfony 2.5, more possibilities to restrict firewalls have been added. You can read everything about
all the possibilities (including host) in "How to Restrict Firewalls to a Specific Request".

Chapter 114: How to Restrict Firewalls to a Specific Host | 398
Chapter 115
How to Use Voters to Check User Permissions

In Symfony, you can check the permission to access data by using the ACL module, which is a bit
overwhelming for many applications. A much easier solution is to work with custom voters, which are
like simple conditional statements.
Take a look at the authorization chapter for an even deeper understanding on voters.
How Symfony Uses Voters

In order to use voters, you have to understand how Symfony works with them. All voters are called
each time you use the isGranted() method on Symfony's authorization checker (i.e. the
security.authorization_checker service). Each one decides if the current user should have
access to some resource.
Ultimately, Symfony takes the responses from all voters and makes the final decision (to allow or deny
access to the resource) according to the strategy defined in the application, which can be: affirmative,
consensus or unanimous.
For more information take a look at the section about access decision managers.
The Voter Interface

A custom voter needs to implement
creating a voter even easier.
Listing 115-1
1
2
3
VoterInterface1 or extend AbstractVoter2, which makes
abstract class AbstractVoter implements VoterInterface

{
abstract protected function getSupportedClasses();
1. http://api.symfony.com/2.7/Symfony/Component/Security/Core/Authorization/Voter/VoterInterface.html
2. http://api.symfony.com/2.7/Symfony/Component/Security/Core/Authorization/Voter/AbstractVoter.html

Chapter 115: How to Use Voters to Check User Permissions | 399
4
5
6
abstract protected function getSupportedAttributes();

abstract protected function isGranted($attribute, $object, $user = null);
}
In this example, the voter will check if the user has access to a specific object according to your
custom conditions (e.g. they must be the owner of the object). If the condition fails, you'll return
VoterInterface::ACCESS_DENIED,
otherwise
you'll
return
VoterInterface::ACCESS_GRANTED. In case the responsibility for this decision does not belong to
this voter, it will return VoterInterface::ACCESS_ABSTAIN.
Creating the custom Voter

The goal is to create a voter that checks if a user has access to view or edit a particular object. Here's an
example implementation:
Listing 115-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
// src/AppBundle/Security/Authorization/Voter/PostVoter.php
namespace AppBundle\Security\Authorization\Voter;
use Symfony\Component\Security\Core\Authorization\Voter\AbstractVoter;
use AppBundle\Entity\User;
class PostVoter extends AbstractVoter
{
const VIEW = 'view';
const EDIT = 'edit';
protected function getSupportedAttributes()
{
return array(self::VIEW, self::EDIT);
}
protected function getSupportedClasses()
{
return array('AppBundle\Entity\Post');
}
protected function isGranted($attribute, $post, $user = null)
{
// make sure there is a user object (i.e. that the user is logged in)
if (!$user instanceof UserInterface) {
return false;
}
// double-check that the User object is the expected entity (this

// only happens when you did not configure the security system properly)
if (!$user instanceof User) {
throw new \LogicException('The user is somehow not our User class!');
}
switch ($attribute) {
case self::VIEW:
// the data object could have for example a method isPrivate()
// which checks the Boolean attribute $private
if (!$post->isPrivate()) {
return true;
}
break;
case self::EDIT:
// this assumes that the data object has a getOwner() method
// to get the entity of the user who owns this data object
if ($user->getId() === $post->getOwner()->getId()) {
return true;
}

51
52
53
54
55
56
57
break;
}
return false;
}
}
That's it! The voter is done. The next step is to inject the voter into the security layer.
To recap, here's what's expected from the three abstract methods:
getSupportedClasses()3
It tells Symfony that your voter should be called whenever an object of one of the given classes
is passed to isGranted(). For example, if you return array('AppBundle\Model\Product'), Symfony will call
your voter when a Product object is passed to isGranted().
getSupportedAttributes()4
It tells Symfony that your voter should be called whenever one of these strings is passed as the first
argument to isGranted(). For example, if you return array('CREATE', 'READ'), then Symfony will call your
voter when one of these is passed to isGranted().
isGranted()5
It implements the business logic that verifies whether or not a given user is allowed access to a given
attribute (e.g. CREATE or READ) on a given object. This method must return a boolean.
Currently, to use the AbstractVoter base class, you must be creating a voter where an object is
always passed to isGranted().
Declaring the Voter as a Service

To inject the voter into the security layer, you must declare it as a service and tag it with
security.voter:
Listing 115-3
services:
security.access.post_voter:
class:
AppBundle\Security\Authorization\Voter\PostVoter
public:
false
tags:
- { name: security.voter }
1
2
3
4
5
6
7
How to Use the Voter in a Controller

The registered voter will then always be asked as soon as the method isGranted() from the
authorization checker is called. When extending the base Controller class, you can simply call the
denyAccessUnlessGranted()()6 method:
Listing 115-4
1
2
// src/AppBundle/Controller/PostController.php
3. http://api.symfony.com/2.7/Symfony/Component/Security/Core/Authorization/Voter/AbstractVoter.html#method_getSupportedClasses
4. http://api.symfony.com/2.7/Symfony/Component/Security/Core/Authorization/Voter/AbstractVoter.html#method_getSupportedAttributes
5. http://api.symfony.com/2.7/Symfony/Component/Security/Core/Authorization/Voter/AbstractVoter.html#method_isGranted
6. http://api.symfony.com/2.7/Symfony/Bundle/FrameworkBundle/Controller/Controller.html#method_denyAccessUnlessGranted()

3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
class PostController extends Controller
{
public function showAction($id)
{
// get a Post instance
$post = ...;
// keep in mind that this will call all registered security voters
$this->denyAccessUnlessGranted('view', $post, 'Unauthorized access!');
return new Response('<h1>'.$post->getName().'</h1>');
}
}
New in version 2.6: The denyAccessUnlessGranted() method was introduced in Symfony 2.6. Prior
to Symfony 2.6, you had to call the isGranted() method of the security.context service and
throw the exception yourself.
It's that easy!
Changing the Access Decision Strategy

Imagine you have multiple voters for one action for an object. For instance, you have one voter that
checks if the user is a member of the site and a second one checking if the user is older than 18.
To handle these cases, the access decision manager uses an access decision strategy. You can configure
this to suite your needs. There are three strategies available:
affirmative (default)
This grants access as soon as there is one voter granting access;
consensus
This grants access if there are more voters granting access than denying;
unanimous
This only grants access once all voters grant access.

In the above scenario, both voters should grant access in order to grant access to the user to read the post.
In this case, the default strategy is no longer valid and unanimous should be used instead. You can set
this in the security configuration:
Listing 115-5
1
2
3
4
security:
access_decision_manager:
strategy: unanimous

Chapter 116
How to Use Access Control Lists (ACLs)

In complex applications, you will often face the problem that access decisions cannot only be based
on the person (Token) who is requesting access, but also involve a domain object that access is being
requested for. This is where the ACL system comes in.
Alternatives to ACLs
Using ACL's isn't trivial, and for simpler use cases, it may be overkill. If your permission logic
could be described by just writing some code (e.g. to check if a Blog is owned by the current
User), then consider using voters. A voter is passed the object being voted on, which you can use to
make complex decisions and effectively implement your own ACL. Enforcing authorization (e.g. the
isGranted part) will look similar to what you see in this entry, but your voter class will handle the
logic behind the scenes, instead of the ACL system.
Imagine you are designing a blog system where your users can comment on your posts. Now, you want a
user to be able to edit their own comments, but not those of other users; besides, you yourself want to be
able to edit all comments. In this scenario, Comment would be the domain object that you want to restrict
access to. You could take several approaches to accomplish this using Symfony, two basic approaches are
(non-exhaustive):
Enforce security in your business methods: Basically, that means keeping a reference inside each
Comment to all users who have access, and then compare these users to the provided Token.
Enforce security with roles: In this approach, you would add a role for each Comment object, i.e.
ROLE_COMMENT_1, ROLE_COMMENT_2, etc.
Both approaches are perfectly valid. However, they couple your authorization logic to your business code
which makes it less reusable elsewhere, and also increases the difficulty of unit testing. Besides, you could
run into performance issues if many users would have access to a single domain object.
Fortunately, there is a better way, which you will find out about now.

Chapter 116: How to Use Access Control Lists (ACLs) | 403
Bootstrapping
Now, before you can finally get into action, you need to do some bootstrapping. First, you need to
configure the connection the ACL system is supposed to use:
Listing 116-1
1
2
3
4
5
6
security:
# ...
acl:
connection: default
The ACL system requires a connection from either Doctrine DBAL (usable by default) or Doctrine
MongoDB (usable with MongoDBAclBundle1). However, that does not mean that you have to use
Doctrine ORM or ODM for mapping your domain objects. You can use whatever mapper you like
for your objects, be it Doctrine ORM, MongoDB ODM, Propel, raw SQL, etc. The choice is yours.
After the connection is configured, you have to import the database structure. Fortunately, there is a task
for this. Simply run the following command:
Listing 116-2
$ php app/console init:acl
Getting Started
Coming back to the small example from the beginning, you can now implement ACL for it.
Once the ACL is created, you can grant access to objects by creating an Access Control Entry (ACE) to
solidify the relationship between the entity and your user.
Creating an ACL and Adding an ACE

Listing 116-3
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
// src/AppBundle/Controller/BlogController.php
use
use
use
use
use
Symfony\Component\Security\Core\Exception\AccessDeniedException;
Symfony\Component\Security\Acl\Domain\ObjectIdentity;
Symfony\Component\Security\Acl\Domain\UserSecurityIdentity;
Symfony\Component\Security\Acl\Permission\MaskBuilder;
class BlogController extends Controller

{
// ...
public function addCommentAction(Post $post)
{
$comment = new Comment();
// ... setup $form, and submit data

$entityManager = $this->getDoctrine()->getManager();
$entityManager->persist($comment);
$entityManager->flush();
// creating the ACL

$aclProvider = $this->get('security.acl.provider');
1. https://github.com/IamPersistent/MongoDBAclBundle

27
28
29
30
31
32
33
34
35
36
37
38
39
40
$objectIdentity = ObjectIdentity::fromDomainObject($comment);
$acl = $aclProvider->createAcl($objectIdentity);
// retrieving the security identity of the currently logged-in user

$tokenStorage = $this->get('security.token_storage');
$user = $tokenStorage->getToken()->getUser();
$securityIdentity = UserSecurityIdentity::fromAccount($user);
// grant owner access
$acl->insertObjectAce($securityIdentity, MaskBuilder::MASK_OWNER);
$aclProvider->updateAcl($acl);
}
}
}
There are a couple of important implementation decisions in this code snippet. For now, I only want to
highlight two:
First, you may have noticed that ->createAcl() does not accept domain objects directly, but only
implementations of the ObjectIdentityInterface. This additional step of indirection allows you to
work with ACLs even when you have no actual domain object instance at hand. This will be extremely
helpful if you want to check permissions for a large number of objects without actually hydrating these
objects.
The other interesting part is the ->insertObjectAce() call. In the example, you are granting the user
who is currently logged in owner access to the Comment. The MaskBuilder::MASK_OWNER is a predefined integer bitmask; don't worry the mask builder will abstract away most of the technical details,
but using this technique you can store many different permissions in one database row which gives a
considerable boost in performance.
The order in which ACEs are checked is significant. As a general rule, you should place more specific
entries at the beginning.
Checking Access
Listing 116-4
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
// src/AppBundle/Controller/BlogController.php
// ...
class BlogController
{
// ...
public function editCommentAction(Comment $comment)
{
$authorizationChecker = $this->get('security.authorization_checker');
// check for edit access

if (false === $authorizationChecker->isGranted('EDIT', $comment)) {
throw new AccessDeniedException();
}
// ... retrieve actual comment object, and do your editing here
}
}
In this example, you check whether the user has the EDIT permission. Internally, Symfony maps the
permission to several integer bitmasks, and checks whether the user has any of them.

You can define up to 32 base permissions (depending on your OS PHP might vary between 30 to
32). In addition, you can also define cumulative permissions.
Cumulative Permissions
In the first example above, you only granted the user the OWNER base permission. While this effectively
also allows the user to perform any operation such as view, edit, etc. on the domain object, there are
cases where you may want to grant these permissions explicitly.
The MaskBuilder can be used for creating bit masks easily by combining several base permissions:
Listing 116-5
1
2
3
4
5
6
7
8
$builder = new MaskBuilder();

$builder
->add('view')
->add('edit')
->add('delete')
->add('undelete')
;
$mask = $builder->get(); // int(29)
This integer bitmask can then be used to grant a user the base permissions you added above:
Listing 116-6
1
2
$identity = new UserSecurityIdentity('johannes', 'AppBundle\Entity\User');

$acl->insertObjectAce($identity, $mask);
The user is now allowed to view, edit, delete, and un-delete objects.

Chapter 117
How to Use advanced ACL Concepts

The aim of this chapter is to give a more in-depth view of the ACL system, and also explain some of the
design decisions behind it.
Design Concepts
Symfony's object instance security capabilities are based on the concept of an Access Control List. Every
domain object instance has its own ACL. The ACL instance holds a detailed list of Access Control
Entries (ACEs) which are used to make access decisions. Symfony's ACL system focuses on two main
objectives:
providing a way to efficiently retrieve a large amount of ACLs/ACEs for your domain objects, and
to modify them;
providing a way to easily make decisions of whether a person is allowed to perform an action on a
domain object or not.
As indicated by the first point, one of the main capabilities of Symfony's ACL system is a highperformance way of retrieving ACLs/ACEs. This is extremely important since each ACL might have
several ACEs, and inherit from another ACL in a tree-like fashion. Therefore, no ORM is leveraged,
instead the default implementation interacts with your connection directly using Doctrine's DBAL.
Object Identities
The ACL system is completely decoupled from your domain objects. They don't even have to be stored
in the same database, or on the same server. In order to achieve this decoupling, in the ACL system your
objects are represented through object identity objects. Every time you want to retrieve the ACL for a
domain object, the ACL system will first create an object identity from your domain object, and then pass
this object identity to the ACL provider for further processing.
Security Identities
This is analog to the object identity, but represents a user, or a role in your application. Each role, or user
has its own security identity.

Chapter 117: How to Use advanced ACL Concepts | 407
For users, the security identity is based on the username. This means that, if for any reason,
a user's username was to change, you must ensure its security identity is updated too. The
MutableAclProvider::updateUserSecurityIdentity()1 method is there to handle the
update.
Database Table Structure

The default implementation uses five database tables as listed below. The tables are ordered from least
rows to most rows in a typical application:
acl_security_identities: This table records all security identities (SID) which hold ACEs. The default
implementation ships with two security identities: RoleSecurityIdentity2 and UserSecurityIdentity3.
acl_classes: This table maps class names to a unique ID which can be referenced from other tables.
acl_object_identities: Each row in this table represents a single domain object instance.
acl_object_identity_ancestors: This table allows all the ancestors of an ACL to be determined in a
very efficient way.
acl_entries: This table contains all ACEs. This is typically the table with the most rows. It can
contain tens of millions without significantly impacting performance.
Scope of Access Control Entries

Access control entries can have different scopes in which they apply. In Symfony, there are basically two
different scopes:
Class-Scope: These entries apply to all objects with the same class.
Object-Scope: This was the scope solely used in the previous chapter, and it only applies to one
specific object.
Sometimes, you will find the need to apply an ACE only to a specific field of the object. Suppose you
want the ID only to be viewable by an administrator, but not by your customer service. To solve this
common problem, two more sub-scopes have been added:
Class-Field-Scope: These entries apply to all objects with the same class, but only to a specific field
of the objects.
Object-Field-Scope: These entries apply to a specific object, and only to a specific field of that
object.
Pre-Authorization Decisions
For pre-authorization decisions, that is decisions made before any secure method (or secure action) is
invoked, the proven AccessDecisionManager service is used. The AccessDecisionManager is also used
for reaching authorization decisions based on roles. Just like roles, the ACL system adds several new
attributes which may be used to check for different permissions.
1. http://api.symfony.com/2.7/Symfony/Component/Security/Acl/Dbal/MutableAclProvider.html#method_updateUserSecurityIdentity
2. http://api.symfony.com/2.7/Symfony/Component/Security/Acl/Domain/RoleSecurityIdentity.html
3. http://api.symfony.com/2.7/Symfony/Component/Security/Acl/Domain/UserSecurityIdentity.html

Built-in Permission Map

Attribute
Intended Meaning
Integer Bitmasks
VIEW
Whether someone is allowed to

view the domain object.
VIEW, EDIT, OPERATOR,

MASTER, or OWNER
EDIT

EDIT, OPERATOR, MASTER, or
make changes to the domain object. OWNER
CREATE

create the domain object.
CREATE, OPERATOR, MASTER, or

OWNER
DELETE

delete the domain object.
DELETE, OPERATOR, MASTER, or

OWNER
UNDELETE

restore a previously deleted domain
object.
UNDELETE, OPERATOR,
MASTER, or OWNER
OPERATOR

perform all of the above actions.
OPERATOR, MASTER, or OWNER
MASTER

perform all of the above actions,
and in addition is allowed to grant
any of the above permissions to
others.
MASTER, or OWNER
OWNER
Whether someone owns the

domain object. An owner can
perform any of the above actions
and grant master and owner
permissions.
OWNER
Permission Attributes vs. Permission Bitmasks

Attributes are used by the AccessDecisionManager, just like roles. Often, these attributes represent in
fact an aggregate of integer bitmasks. Integer bitmasks on the other hand, are used by the ACL system
internally to efficiently store your users' permissions in the database, and perform access checks using
extremely fast bitmask operations.
Extensibility
The above permission map is by no means static, and theoretically could be completely replaced at will.
However, it should cover most problems you encounter, and for interoperability with other bundles, you
are encouraged to stick to the meaning envisaged for them.
Post Authorization Decisions

Post authorization decisions are made after a secure method has been invoked, and typically involve the
domain object which is returned by such a method. After invocation providers also allow to modify, or
filter the domain object before it is returned.

Due to current limitations of the PHP language, there are no post-authorization capabilities build into the
core Security component. However, there is an experimental JMSSecurityExtraBundle4 which adds these
capabilities. See its documentation for further information on how this is accomplished.
Process for Reaching Authorization Decisions

The ACL class provides two methods for determining whether a security identity has the required
bitmasks, isGranted and isFieldGranted. When the ACL receives an authorization request
through one of these methods, it delegates this request to an implementation of
PermissionGrantingStrategy5. This allows you to replace the way access decisions are reached
without actually modifying the ACL class itself.
The PermissionGrantingStrategy first checks all your object-scope ACEs. If none is applicable,
the class-scope ACEs will be checked. If none is applicable, then the process will be repeated with the
ACEs of the parent ACL. If no parent ACL exists, an exception will be thrown.
4. https://github.com/schmittjoh/JMSSecurityExtraBundle
5. http://api.symfony.com/2.7/Symfony/Component/Security/Acl/Domain/PermissionGrantingStrategy.html

Chapter 118
How to Force HTTPS or HTTP for different URLs

You can force areas of your site to use the HTTPS protocol in the security config. This is done through
the access_control rules using the requires_channel option. For example, if you want to force
all URLs starting with /secure to use HTTPS then you could use the following configuration:
Listing 118-1
1
2
3
4
5
6
security:
# ...
access_control:
- { path: ^/secure, roles: ROLE_ADMIN, requires_channel: https }
The login form itself needs to allow anonymous access, otherwise users will be unable to authenticate.
To force it to use HTTPS you can still use access_control rules by using the
IS_AUTHENTICATED_ANONYMOUSLY role:
Listing 118-2
1
2
3
4
5
6
security:
# ...
access_control:
- { path: ^/login, roles: IS_AUTHENTICATED_ANONYMOUSLY, requires_channel: https }
It is also possible to specify using HTTPS in the routing configuration, see How to Force Routes to always
Use HTTPS or HTTP for more details.

Chapter 118: How to Force HTTPS or HTTP for different URLs | 411
Chapter 119
How to Secure any Service or Method in your

Application
In the security chapter, you can see how to secure a controller by requesting the
security.authorization_checker service from the Service Container and checking the current
user's role:
Listing 119-1
1
2
3
4
5
6
7
8
9
// ...
use Symfony\Component\Security\Core\Exception\AccessDeniedException;
public function helloAction($name)
{
$this->denyAccessUnlessGranted('ROLE_ADMIN');
// ...
}
You can also secure any service by injecting the security.authorization_checker service into it.
For a general introduction to injecting dependencies into services see the Service Container chapter of the
book. For example, suppose you have a NewsletterManager class that sends out emails and you want
to restrict its use to only users who have some ROLE_NEWSLETTER_ADMIN role. Before you add security,
the class looks something like this:
Listing 119-2
1
2
3
4
5
6
7
8
9
10
11
12
// src/AppBundle/Newsletter/NewsletterManager.php
namespace AppBundle\Newsletter;
class NewsletterManager
{
public function sendNewsletter()
{
// ... where you actually do the work
}
// ...
}
Your goal is to check the user's role when the sendNewsletter() method is called. The first step
towards this is to inject the security.authorization_checker service into the object. Since it
Chapter 119: How to Secure any Service or Method in your Application | 412
won't make sense not to perform the security check, this is an ideal candidate for constructor injection,
which guarantees that the authorization checker object will be available inside the
NewsletterManager class:
Listing 119-3
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
// src/AppBundle/Newsletter/NewsletterManager.php
// ...
{
{
}
// ...
}
Then in your service configuration, you can inject the service:

Listing 119-4
1
2
3
4
5
services:
newsletter_manager:
class:
AppBundle\Newsletter\NewsletterManager
arguments: ['@security.authorization_checker']
The injected service can then be used to perform the security check when the sendNewsletter()
method is called:
Listing 119-5
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
use Symfony\Component\Security\Core\Exception\AccessDeniedException;
// ...
{
{
}
{
if (false === $this->authorizationChecker->isGranted('ROLE_NEWSLETTER_ADMIN')) {
throw new AccessDeniedException();
}
// ...
}
// ...
}
If the current user does not have the ROLE_NEWSLETTER_ADMIN, they will be prompted to log in.

Securing Methods Using Annotations

You can also secure method calls in any service with annotations by using the optional
JMSSecurityExtraBundle1 bundle. This bundle is not included in the Symfony Standard Distribution, but
you can choose to install it.
To enable the annotations functionality, tag the service you want to secure with the
security.secure_service tag (you can also automatically enable this functionality for all services,
see the sidebar below):
Listing 119-6
services:
newsletter_manager:
class: AppBundle\Newsletter\NewsletterManager
tags:
- { name: security.secure_service }
1
2
3
4
5
6
You can then achieve the same results as above using an annotation:
Listing 119-7
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
use JMS\SecurityExtraBundle\Annotation\Secure;
// ...
{
/**
* @Secure(roles="ROLE_NEWSLETTER_ADMIN")
*/
{
// ...
}
// ...
}
The annotations work because a proxy class is created for your class which performs the security
checks. This means that, whilst you can use annotations on public and protected methods, you
cannot use them with private methods or methods marked final.
The JMSSecurityExtraBundle also allows you to secure the parameters and return values of methods. For
more information, see the JMSSecurityExtraBundle2 documentation.
Activating the Annotations Functionality for all Services

When securing the method of a service (as shown above), you can either tag each service
individually, or activate the functionality for all services at once. To do so, set the
secure_all_services configuration option to true:
Listing 119-8
1
2
3
4
jms_security_extra:
# ...
secure_all_services: true
The disadvantage of this method is that, if activated, the initial page load may be very slow
depending on how many services you have defined.

Chapter 120
How Does the Security access_control Work?

For each incoming request, Symfony checks each access_control entry to find one that matches the
current request. As soon as it finds a matching access_control entry, it stops - only the first matching
access_control is used to enforce access.
Each access_control has several options that configure two different things:
1. should the incoming request match this access control entry
2. once it matches, should some sort of access restriction be enforced:
1. Matching Options
Symfony creates an instance of RequestMatcher1 for each access_control entry, which determines
whether or not a given access control should be used on this request. The following access_control
options are used for matching:
path
ip
or ips
host
methods
Take the following access_control entries as an example:

Listing 120-1
1
2
3
4
5
6
7
8
security:
# ...
access_control:
- { path: ^/admin,
- { path: ^/admin,
- { path: ^/admin,
- { path: ^/admin,
roles:
roles:
roles:
roles:
ROLE_USER_IP, ip: 127.0.0.1 }

ROLE_USER_HOST, host: symfony\.com$ }
ROLE_USER_METHOD, methods: [POST, PUT] }
ROLE_USER }
For each incoming request, Symfony will decide which access_control to use based on the URI,
the client's IP address, the incoming host name, and the request method. Remember, the first rule that
1. http://api.symfony.com/2.7/Symfony/Component/HttpFoundation/RequestMatcher.html

Chapter 120: How Does the Security access_control Work? | 415
matches is used, and if ip, host or method are not specified for an entry, that access_control will
match any ip, host or method:
URI
IP
/admin/
HOST
METHOD
access_control
Why?
127.0.0.1 example.com GET
rule #1
(ROLE_USER_IP)
The URI matches path and the IP

matches ip.
127.0.0.1 symfony.com GET
rule #1
(ROLE_USER_IP)
The path and ip still match. This

would also match the
ROLE_USER_HOST entry, but only the
first access_control match is
used.
168.0.0.1 symfony.com GET
rule #2
(ROLE_USER_HOST)
The ip doesn't match the first

rule, so the second rule (which
matches) is used.
168.0.0.1 symfony.com POST
rule #2
(ROLE_USER_HOST)
The second rule still matches.

This would also match the third
rule (ROLE_USER_METHOD), but only
the first matched access_control
is used.
168.0.0.1 example.com POST
rule #3
(ROLE_USER_METHOD)
The ip and host don't match the

first two entries, but the third ROLE_USER_METHOD - matches and is
used.
168.0.0.1 example.com GET
rule #4
(ROLE_USER)
The ip, host and method prevent

the first three entries from
matching. But since the URI
matches the path pattern of the
ROLE_USER entry, it is used.
127.0.0.1 symfony.com POST
matches no
entries
This doesn't match any

access_control rules, since its URI
doesn't match any of the path
values.
user
/admin/
user
/admin/
user
/admin/
user
/admin/
user
/admin/
user
/foo
2. Access Enforcement
Once Symfony has decided which access_control entry matches (if any), it then enforces access
restrictions based on the roles, allow_if and requires_channel options:
If the user does not have the given role(s), then access is denied (internally, an
is thrown);
allow_if If the expression returns false, then access is denied;
requires_channel If the incoming request's channel (e.g. http) does not match this value (e.g. https), the
user will be redirected (e.g. redirected from http to https, or vice versa).
role
AccessDeniedException2
If access is denied, the system will try to authenticate the user if not already (e.g. redirect the user to
the login page). If the user is already logged in, the 403 "access denied" error page will be shown. See
How to Customize Error Pages for more information.
2. http://api.symfony.com/2.7/Symfony/Component/Security/Core/Exception/AccessDeniedException.html

Matching access_control By IP
Certain situations may arise when you need to have an access_control entry that only matches
requests coming from some IP address or range. For example, this could be used to deny access to a URL
pattern to all requests except those from a trusted, internal server.
As you'll read in the explanation below the example, the ips option does not restrict to a specific IP
address. Instead, using the ips key means that the access_control entry will only match this IP
address, and users accessing it from a different IP address will continue down the access_control
list.
Here is an example of how you configure some example /internal* URL pattern so that it is only
accessible by requests from the local server itself:
Listing 120-2
1
2
3
4
5
6
7
security:
# ...
access_control:
#
- { path: ^/internal, roles: IS_AUTHENTICATED_ANONYMOUSLY, ips: [127.0.0.1, ::1] }
- { path: ^/internal, roles: ROLE_NO_ACCESS }
Here is how it works when the path is /internal/something coming from the external IP address
10.0.0.1:
The first access control rule is ignored as the path matches but the IP address does not match either
of the IPs listed;
The second access control rule is enabled (the only restriction being the path) and so it matches.
If you make sure that no users ever have ROLE_NO_ACCESS, then access is denied (ROLE_NO_ACCESS can be
anything that does not match an existing role, it just serves as a trick to always deny access).
But if the same request comes from 127.0.0.1 or ::1 (the IPv6 loopback address):
Now, the first access control rule is enabled as both the path and the ip match: access is allowed as
the user always has the IS_AUTHENTICATED_ANONYMOUSLY role.
The second access rule is not examined as the first rule matched.
Securing by an Expression
Once an access_control entry is matched, you can deny access via the roles key or use more
complex logic with an expression in the allow_if key:
Listing 120-3
1
2
3
4
5
6
7
security:
# ...
access_control:
path: ^/_internal/secure
allow_if: "'127.0.0.1' == request.getClientIp() or has_role('ROLE_ADMIN')"
In this case, when the user tries to access any URL starting with /_internal/secure, they will only
be granted access if the IP address is 127.0.0.1 or if the user has the ROLE_ADMIN role.
Inside the expression, you have access to a number of different variables and functions including
request, which is the Symfony Request3 object (see Request).

For a list of the other functions and variables, see functions and variables.
Forcing a Channel (http, https)

You can also require a user to access a URL via SSL; just use the requires_channel argument in
any access_control entries. If this access_control is matched and the request is using the http
channel, the user will be redirected to https:
Listing 120-4
1
2
3
4
5
security:
# ...
access_control:
- { path: ^/cart/checkout, roles: IS_AUTHENTICATED_ANONYMOUSLY, requires_channel: https }

Chapter 121
How to Use the Serializer

Serializing and deserializing to and from objects and different formats (e.g. JSON or XML) is a very
complex topic. Symfony comes with a Serializer Component, which gives you some tools that you can
leverage for your solution.
In fact, before you start, get familiar with the serializer, normalizers and encoders by reading the Serializer
Component.
Activating the Serializer

New in version 2.3: The Serializer has always existed in Symfony, but prior to Symfony 2.3, you needed
to build the serializer service yourself.
The serializer service is not available by default. To turn it on, activate it in your configuration:
Listing 121-1
framework:
# ...
serializer:
enabled: true
1
2
3
4
5
Using the Serializer Service

Once enabled, the serializer service can be injected in any service where you need it or it can be used
in a controller like the following:
Listing 121-2
1
2
3
4
5
6
7
8
9
// src/AppBundle/Controller/DefaultController.php
class DefaultController extends Controller
{
{

Chapter 121: How to Use the Serializer | 419
10
11
12
13
14
$serializer = $this->get('serializer');
// ...
}
}
Adding Normalizers and Encoders

New in version 2.7: The ObjectNormalizer1 is enabled by default in Symfony 2.7. In prior versions,
you needed to load your own normalizer.
Once enabled, the serializer service will be available in the container and will be loaded with two
encoders (JsonEncoder2 and XmlEncoder3) and the ObjectNormalizer normalizer.
You can load normalizers and/or encoders by tagging them as serializer.normalizer and
serializer.encoder. It's also possible to set the priority of the tag in order to decide the matching order.
Here is an example on how to load the GetSetMethodNormalizer4:
Listing 121-3
1
2
3
4
5
6
services:
get_set_method_normalizer:
class: Symfony\Component\Serializer\Normalizer\GetSetMethodNormalizer
tags:
- { name: serializer.normalizer }
Using Serialization Groups Annotations

New in version 2.7: Support for serialization groups was introduced in Symfony 2.7.
Enable serialization groups annotation with the following configuration:
Listing 121-4
1
2
3
4
5
framework:
# ...
serializer:
enable_annotations: true
Next, add the @Groups annotations to your class and choose which groups to use when serializing:
Listing 121-5
1
2
3
4
5
$serializer = $this->get('serializer');
$json = $serializer->serialize(
$someObject,
'json', array('groups' => array('group1'))
);
In addition to the @Groups annotation, the Serializer component also supports Yaml or XML files. These
files are automatically loaded when being stored in one of the following locations:
The serialization.yml or serialization.xml file in the Resources/config/ directory of a bundle;
All *.yml and *.xml files in the Resources/config/serialization/ directory of a bundle.
1. http://api.symfony.com/2.7/Symfony/Component/Serializer/Normalizer/ObjectNormalizer.html
2. http://api.symfony.com/2.7/Symfony/Component/Serializer/Encoder/JsonEncoder.html
3. http://api.symfony.com/2.7/Symfony/Component/Serializer/Encoder/XmlEncoder.html
4. http://api.symfony.com/2.7/Symfony/Component/Serializer/Normalizer/GetSetMethodNormalizer.html

Enabling the Metadata Cache

New in version 2.7: Serializer metadata and the ability to cache them were introduced in Symfony 2.7.
Metadata used by the Serializer component such as groups can be cached to enhance application
performance. Any service implementing the Doctrine\Common\Cache\Cache interface can be used.
A service leveraging APCu5 (and APC for PHP < 5.5) is built-in.
Listing 121-6
1
2
3
4
5
framework:
# ...
serializer:
cache: serializer.mapping.cache.apc
Going Further with the Serializer Component

ApiPlatform6 provides an API system supporting JSON-LD7 and Hydra Core Vocabulary8 hypermedia
formats. It is built on top of the Symfony Framework and its Serializer component. It provides custom
normalizers and a custom encoder, custom metadata and a caching system.
If you want to leverage the full power of the Symfony Serializer component, take a look at how this
bundle works.
5. https://github.com/krakjoe/apcu
6. https://github.com/api-platform/core
7. http://json-ld.org
8. http://hydra-cg.com

Chapter 122
How to Work with Scopes

This article is all about scopes, a somewhat advanced topic related to the Service Container. If you've ever
gotten an error mentioning "scopes" when creating services, then this article is for you.
If you are trying to inject the request service, the simple solution is to inject the request_stack
service instead and access the current Request by calling the getCurrentRequest()1 method (see
Injecting the Request). The rest of this entry talks about scopes in a theoretical and more advanced
way. If you're dealing with scopes for the request service, simply inject request_stack.
Understanding Scopes
The scope of a service controls how long an instance of a service is used by the container. The
DependencyInjection component provides two generic scopes:
container (the default one):
The same instance is used each time you ask for it from this container.
prototype:
A new instance is created each time you ask for the service.
The ContainerAwareHttpKernel2 also defines a third scope: request. This scope is tied to the
request, meaning a new instance is created for each subrequest and is unavailable outside the request (for
instance in the CLI).
An Example: Client Scope

Other than the request service (which has a simple solution, see the above note), no services in the
default Symfony container belong to any scope other than container and prototype. But for the
purposes of this article, imagine there is another scope client and a service client_configuration
1. http://api.symfony.com/2.7/Symfony/Component/HttpFoundation/RequestStack.html#method_getCurrentRequest
2. http://api.symfony.com/2.7/Symfony/Component/HttpKernel/DependencyInjection/ContainerAwareHttpKernel.html

Chapter 122: How to Work with Scopes | 422
that belongs to it. This is not a common situation, but the idea is that you may enter and exit multiple
client scopes during a request, and each has its own client_configuration service.
Scopes add a constraint on the dependencies of a service: a service cannot depend on services from
a narrower scope. For example, if you create a generic my_foo service, but try to inject the
client_configuration service, you will receive a ScopeWideningInjectionException3 when
compiling the container. Read the sidebar below for more details.
Scopes and Dependencies

Imagine you've configured a my_mailer service. You haven't configured the scope of the service,
so it defaults to container. In other words, every time you ask the container for the my_mailer
service, you get the same object back. This is usually how you want your services to work.
Imagine, however, that you need the client_configuration service in your my_mailer
service, maybe because you're reading some details from it, such as what the "sender" address should
be. You add it as a constructor argument. There are several reasons why this presents a problem:
When requesting my_mailer, an instance of my_mailer (called MailerA here) is created
and the client_configuration service ( called ConfigurationA here) is passed to it. Life is
good!
Your application now needs to do something with another client, and you've designed your
application in such a way that you handle this by entering a new client_configuration
scope and setting a new client_configuration service into the container. Call this
ConfigurationB.
Somewhere in your application, you once again ask for the my_mailer service. Since your
service is in the container scope, the same instance (MailerA) is just re-used. But here's
the problem: the MailerA instance still contains the old ConfigurationA object, which is now
not the correct configuration object to have (ConfigurationB is now the current
client_configuration service). This is subtle, but the mis-match could cause major
problems, which is why it's not allowed.
So, that's the reason why scopes exist, and how they can cause problems. Keep reading to find
out the common solutions.
A service can of course depend on a service from a wider scope without any issue.
Using a Service from a Narrower Scope

There are two solutions to the scope problem:
A) Put your service in the same scope as the dependency (or a narrower one). If you depend on the
client_configuration service, this means putting your new service in the client scope (see A) Changing
the Scope of your Service);
B) Pass the entire container to your service and retrieve your dependency from the container each
time you need it to be sure you have the right instance -- your service can live in the default container
scope (see B) Passing the Container as a Dependency of your Service).
Each scenario is detailed in the following sections.
3. http://api.symfony.com/2.7/Symfony/Component/DependencyInjection/Exception/ScopeWideningInjectionException.html

Prior to Symfony 2.7, there was another alternative based on synchronized services. However,
these kind of services have been deprecated starting from Symfony 2.7.
A) Changing the Scope of your Service

Changing the scope of a service should be done in its definition. This example assumes that the Mailer
class has a __construct function whose first argument is the ClientConfiguration object:
Listing 122-1
1
2
3
4
5
6
services:
my_mailer:
class: AppBundle\Mail\Mailer
scope: client
arguments: ['@client_configuration']
B) Passing the Container as a Dependency of your Service

Setting the scope to a narrower one is not always possible (for instance, a twig extension must be in the
container scope as the Twig environment needs it as a dependency). In these cases, you can pass the
entire container into your service:
Listing 122-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
// src/AppBundle/Mail/Mailer.php
namespace AppBundle\Mail;
use Symfony\Component\DependencyInjection\ContainerInterface;
class Mailer
{
protected $container;
public function __construct(ContainerInterface $container)
{
$this->container = $container;
}
public function sendEmail()
{
$request = $this->container->get('client_configuration');
// ... do something using the client configuration here
}
}
Take care not to store the client configuration in a property of the object for a future call of the
service as it would cause the same issue described in the first section (except that Symfony cannot
detect that you are wrong).
The service configuration for this class would look something like this:
Listing 122-3
1
2
3
4
5
6
services:
my_mailer:
class:
AppBundle\Mail\Mailer
arguments: ['@service_container']
# scope: container can be omitted as it is the default
Injecting the whole container into a service is generally not a good idea (only inject what you need).

Chapter 123
How to Work with Compiler Passes in Bundles

Compiler passes give you an opportunity to manipulate other service definitions that have been registered
with the service container. You can read about how to create them in the components section "Compiling
the Container". To register a compiler pass from a bundle you need to add it to the build method of the
bundle definition class:
Listing 123-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
// src/Acme/MailerBundle/AcmeMailerBundle.php
namespace Acme\MailerBundle;
use Acme\MailerBundle\DependencyInjection\Compiler\CustomCompilerPass;
class AcmeMailerBundle extends Bundle
{
{
$container->addCompilerPass(new CustomCompilerPass());
}
}
One of the most common use-cases of compiler passes is to work with tagged services (read more about
tags in the components section "Working with Tagged Services"). If you are using custom tags in a bundle
then by convention, tag names consist of the name of the bundle (lowercase, underscores as separators),
followed by a dot, and finally the "real" name. For example, if you want to introduce some sort of
"transport" tag in your AcmeMailerBundle, you should call it acme_mailer.transport.

Chapter 123: How to Work with Compiler Passes in Bundles | 425
Chapter 124
Session Proxy Examples

The session proxy mechanism has a variety of uses and this article demonstrates two common uses.
Rather than using the regular session handler, you can create a custom save handler just by defining a
class that extends the SessionHandlerProxy1 class.
Then, define a new service related to the custom session handler:
Listing 124-1
services:
app.session_handler:
class: AppBundle\Session\CustomSessionHandler
1
2
3
4
Finally, use the framework.session.handler_id configuration option to tell Symfony to use your
own session handler instead of the default one:
Listing 124-2
framework:
session:
# ...
handler_id: app.session_handler
1
2
3
4
5
Keep reading the next sections to learn how to use the session handlers in practice to solve two common
use cases: encrypt session information and define readonly guest sessions.
Encryption of Session Data

If you wanted to encrypt the session data, you could use the proxy to encrypt and decrypt the session as
required:
Listing 124-3
1
2
3
4
5
6
// src/AppBundle/Session/EncryptedSessionProxy.php
namespace AppBundle\Session;
use Symfony\Component\HttpFoundation\Session\Storage\Proxy\SessionHandlerProxy;
class EncryptedSessionProxy extends SessionHandlerProxy
1. http://api.symfony.com/2.7/Symfony/Component/HttpFoundation/Session/Storage/Proxy/SessionHandlerProxy.html

Chapter 124: Session Proxy Examples | 426
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
{
private $key;
public function __construct(\SessionHandlerInterface $handler, $key)
{
$this->key = $key;
parent::__construct($handler);
}
public function read($id)
{
$data = parent::read($id);
return mcrypt_decrypt(\MCRYPT_3DES, $this->key, $data);
}
public function write($id, $data)
{
$data = mcrypt_encrypt(\MCRYPT_3DES, $this->key, $data);
return parent::write($id, $data);
}
}
Readonly Guest Sessions

There are some applications where a session is required for guest users, but where there is no particular
need to persist the session. In this case you can intercept the session before it is written:
Listing 124-4
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
// src/AppBundle/Session/ReadOnlySessionProxy.php
namespace AppBundle\Session;
use AppBundle\Entity\User;
use Symfony\Component\HttpFoundation\Session\Storage\Proxy\SessionHandlerProxy;
class ReadOnlySessionProxy extends SessionHandlerProxy
{
private $user;
public function __construct(\SessionHandlerInterface $handler, User $user)
{
$this->user = $user;
parent::__construct($handler);
}
public function write($id, $data)
{
if ($this->user->isGuest()) {
return;
}
return parent::write($id, $data);
}
}

Chapter 124: Session Proxy Examples | 427
Chapter 125
Making the Locale "Sticky" during a User's

Session
Prior to Symfony 2.1, the locale was stored in a session attribute called _locale. Since 2.1, it is stored in
the Request, which means that it's not "sticky" during a user's request. In this article, you'll learn how to
make the locale of a user "sticky" so that once it's set, that same locale will be used for every subsequent
request.
Creating a LocaleListener
To simulate that the locale is stored in a session, you need to create and register a new event listener. The
listener will look something like this. Typically, _locale is used as a routing parameter to signify the
locale, though it doesn't really matter how you determine the desired locale from the request:
Listing 125-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
// src/AppBundle/EventListener/LocaleListener.php
use Symfony\Component\HttpKernel\Event\GetResponseEvent;
use Symfony\Component\HttpKernel\KernelEvents;
class LocaleListener implements EventSubscriberInterface
{
private $defaultLocale;
public function __construct($defaultLocale = 'en')
{
$this->defaultLocale = $defaultLocale;
}
public function onKernelRequest(GetResponseEvent $event)
{
$request = $event->getRequest();
if (!$request->hasPreviousSession()) {
return;
}

Chapter 125: Making the Locale "Sticky" during a User's Session | 428
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
// try to see if the locale has been set as a _locale routing parameter
if ($locale = $request->attributes->get('_locale')) {
$request->getSession()->set('_locale', $locale);
} else {
// if no explicit locale has been set on this request, use one from the session
$request->setLocale($request->getSession()->get('_locale', $this->defaultLocale));
}
}
{
return array(
// must be registered after the default Locale listener
KernelEvents::REQUEST => array(array('onKernelRequest', 15)),
);
}
}
Then register the listener:

Listing 125-2
1
2
3
4
5
6
services:
app.locale_listener:
class: AppBundle\EventListener\LocaleListener
arguments: ['%kernel.default_locale%']
tags:
- { name: kernel.event_subscriber }
That's it! Now celebrate by changing the user's locale and seeing that it's sticky throughout the request.
Remember, to get the user's locale, always use the Request::getLocale1 method:
Listing 125-3
// from a controller...
1
2
3
4
5
6
7
public function indexAction(Request $request)

{
$locale = $request->getLocale();
}
Setting the Locale Based on the User's Preferences

You might want to improve this technique even further and define the locale based on the user entity
of the logged in user. However, since the LocaleListener is called before the FirewallListener,
which is responsible for handling authentication and setting the user token on the TokenStorage, you
have no access to the user which is logged in.
Suppose you have defined a locale property on your User entity and you want to use this as the locale
for the given user. To accomplish this, you can hook into the login process and update the user's session
with this locale value before they are redirected to their first page.
To do this, you need an event listener for the security.interactive_login event:
Listing 125-4
1
2
3
4
5
6
7
8
9
// src/AppBundle/EventListener/UserLocaleListener.php
use Symfony\Component\HttpFoundation\Session\Session;
use Symfony\Component\Security\Http\Event\InteractiveLoginEvent;
/**
* Stores the locale of the user in the session after the
* login. This can be used by the LocaleListener afterwards.
1. http://api.symfony.com/2.7/Symfony/Component/HttpFoundation/Request.html#method_getLocale

10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
*/
class UserLocaleListener
{
/**
* @var Session
*/
private $session;
public function __construct(Session $session)
{
$this->session = $session;
}
/**
* @param InteractiveLoginEvent $event
*/
public function onInteractiveLogin(InteractiveLoginEvent $event)
{
$user = $event->getAuthenticationToken()->getUser();
if (null !== $user->getLocale()) {
$this->session->set('_locale', $user->getLocale());
}
}
}
Then register the listener:

Listing 125-5
1
2
3
4
5
6
7
services:
app.user_locale_listener:
class: AppBundle\EventListener\UserLocaleListener
tags:
- { name: kernel.event_listener, event: security.interactive_login, method: onInteractiveLogin }
In order to update the language immediately after a user has changed their language preferences, you
need to update the session after an update to the User entity.

Chapter 126
Configuring the Directory where Session Files

are Saved
default, the Symfony Standard Edition uses the global php.ini values for
session.save_handler and session.save_path to determine where to store session data. This
By
is because of the following configuration:

Listing 126-1
1
2
3
4
5
framework:
session:
# handler_id set to null will use default session handler from php.ini
handler_id: ~
With this configuration, changing where your session metadata is stored is entirely up to your php.ini
configuration.
However, if you have the following configuration, Symfony will store the session data in files in the cache
directory %kernel.cache_dir%/sessions. This means that when you clear the cache, any current
sessions will also be deleted:
Listing 126-2
1
2
3
framework:
session: ~
Using a different directory to save session data is one method to ensure that your current sessions aren't
lost when you clear Symfony's cache.
Using a different session save handler is an excellent (yet more complex) method of session
management available within Symfony. See Configuring Sessions and Save Handlers for a discussion
of session save handlers. There are also entries in the cookbook about storing sessions in a relational
database or a NoSQL database.
To change the directory in which Symfony saves session data, you only need change the framework
configuration. In this example, you will change the session directory to app/sessions:
Listing 126-3

Chapter 126: Configuring the Directory where Session Files are Saved | 431
1
2
3
4
5
framework:
session:
handler_id: session.handler.native_file
save_path: '%kernel.root_dir%/sessions'

Chapter 126: Configuring the Directory where Session Files are Saved | 432
Chapter 127
Bridge a legacy Application with Symfony

Sessions
New in version 2.3: The ability to integrate with a legacy PHP session was introduced in Symfony 2.3.
If you're integrating the Symfony full-stack Framework into a legacy application that starts the session
with session_start(), you may still be able to use Symfony's session management by using the PHP
Bridge session.
If the application has it's own PHP save handler, you can specify null for the handler_id:
Listing 127-1
1
2
3
4
framework:
session:
storage_id: session.storage.php_bridge
handler_id: ~
Otherwise, if the problem is simply that you cannot avoid the application starting the session with
session_start(), you can still make use of a Symfony based session save handler by specifying the
save handler as in the example below:
Listing 127-2
1
2
3
4
framework:
session:
storage_id: session.storage.php_bridge
handler_id: session.handler.native_file
If the legacy application requires its own session save handler, do not override this. Instead set
handler_id: ~. Note that a save handler cannot be changed once the session has been started.
If the application starts the session before Symfony is initialized, the save handler will have already
been set. In this case, you will need handler_id: ~. Only override the save handler if you are sure
the legacy application can use the Symfony save handler without side effects and that the session has
not been started before Symfony is initialized.
For more details, see Integrating with Legacy Sessions.

Chapter 127: Bridge a legacy Application with Symfony Sessions | 433
Chapter 128
Limit Session Metadata Writes

The default behavior of PHP session is to persist the session regardless of whether the session data has
changed or not. In Symfony, each time the session is accessed, metadata is recorded (session created/last
used) which can be used to determine session age and idle time.
If for performance reasons you wish to limit the frequency at which the session persists, this feature can
adjust the granularity of the metadata updates and persist the session less often while still maintaining
relatively accurate metadata. If other session data is changed, the session will always persist.
You can tell Symfony not to update the metadata "session last updated" time until a certain amount
of time has passed, by setting framework.session.metadata_update_threshold to a value in
seconds greater than zero:
Listing 128-1
1
2
3
framework:
session:
metadata_update_threshold: 120
PHP default's behavior is to save the session whether it has been changed or not. When using
framework.session.metadata_update_threshold Symfony will wrap the session handler
(configured at framework.session.handler_id) into the WriteCheckSessionHandler. This
will prevent any session write if the session was not modified.
Be aware that if the session is not written at every request, it may be garbage collected sooner than
usual. This means that your users may be logged out sooner than expected.

Chapter 128: Limit Session Metadata Writes | 434
Chapter 129
Avoid Starting Sessions for Anonymous Users

Sessions are automatically started whenever you read, write or even check for the existence of data in the
session. This means that if you need to avoid creating a session cookie for some users, it can be difficult:
you must completely avoid accessing the session.
For example, one common problem in this situation involves checking for flash messages, which are
stored in the session. The following code would guarantee that a session is always started:
Listing 129-1
1
2
3
4
5
{% for flashMessage in app.session.flashBag.get('notice') %}

<div class="flash-notice">
{{ flashMessage }}
</div>
{% endfor %}
Even if the user is not logged in and even if you haven't created any flash messages, just calling the
get() (or even has()) method of the flashBag will start a session. This may hurt your application
performance because all users will receive a session cookie. To avoid this behavior, add a check before
trying to access the flash messages:
Listing 129-2
1
2
3
4
5
6
7
{% if app.request.hasPreviousSession %}
{% for flashMessage in app.session.flashBag.get('notice') %}
<div class="flash-notice">
{{ flashMessage }}
</div>
{% endfor %}
{% endif %}

Chapter 129: Avoid Starting Sessions for Anonymous Users | 435
Chapter 130
How Symfony2 Differs from Symfony1

The Symfony2 Framework embodies a significant evolution when compared with the first version of the
framework. Fortunately, with the MVC architecture at its core, the skills used to master a symfony1
project continue to be very relevant when developing in Symfony2. Sure, app.yml is gone, but routing,
controllers and templates all remain.
This chapter walks through the differences between symfony1 and Symfony2. As you'll see, many tasks
are tackled in a slightly different way. You'll come to appreciate these minor differences as they promote
stable, predictable, testable and decoupled code in your Symfony2 applications.
So, sit back and relax as you travel from "then" to "now".
Directory Structure
When looking at a Symfony2 project - for example, the Symfony Standard Edition1 - you'll notice a very
different directory structure than in symfony1. The differences, however, are somewhat superficial.
The app/ Directory

In symfony1, your project has one or more applications, and each lives inside the apps/ directory (e.g.
apps/frontend). By default in Symfony2, you have just one application represented by the app/
directory. Like in symfony1, the app/ directory contains configuration specific to that application. It also
contains application-specific cache, log and template directories as well as a Kernel class (AppKernel),
which is the base object that represents the application.
Unlike symfony1, almost no PHP code lives in the app/ directory. This directory is not meant to house
modules or library files as it did in symfony1. Instead, it's simply the home of configuration and other
resources (templates, translation files).

Chapter 130: How Symfony2 Differs from Symfony1 | 436
The src/ Directory

Put simply, your actual code goes here. In Symfony2, all actual application-code lives inside a bundle
(roughly equivalent to a symfony1 plugin) and, by default, each bundle lives inside the src directory.
In that way, the src directory is a bit like the plugins directory in symfony1, but much more flexible.
Additionally, while your bundles will live in the src/ directory, third-party bundles will live somewhere
in the vendor/ directory.
To get a better picture of the src/ directory, first think of the structure of a symfony1 application. First,
part of your code likely lives inside one or more applications. Most commonly these include modules,
but could also include any other PHP classes you put in your application. You may have also created
a schema.yml file in the config directory of your project and built several model files. Finally, to
help with some common functionality, you're using several third-party plugins that live in the plugins/
directory. In other words, the code that drives your application lives in many different places.
In Symfony2, life is much simpler because all Symfony2 code must live in a bundle. In the pretend
symfony1 project, all the code could be moved into one or more plugins (which is a very good practice,
in fact). Assuming that all modules, PHP classes, schema, routing configuration, etc. were moved into a
plugin, the symfony1 plugins/ directory would be very similar to the Symfony2 src/ directory.
Put simply again, the src/ directory is where your code, assets, templates and most anything else specific
to your project will live.
The vendor/ Directory

The vendor/ directory is basically equivalent to the lib/vendor/ directory in symfony1, which was
the conventional directory for all vendor libraries and bundles. By default, you'll find the Symfony2
library files in this directory, along with several other dependent libraries such as Doctrine2, Twig and
Swift Mailer. 3rd party Symfony2 bundles live somewhere in the vendor/.
The web/ Directory

Not much has changed in the web/ directory. The most noticeable difference is the absence of the css/,
js/ and images/ directories. This is intentional. Like with your PHP code, all assets should also live
inside a bundle. With the help of a console command, the Resources/public/ directory of each
bundle is copied or symbolically-linked to the web/bundles/ directory. This allows you to keep assets
organized inside your bundle, but still make them available to the public. To make sure that all bundles
are available, run the following command:
Listing 130-1
$ php app/console assets:install web
This command is the Symfony2 equivalent to the symfony1 plugin:publish-assets command.
Autoloading
One of the advantages of modern frameworks is never needing to worry about requiring files. By making
use of an autoloader, you can refer to any class in your project and trust that it's available. Autoloading
has changed in Symfony2 to be more universal, faster, and independent of needing to clear your cache.

In symfony1, autoloading was done by searching the entire project for the presence of PHP class files
and caching this information in a giant array. That array told symfony1 exactly which file contained each
class. In the production environment, this caused you to need to clear the cache when classes were added
or moved.
In Symfony2, a tool named Composer2 handles this process. The idea behind the autoloader is simple:
the name of your class (including the namespace) must match up with the path to the file containing that
class. Take the FrameworkExtraBundle from the Symfony2 Standard Edition as an example:
Listing 130-2
1
2
3
4
5
6
7
8
9
namespace Sensio\Bundle\FrameworkExtraBundle;
// ...
class SensioFrameworkExtraBundle extends Bundle
{
// ...
}
vendor/sensio/framework-extra-bundle/Sensio/Bundle/
FrameworkExtraBundle/SensioFrameworkExtraBundle.php. As you can see, the second part
The
file
itself
lives
at
of the path follows the namespace of the class. The first part is equal to the package name of the
SensioFrameworkExtraBundle.
The namespace, Sensio\Bundle\FrameworkExtraBundle, and package name, sensio/
framework-extra-bundle, spells out the directory that the file should live in (vendor/sensio/
framework-extra-bundle/Sensio/Bundle/FrameworkExtraBundle/). Composer can then
look for the file at this specific place and load it very fast.
the file did not live at this exact location, you'd receive a Class
"Sensio\Bundle\FrameworkExtraBundle\SensioFrameworkExtraBundle" does not
exist. error. In Symfony2, a "class does not exist" error means that the namespace of the class and
If
physical location do not match. Basically, Symfony2 is looking in one exact location for that class, but
that location doesn't exist (or contains a different class). In order for a class to be autoloaded, you never
need to clear your cache in Symfony2.
As mentioned before, for the autoloader to work, it needs to know that the Sensio namespace lives
in the vendor/sensio/framework-extra-bundle directory and that, for example, the Doctrine
namespace lives in the vendor/doctrine/orm/lib/ directory. This mapping is entirely controlled by
Composer. Each third-party library you load through Composer has its settings defined and Composer
takes care of everything for you.
For this to work, all third-party libraries used by your project must be defined in the composer.json
file.
If you look at the HelloController from the Symfony Standard Edition you can see that it lives in
the Acme\DemoBundle\Controller namespace. Yet, the AcmeDemoBundle is not defined in your
composer.json file. Nonetheless are the files autoloaded. This is because you can tell Composer to
autoload files from specific directories without defining a dependency:
Listing 130-3
1
2
3
4
5
{
"autoload": {
"psr-0": { "": "src/" }
}
}
2. https://getcomposer.org

This means that if a class is not found in the vendor directory, Composer will search in the src
directory before throwing a "class does not exist" exception. Read more about configuring the Composer
autoloader in the Composer documentation3.
Using the Console

In symfony1, the console is in the root directory of your project and is called symfony:
Listing 130-4
$ php symfony
In Symfony2, the console is now in the app sub-directory and is called console:
Listing 130-5
$ php app/console
Applications
In a symfony1 project, it is common to have several applications: one for the front-end and one for the
back-end for instance.
In a Symfony2 project, you only need to create one application (a blog application, an intranet
application, ...). Most of the time, if you want to create a second application, you might instead create
another project and share some bundles between them.
And if you need to separate the front-end and the back-end features of some bundles, you can create
sub-namespaces for controllers, sub-directories for templates, different semantic configurations, separate
routing configurations, and so on.
Of course, there's nothing wrong with having multiple applications in your project, that's entirely up to
you. A second application would mean a new directory, e.g. my_app/, with the same basic setup as the
app/ directory.
Bundles and Plugins

In a symfony1 project, a plugin could contain configuration, modules, PHP libraries, assets and anything
else related to your project. In Symfony2, the idea of a plugin is replaced by the "bundle". A bundle is
even more powerful than a plugin because the core Symfony2 Framework is brought in via a series of
bundles. In Symfony2, bundles are first-class citizens that are so flexible that even core code itself is a
bundle.
In symfony1, a plugin must be enabled inside the ProjectConfiguration class:
Listing 130-6
// config/ProjectConfiguration.class.php
public function setup()
{
// some plugins here
$this->enableAllPluginsExcept(array(...));
}
1
2
3
4
5
6
In Symfony2, the bundles are activated inside the application kernel:

Listing 130-7
1
2
3. https://getcomposer.org/doc/04-schema.md#autoload

3
4
5
6
7
8
9
10
11
12
{
$bundles = array(
new Symfony\Bundle\FrameworkBundle\FrameworkBundle(),
new Symfony\Bundle\TwigBundle\TwigBundle(),
...,
new Acme\DemoBundle\AcmeDemoBundle(),
);
return $bundles;
}
Routing (routing.yml) and Configuration (config.yml)

In symfony1, the routing.yml and app.yml configuration files were automatically loaded inside any
plugin. In Symfony2, routing and application configuration inside a bundle must be included manually.
For example, to include a routing resource from a bundle called AcmeDemoBundle, you can do the
following:
1
2
3
Listing 130-8
_hello:
resource: '@AcmeDemoBundle/Resources/config/routing.yml'
This will load the routes found in the Resources/config/routing.yml file of the
AcmeDemoBundle. The special @AcmeDemoBundle is a shortcut syntax that, internally, resolves to the
full path to that bundle.
You can use this same strategy to bring in configuration from a bundle:
1
2
3
Listing 130-9
imports:
- { resource: "@AcmeDemoBundle/Resources/config/config.yml" }
In Symfony2, configuration is a bit like app.yml in symfony1, except much more systematic. With
app.yml, you could simply create any keys you wanted. By default, these entries were meaningless and
depended entirely on how you used them in your application:
Listing 130-10 1
2
3
4
# some app.yml file from symfony1

all:
email:
from_address: '[email protected]'
In Symfony2, you can also create arbitrary entries under the parameters key of your configuration:
Listing 130-11 1
parameters:
email.from_address: '[email protected]'
You can now access this from a controller, for example:

public function helloAction($name)
{
$fromAddress = $this->container->getParameter('email.from_address');
}
Listing 130-12
In reality, the Symfony2 configuration is much more powerful and is used primarily to configure objects
that you can use. For more information, see the chapter titled "Service Container".

Chapter 131
How to Inject Variables into all Templates (i.e.

global Variables)
Sometimes you want a variable to be accessible to all the templates you use. This is possible inside your
app/config/config.yml file:
Listing 131-1
1
2
3
4
5
twig:
# ...
globals:
ga_tracking: UA-xxxxx-x
Now, the variable ga_tracking is available in all Twig templates:

Listing 131-2
<p>The google tracking code is: {{ ga_tracking }}</p>
It's that easy!
Using Service Container Parameters

You can also take advantage of the built-in Service Parameters system, which lets you isolate or reuse the
value:
Listing 131-3
1
2
3
parameters:
ga_tracking: UA-xxxxx-x
Listing 131-4
1
2
3
4
twig:
globals:
ga_tracking: '%ga_tracking%'
The same variable is available exactly as before.

Chapter 131: How to Inject Variables into all Templates (i.e. global Variables) | 441
Referencing Services
Instead of using static values, you can also set the value to a service. Whenever the global variable is
accessed in the template, the service will be requested from the service container and you get access to
that object.
The service is not loaded lazily. In other words, as soon as Twig is loaded, your service is instantiated,
even if you never use that global variable.
To define a service as a global Twig variable, prefix the string with @. This should feel familiar, as it's the
same syntax you use in service configuration.
Listing 131-5
1
2
3
4
5
twig:
# ...
globals:
user_management: '@app.user_management'
Using a Twig Extension

If the global variable you want to set is more complicated - say an object - then you won't be able to use
the above method. Instead, you'll need to create a Twig Extension and return the global variable as one
of the entries in the getGlobals method.

Chapter 131: How to Inject Variables into all Templates (i.e. global Variables) | 442
Chapter 132
How to Use and Register Namespaced Twig

Paths
Usually, when you refer to a template, you'll use the MyBundle:Subdir:filename.html.twig
format (see Template Naming and Locations).
Twig also natively offers a feature called "namespaced paths", and support is built-in automatically for all
of your bundles.
Take the following paths as an example:
Listing 132-1
1
2
{% extends "AppBundle::layout.html.twig" %}
{{ include('AppBundle:Foo:bar.html.twig') }}
With namespaced paths, the following works as well:

Listing 132-2
1
2
{% extends "@App/layout.html.twig" %}
{{ include('@App/Foo/bar.html.twig') }}
Both paths are valid and functional by default in Symfony.

As an added bonus, the namespaced syntax is faster.
Registering your own Namespaces

You can also register your own custom namespaces. Suppose that you're using some third-party library
that includes Twig templates that live in vendor/acme/foo-bar/templates. First, register a
namespace for this directory:
Listing 132-3
1
2
3
twig:
# ...

Chapter 132: How to Use and Register Namespaced Twig Paths | 443
4
5
paths:
"%kernel.root_dir%/../vendor/acme/foo-bar/templates": foo_bar
Prior to 2.8, templates in custom namespaces are not pre-compiled by Symfony's cache warmup
process. They are compiled on demand. This may cause problems if two simultaneous requests are
trying to use the template for the first time.
The registered namespace is called foo_bar, which refers to the vendor/acme/foo-bar/

templates directory. Assuming there's a file called sidebar.twig in that directory, you can use it
easily:
Listing 132-4
{{ include('@foo_bar/sidebar.twig') }}
Multiple Paths per Namespace

You can also assign several paths to the same template namespace. The order in which paths are
configured is very important, because Twig will always load the first template that exists, starting from
the first configured path. This feature can be used as a fallback mechanism to load generic templates
when the specific template doesn't exist.
Listing 132-5
1
2
3
4
5
6
7
twig:
# ...
paths:
"%kernel.root_dir%/../vendor/acme/themes/theme1": theme
"%kernel.root_dir%/../vendor/acme/themes/theme2": theme
"%kernel.root_dir%/../vendor/acme/themes/common": theme
Now, you can use the same @theme namespace to refer to any template located in the previous three
directories:
Listing 132-6
{{ include('@theme/header.twig') }}

Chapter 132: How to Use and Register Namespaced Twig Paths | 444
Chapter 133
How to Use PHP instead of Twig for Templates

Symfony defaults to Twig for its template engine, but you can still use plain PHP code if you want. Both
templating engines are supported equally in Symfony. Symfony adds some nice features on top of PHP to
make writing templates with PHP more powerful.
Rendering PHP Templates

If you want to use the PHP templating engine, first, make sure to enable it in your application
configuration file:
Listing 133-1
framework:
# ...
templating:
engines: ['twig', 'php']
1
2
3
4
5
You can now render a PHP template instead of a Twig one simply by using the .php extension in the
template name instead of .twig. The controller below renders the index.html.php template:
Listing 133-2
1
2
3
4
5
6
7
8
9
10
// ...
{
'AppBundle:Hello:index.html.php',
);
}
@Template1
AppBundle:Hello:index.html.php template:
You
Listing 133-3
1
2
can
also
use
the
shortcut
to
render
the
default
use Sensio\Bundle\FrameworkExtraBundle\Configuration\Template;
1. https://symfony.com/doc/current/bundles/SensioFrameworkExtraBundle/annotations/view

Chapter 133: How to Use PHP instead of Twig for Templates | 445
3
4
5
6
7
8
9
10
11
12
// ...
/**
* @Template(engine="php")
*/
{
return array('name' => $name);
}
Enabling the php and twig template engines simultaneously is allowed, but it will produce an
undesirable side effect in your application: the @ notation for Twig namespaces will no longer be
supported for the render() method:
Listing 133-4
Listing 133-5
1
2
3
4
5
6
7
8
9
10
1
2
3
4
5

{
// ...
// namespaced templates will no longer work in controllers

$this->render('@App/Default/index.html.twig');
// you must use the traditional template notation
$this->render('AppBundle:Default:index.html.twig');
}
{# inside a Twig template, namespaced templates work as expected #}

{{ include('@App/Default/index.html.twig') }}
{# traditional template notation will also work #}
{{ include('AppBundle:Default:index.html.twig') }}
Decorating Templates
More often than not, templates in a project share common elements, like the well-known header and
footer. In Symfony, this problem is thought about differently: a template can be decorated by another
one.
The index.html.php template is decorated by layout.html.php, thanks to the extend() call:
Listing 133-6
1
2
3
4


<?php $view->extend('AppBundle::layout.html.php') ?>
Hello <?php echo $name ?>!
The AppBundle::layout.html.php notation sounds familiar, doesn't it? It is the same notation
used to reference a template. The :: part simply means that the controller element is empty, so the
corresponding file is directly stored under views/.
Now, have a look at the layout.html.php file:
Listing 133-7
1
2
3
4
5
6


<?php $view->extend('::base.html.php') ?>
<h1>Hello Application</h1>
<?php $view['slots']->output('_content') ?>

The layout is itself decorated by another one (::base.html.php). Symfony supports multiple
decoration levels: a layout can itself be decorated by another one. When the bundle part of the template
name is empty, views are looked for in the app/Resources/views/ directory. This directory stores
global views for your entire project:
Listing 133-8
1
2
3
4
5
6
7
8
9
10
11


<!DOCTYPE html>
<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
<title><?php $view['slots']->output('title', 'Hello Application') ?></title>
</head>
<body>
<?php $view['slots']->output('_content') ?>
</body>
</html>
For both layouts, the $view['slots']->output('_content') expression is replaced by the

content of the child template, index.html.php and layout.html.php respectively (more on slots
in the next section).
As you can see, Symfony provides methods on a mysterious $view object. In a template, the $view
variable is always available and refers to a special object that provides a bunch of methods that makes the
template engine tick.
Working with Slots

A slot is a snippet of code, defined in a template, and reusable in any layout decorating the template. In
the index.html.php template, define a title slot:
Listing 133-9
1
2
3
4
5
6

<?php $view['slots']->set('title', 'Hello World Application') ?>
The base layout already has the code to output the title in the header:
Listing 133-10 1
2
3
4
5


<head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
<title><?php $view['slots']->output('title', 'Hello Application') ?></title>
</head>
The output() method inserts the content of a slot and optionally takes a default value if the slot is not
defined. And _content is just a special slot that contains the rendered child template.
For large slots, there is also an extended syntax:
Listing 133-11 1
2
3
<?php $view['slots']->start('title') ?>

Some large amount of HTML
<?php $view['slots']->stop() ?>
Including other Templates

The best way to share a snippet of template code is to define a template that can then be included into
other templates.
Create a hello.html.php template:

Listing 133-12 1


And change the index.html.php template to include it:

Listing 133-13 1
2
3
4

<?php echo $view->render('AppBundle:Hello:hello.html.php', array('name' => $name)) ?>
The render() method evaluates and returns the content of another template (this is the exact same
method as the one used in the controller).
Embedding other Controllers

And what if you want to embed the result of another controller in a template? That's very useful when
working with Ajax, or when the embedded template needs some variable not available in the main
template.
If you create a fancy action, and want to include it into the index.html.php template, simply use the
following code:
Listing 133-14 1
2
3
4
5
6
7

<?php echo $view['actions']->render(
new \Symfony\Component\HttpKernel\Controller\ControllerReference('AppBundle:Hello:fancy', array(
'name' => $name,
'color' => 'green',
))
) ?>
Here, the AppBundle:Hello:fancy string refers to the fancy action of the Hello controller:
Listing 133-15
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
class HelloController extends Controller
{
public function fancyAction($name, $color)
{
// create some object, based on the $color variable
$object = ...;
return $this->render('AppBundle:Hello:fancy.html.php', array(
'name'
=> $name,
'object' => $object
));
}
// ...
}
But where is the $view['actions'] array element defined? Like $view['slots'], it's called a
template helper, and the next section tells you more about those.
Using Template Helpers

The Symfony templating system can be easily extended via helpers. Helpers are PHP objects that provide
features useful in a template context. actions and slots are two of the built-in Symfony helpers.
Creating Links between Pages

Speaking of web applications, creating links between pages is a must. Instead of hardcoding URLs in
templates, the router helper knows how to generate URLs based on the routing configuration. That
way, all your URLs can be easily updated by changing the configuration:
Listing 133-16 1
2
3
<a href="<?php echo $view['router']->generate('hello', array('name' => 'Thomas')) ?>">

Greet Thomas!
</a>
The generate() method takes the route name and an array of parameters as arguments. The route
name is the main key under which routes are referenced and the parameters are the values of the
placeholders defined in the route pattern:
Listing 133-17 1
2
3
4
# src/AppBundle/Resources/config/routing.yml
hello: # The route name
path:
/hello/{name}
defaults: { _controller: AppBundle:Hello:index }
Using Assets: Images, JavaScripts and Stylesheets

What would the Internet be without images, JavaScripts, and stylesheets? Symfony provides the assets
tag to deal with them easily:
Listing 133-18 1
2
3
<link href="<?php echo $view['assets']->getUrl('css/blog.css') ?>" rel="stylesheet" type="text/css" />

<img src="<?php echo $view['assets']->getUrl('images/logo.png') ?>" />
The assets helper's main purpose is to make your application more portable. Thanks to this helper,
you can move the application root directory anywhere under your web root directory without changing
anything in your template's code.
Profiling Templates
By using the stopwatch helper, you are able to time parts of your template and display it on the timeline
of the WebProfilerBundle:
<?php $view['stopwatch']->start('foo') ?>
... things that get timed
<?php $view['stopwatch']->stop('foo') ?>
Listing 133-19
If you use the same name more than once in your template, the times are grouped on the same line
in the timeline.
Output Escaping
When using PHP templates, escape variables whenever they are displayed to the user:
<?php echo $view->escape($var) ?>
Listing 133-20
By default, the escape() method assumes that the variable is outputted within an HTML context. The
second argument lets you change the context. For instance, to output something in a JavaScript script,
use the js context:
<?php echo $view->escape($var, 'js') ?>
Listing 133-21

Chapter 134
How to Write a custom Twig Extension

The main motivation for writing an extension is to move often used code into a reusable class like adding
support for internationalization. An extension can define tags, filters, tests, operators, global variables,
functions, and node visitors.
Creating an extension also makes for a better separation of code that is executed at compilation time and
code needed at runtime. As such, it makes your code faster.
Before writing your own extensions, have a look at the Twig official extension repository1.
Create the Extension Class

This cookbook describes how to write a custom Twig extension as of Twig 1.12. If you are using an
older version, please read Twig extensions documentation legacy2.
To get your custom functionality you must first create a Twig Extension class. As an example you'll create
a price filter to format a given number into price:
Listing 134-1
1
2
3
4
5
6
7
8
9
10
11
12
// src/AppBundle/Twig/AppExtension.php
namespace AppBundle\Twig;
class AppExtension extends \Twig_Extension
{
public function getFilters()
{
return array(
new \Twig_SimpleFilter('price', array($this, 'priceFilter')),
);
}
1. https://github.com/twigphp/Twig-extensions
2. http://twig.sensiolabs.org/doc/advanced_legacy.html#creating-an-extension

Chapter 134: How to Write a custom Twig Extension | 450
13
14
15
16
17
18
19
20
21
22
23
24
25
public function priceFilter($number, $decimals = 0, $decPoint = '.', $thousandsSep = ',')

{
$price = number_format($number, $decimals, $decPoint, $thousandsSep);
$price = '$'.$price;
return $price;
}
{
return 'app_extension';
}
}
Along with custom filters, you can also add custom functions and register global variables.
Register an Extension as a Service

Now you must let the Service Container know about your newly created Twig Extension:
Listing 134-2
1
2
3
4
5
6
7
services:
app.twig_extension:
class: AppBundle\Twig\AppExtension
public: false
tags:
- { name: twig.extension }
Using the custom Extension

Using your newly created Twig Extension is no different than any other:
Listing 134-3
1
2
{# outputs $5,500.00 #}
{{ '5500'|price }}
Passing other arguments to your filter:

Listing 134-4
1
2
{# outputs $5500,2516 #}
{{ '5500.25155'|price(4, ',', '') }}
Learning further
For a more in-depth look into Twig Extensions, please take a look at the Twig extensions documentation3.
3. http://twig.sensiolabs.org/doc/advanced.html#creating-an-extension

Chapter 134: How to Write a custom Twig Extension | 451
Chapter 135
How to Render a Template without a custom

Controller
Usually, when you need to create a page, you need to create a controller and render a template from
within that controller. But if you're rendering a simple template that doesn't need any data passed
into it, you can avoid creating the controller entirely, by using the built-in
FrameworkBundle:Template:template controller.
For example, suppose you want to render a static/privacy.html.twig template, which doesn't
require that any variables are passed to it. You can do this without creating a controller:
Listing 135-1
1
2
3
4
5
acme_privacy:
path: /privacy
defaults:
_controller: FrameworkBundle:Template:template
template:
static/privacy.html.twig
The FrameworkBundle:Template:template controller will simply render whatever template

you've passed as the template default value.
You can of course also use this trick when rendering embedded controllers from within a template. But
since the purpose of rendering a controller from within a template is typically to prepare some data in
a custom controller, this is probably only useful if you'd like to cache this page partial (see Caching the
static Template).
Listing 135-2
{{ render(url('acme_privacy')) }}
Caching the static Template

Since templates that are rendered in this way are typically static, it might make sense to cache them.
Fortunately, this is easy! By configuring a few other variables in your route, you can control exactly how
your page is cached:
Listing 135-3

Chapter 135: How to Render a Template without a custom Controller | 452
1
2
3
4
5
6
7
acme_privacy:
path: /privacy
defaults:
_controller:
template:
maxAge:
sharedAge:
FrameworkBundle:Template:template
'static/privacy.html.twig'
86400
86400
The maxAge and sharedAge values are used to modify the Response object created in the controller.
For more information on caching, see HTTP Cache.
There is also a private variable (not shown here). By default, the Response will be made public, as long
as maxAge or sharedAge are passed. If set to true, the Response will be marked as private.

Chapter 135: How to Render a Template without a custom Controller | 453
Chapter 136
How to Simulate HTTP Authentication in a

Functional Test
If your application needs HTTP authentication, pass the username and password as server variables to
createClient():
Listing 136-1
$client = static::createClient(array(), array(

'PHP_AUTH_USER' => 'username',
'PHP_AUTH_PW' => 'pa$$word',
));
You can also override it on a per request basis:

Listing 136-2
$client->request('DELETE', '/post/12', array(), array(), array(

'PHP_AUTH_USER' => 'username',
'PHP_AUTH_PW' => 'pa$$word',
));
When your application is using a form_login, you can simplify your tests by allowing your test
configuration to make use of HTTP authentication. This way you can use the above to authenticate
in tests, but still have your users log in via the normal form_login. The trick is to include the
http_basic key in your firewall, along with the form_login key:
Listing 136-3
1
2
3
4
5
security:
firewalls:
your_firewall_name:
http_basic: ~

Chapter 136: How to Simulate HTTP Authentication in a Functional Test | 454
Chapter 137
How to Simulate Authentication with a Token

in a Functional Test
Authenticating requests in functional tests might slow down the suite. It could become an issue especially
when form_login is used, since it requires additional requests to fill in and submit the form.
One of the solutions is to configure your firewall to use http_basic in the test environment as
explained in How to Simulate HTTP Authentication in a Functional Test. Another way would be to create
a token yourself and store it in a session. While doing this, you have to make sure that an appropriate
cookie is sent with a request. The following example demonstrates this technique:
Listing 137-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
// src/AppBundle/Tests/Controller/DefaultControllerTest.php
namespace Appbundle\Tests\Controller;
use Symfony\Bundle\FrameworkBundle\Test\WebTestCase;
use Symfony\Component\BrowserKit\Cookie;
use Symfony\Component\Security\Core\Authentication\Token\UsernamePasswordToken;
class DefaultControllerTest extends WebTestCase
{
private $client = null;
public function setUp()
{
$this->client = static::createClient();
}
public function testSecuredHello()
{
$this->logIn();
$crawler = $this->client->request('GET', '/admin');
$this->assertTrue($this->client->getResponse()->isSuccessful());
$this->assertGreaterThan(0, $crawler->filter('html:contains("Admin Dashboard")')->count());
}
private function logIn()
{
$session = $this->client->getContainer()->get('session');

Chapter 137: How to Simulate Authentication with a Token in a Functional Test | 455
31
32
33
34
35
36
37
38
39
40
41
// the firewall context (defaults to the firewall name)

$firewall = 'secured_area';
$token = new UsernamePasswordToken('admin', null, $firewall, array('ROLE_ADMIN'));
$session->set('_security_'.$firewall, serialize($token));
$session->save();
$cookie = new Cookie($session->getName(), $session->getId());
$this->client->getCookieJar()->set($cookie);
}
}
The technique described in How to Simulate HTTP Authentication in a Functional Test is cleaner and
therefore the preferred way.

Chapter 137: How to Simulate Authentication with a Token in a Functional Test | 456
Chapter 138
How to Test the Interaction of several Clients

If you need to simulate an interaction between different clients (think of a chat for instance), create
several clients:
Listing 138-1
1
2
3
4
5
6
7
8
9
10
// ...
$harry = static::createClient();
$sally = static::createClient();
$harry->request('POST', '/say/sally/Hello');
$sally->request('GET', '/messages');
$this->assertEquals(Response::HTTP_CREATED, $harry->getResponse()->getStatusCode());
$this->assertRegExp('/Hello/', $sally->getResponse()->getContent());
This works except when your code maintains a global state or if it depends on a third-party library that
has some kind of global state. In such a case, you can insulate your clients:
Listing 138-2
1
2
3
4
5
6
7
8
9
10
11
12
13
// ...
$harry = static::createClient();
$sally = static::createClient();
$harry->insulate();
$sally->insulate();
$harry->request('POST', '/say/sally/Hello');
$sally->request('GET', '/messages');
$this->assertEquals(Response::HTTP_CREATED, $harry->getResponse()->getStatusCode());
$this->assertRegExp('/Hello/', $sally->getResponse()->getContent());
Insulated clients transparently execute their requests in a dedicated and clean PHP process, thus avoiding
any side-effects.
As an insulated client is slower, you can keep one client in the main process, and insulate the other
ones.

Chapter 138: How to Test the Interaction of several Clients | 457
Chapter 139
How to Use the Profiler in a Functional Test

It's highly recommended that a functional test only tests the Response. But if you write functional tests
that monitor your production servers, you might want to write tests on the profiling data as it gives you
a great way to check various things and enforce some metrics.
The Symfony Profiler gathers a lot of data for each request. Use this data to check the number of database
calls, the time spent in the framework, etc. But before writing assertions, enable the profiler and check
that the profiler is indeed available (it is enabled by default in the test environment):
Listing 139-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
class HelloControllerTest extends WebTestCase

{
public function testIndex()
{
$client = static::createClient();
// Enable the profiler for the next request

// (it does nothing if the profiler is not available)
$client->enableProfiler();
$crawler = $client->request('GET', '/hello/Fabien');
// ... write some assertions about the Response

// Check that the profiler is enabled
if ($profile = $client->getProfile()) {
// check the number of requests
$this->assertLessThan(
10,
$profile->getCollector('db')->getQueryCount()
);
// check the time spent in the framework
500,
$profile->getCollector('time')->getDuration()
);
}
}
}
If a test fails because of profiling data (too many DB queries for instance), you might want to use the Web
Profiler to analyze the request after the tests finish. It's easy to achieve if you embed the token in the error
message:
Chapter 139: How to Use the Profiler in a Functional Test | 458
Listing 139-2
1
2
3
4
5
6
7
8
30,
$profile->getCollector('db')->getQueryCount(),
sprintf(
'Checks that query count is less than 30 (token %s)',
$profile->getToken()
)
);
The profiler store can be different depending on the environment (especially if you use the SQLite
store, which is the default configured one).
The profiler information is available even if you insulate the client or if you use an HTTP layer for
your tests.
Read the API for built-in data collectors to learn more about their interfaces.
Speeding up Tests by not Collecting Profiler Data

To avoid collecting data in each test you can set the collect parameter to false:
Listing 139-3
1
2
3
4
5
6
7
# ...
framework:
profiler:
enabled: true
collect: false
In this way only tests that call $client->enableProfiler() will collect data.

Chapter 139: How to Use the Profiler in a Functional Test | 459
Chapter 140
How to Test Code that Interacts with the

Database
If your code interacts with the database, e.g. reads data from or stores data into it, you need to adjust
your tests to take this into account. There are many ways how to deal with this. In a unit test, you can
create a mock for a Repository and use it to return expected objects. In a functional test, you may need
to prepare a test database with predefined values to ensure that your test always has the same data to
work with.
If you want to test your queries directly, see How to Test Doctrine Repositories.
Mocking the Repository in a Unit Test

If you want to test code which depends on a Doctrine repository in isolation, you need to mock the
Repository. Normally you inject the EntityManager into your class and use it to get the repository.
This makes things a little more difficult as you need to mock both the EntityManager and your
repository class.
It is possible (and a good idea) to inject your repository directly by registering your repository as a
factory service. This is a little bit more work to setup, but makes testing easier as you only need to
mock the repository.
Suppose the class you want to test looks like this:

Listing 140-1
1
2
3
4
5
6
namespace AppBundle\Salary;
use Doctrine\Common\Persistence\ObjectManager;
class SalaryCalculator
{

Chapter 140: How to Test Code that Interacts with the Database | 460
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
private $entityManager;
public function __construct(ObjectManager $entityManager)
{
$this->entityManager = $entityManager;
}
public function calculateTotalSalary($id)
{
$employeeRepository = $this->entityManager
->getRepository('AppBundle:Employee');
$employee = $employeeRepository->find($id);
return $employee->getSalary() + $employee->getBonus();
}
}
Since the ObjectManager gets injected into the class through the constructor, it's easy to pass a mock
object within a test:
Listing 140-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
use AppBundle\Salary\SalaryCalculator;
class SalaryCalculatorTest extends \PHPUnit_Framework_TestCase
{
public function testCalculateTotalSalary()
{
// First, mock the object to be used in the test
$employee = $this->getMock('\AppBundle\Entity\Employee');
$employee->expects($this->once())
->method('getSalary')
->will($this->returnValue(1000));
$employee->expects($this->once())
->method('getBonus')
->will($this->returnValue(1100));
// Now, mock the repository so it returns the mock of the employee

$employeeRepository = $this
->getMockBuilder('\Doctrine\ORM\EntityRepository')
->disableOriginalConstructor()
->getMock();
$employeeRepository->expects($this->once())
->method('find')
->will($this->returnValue($employee));
// Last, mock the EntityManager to return the mock of the repository
$entityManager = $this
->getMockBuilder('\Doctrine\Common\Persistence\ObjectManager')
->disableOriginalConstructor()
->getMock();
$entityManager->expects($this->once())
->method('getRepository')
->will($this->returnValue($employeeRepository));
$salaryCalculator = new SalaryCalculator($entityManager);
$this->assertEquals(2100, $salaryCalculator->calculateTotalSalary(1));
}
}
In this example, you are building the mocks from the inside out, first creating the employee which gets
returned by the Repository, which itself gets returned by the EntityManager. This way, no real class
is involved in testing.

Changing Database Settings for Functional Tests

If you have functional tests, you want them to interact with a real database. Most of the time you want
to use a dedicated database connection to make sure not to overwrite data you entered when developing
the application and also to be able to clear the database before every test.
To do this, you can specify a database configuration which overwrites the default configuration:
Listing 140-3
1
2
3
4
5
6
7
8
doctrine:
# ...
dbal:
host:
localhost
dbname:
testdb
user:
testdb
password: testdb
Make sure that your database runs on localhost and has the defined database and user credentials set up.

Chapter 141
How to Test Doctrine Repositories

Unit testing Doctrine repositories in a Symfony project is not recommended. When you're dealing with
a repository, you're really dealing with something that's meant to be tested against a real database
connection.
Fortunately, you can easily test your queries against a real database, as described below.
Functional Testing
If you need to actually execute a query, you will need to boot the kernel to get a valid connection. In this
case, you'll extend the KernelTestCase, which makes all of this quite easy:
Listing 141-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
// src/AppBundle/Tests/Entity/ProductRepositoryFunctionalTest.php
namespace AppBundle\Tests\Entity;
use Symfony\Bundle\FrameworkBundle\Test\KernelTestCase;
class ProductRepositoryFunctionalTest extends KernelTestCase
{
/**
* @var \Doctrine\ORM\EntityManager
*/
private $em;
/**
* {@inheritDoc}
*/
protected function setUp()
{
self::bootKernel();
$this->em = static::$kernel->getContainer()
->get('doctrine')
->getManager()
;
}
public function testSearchByCategoryName()
{
$products = $this->em
->getRepository('AppBundle:Product')

Chapter 141: How to Test Doctrine Repositories | 463
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
->searchByCategoryName('foo')
;
$this->assertCount(1, $products);
}
/**
* {@inheritDoc}
*/
protected function tearDown()
{
parent::tearDown();
$this->em->close();
$this->em = null; // avoid memory leaks
}
}

Chapter 141: How to Test Doctrine Repositories | 464
Chapter 142
How to Customize the Bootstrap Process

before Running Tests
Sometimes when running tests, you need to do additional bootstrap work before running those tests. For
example, if you're running a functional test and have introduced a new translation resource, then you will
need to clear your cache before running those tests. This cookbook covers how to do that.
First, add the following file:
Listing 142-1
1
2
3
4
5
6
7
8
9
10
// app/tests.bootstrap.php
if (isset($_ENV['BOOTSTRAP_CLEAR_CACHE_ENV'])) {
passthru(sprintf(
'php "%s/console" cache:clear --env=%s --no-warmup',
__DIR__,
$_ENV['BOOTSTRAP_CLEAR_CACHE_ENV']
));
}
require __DIR__.'/bootstrap.php.cache';
Replace the test bootstrap file bootstrap.php.cache in app/phpunit.xml.dist with

tests.bootstrap.php:
Listing 142-2
1
2
3
4
5
6



<phpunit
bootstrap = "tests.bootstrap.php"
>
Now, you can define in your phpunit.xml.dist file which environment you want the cache to be
cleared:
Listing 142-3
1
2
3
4


<php>
<env name="BOOTSTRAP_CLEAR_CACHE_ENV" value="test"/>
</php>

Chapter 142: How to Customize the Bootstrap Process before Running Tests | 465
This now becomes an environment variable (i.e. $_ENV) that's available in the custom bootstrap file
(tests.bootstrap.php).

Chapter 142: How to Customize the Bootstrap Process before Running Tests | 466
Chapter 143
Upgrading a Patch Version (e.g. 2.6.0 to 2.6.1)

When a new patch version is released (only the last number changed), it is a release that only contains
bug fixes. This means that upgrading to a new patch version is really easy:
Listing 143-1
That's it! You should not encounter any backwards-compatibility breaks or need to change anything else
in your code. That's because when you started your project, your composer.json included Symfony
using a constraint like 2.6.*, where only the last version number will change when you update.
It is recommended to update to a new patch version as soon as possible, as important bugs and
security leaks may be fixed in these new releases.
Upgrading other Packages

You may also want to upgrade the rest of your libraries. If you've done a good job with your version
constraints1 in composer.json, you can do this safely by running:
Listing 143-2
$ composer update
Beware, if you have some unspecific version constraints2 in your composer.json (e.g. devmaster), this could upgrade some non-Symfony libraries to new versions that contain backwardscompatibility breaking changes.
1. https://getcomposer.org/doc/01-basic-usage.md#package-versions

Chapter 143: Upgrading a Patch Version (e.g. 2.6.0 to 2.6.1) | 467
Chapter 144
Upgrading a Minor Version (e.g. 2.5.3 to 2.6.1)

If you're upgrading a minor version (where the middle number changes), then you should not encounter
significant backwards compatibility changes. For details, see the Symfony backwards compatibility
promise.
However, some backwards-compatibility breaks are possible and you'll learn in a second how to prepare
for them.
There are two steps to upgrading a minor version:
1. Update the Symfony library via Composer;
2. Update your code to work with the new version.
1) Update the Symfony Library via Composer

First, you need to update Symfony by modifying your composer.json file to use the new version:
Listing 144-1
1
2
3
4
5
6
7
8
{
"...": "...",
"require": {
"symfony/symfony": "2.6.*",
},
"...": "...",
}
Next, use Composer to download new versions of the libraries:

Listing 144-2
Dependency Errors
If you get a dependency error, it may simply mean that you need to upgrade other Symfony dependencies
too. In that case, try the following command:
Listing 144-3
$ composer update symfony/symfony --with-dependencies

Chapter 144: Upgrading a Minor Version (e.g. 2.5.3 to 2.6.1) | 468
This updates symfony/symfony and all packages that it depends on, which will include several other
packages. By using tight version constraints in composer.json, you can control what versions each
library upgrades to.
If this still doesn't work, your composer.json file may specify a version for a library that is not
compatible with the newer Symfony version. In that case, updating that library to a newer version in
composer.json may solve the issue.
Or, you may have deeper issues where different libraries depend on conflicting versions of other libraries.
Check your error message to debug.

Listing 144-4
$ composer update
2) Updating your Code to Work with the new Version

In theory, you should be done! However, you may need to make a few changes to your code to
get everything working. Additionally, some features you're using might still work, but might now be
deprecated. While that's just fine, if you know about these deprecations, you can start to fix them over
time.
Every version of Symfony comes with an UPGRADE file (e.g. UPGRADE-2.7.md3) included in the
Symfony directory that describes these changes. If you follow the instructions in the document and
update your code accordingly, it should be safe to update in the future.
These documents can also be found in the Symfony Repository4.
3. https://github.com/symfony/symfony/blob/2.7/UPGRADE-2.7.md
4. https://github.com/symfony/symfony

Chapter 144: Upgrading a Minor Version (e.g. 2.5.3 to 2.6.1) | 469
Chapter 145
Upgrading a Major Version (e.g. 2.7.0 to 3.0.0)

Every few years, Symfony releases a new major version release (the first number changes). These releases
are the trickiest to upgrade, as they are allowed to contain BC breaks. However, Symfony tries to make
this upgrade process as smooth as possible.
This means that you can update most of your code before the major release is actually released. This is
called making your code future compatible.
There are a couple of steps to upgrading a major version:
1. Make your code deprecation free;
2. Update to the new major version via Composer;
3. Update your code to work with the new version.
1) Make your Code Deprecation Free

During the lifecycle of a major release, new features are added and method signatures and public API
usages are changed. However, minor versions should not contain any backwards incompatible changes.
To accomplish this, the "old" (e.g. functions, classes, etc) code still works, but is marked as deprecated,
indicating that it will be removed/changed in the future and that you should stop using it.
When the major version is released (e.g. 3.0.0), all deprecated features and functionality are removed. So,
as long as you've updated your code to stop using these deprecated features in the last version before the
major (e.g. 2.8.*), you should be able to upgrade without a problem.
To help you with this, deprecation notices are triggered whenever you end up using a deprecated feature.
When visiting your application in the dev environment in your browser, these notices are shown in the
web dev toolbar:

Chapter 145: Upgrading a Major Version (e.g. 2.7.0 to 3.0.0) | 470
Of course ultimately, you want to stop using the deprecated functionality. Sometimes, this is easy: the
warning might tell you exactly what to change.
But other times, the warning might be unclear: a setting somewhere might cause a class deeper to trigger
the warning. In this case, Symfony does its best to give a clear message, but you may need to research
that warning further.
And sometimes, the warning may come from a third-party library or bundle that you're using. If that's
true, there's a good chance that those deprecations have already been updated. In that case, upgrade the
library to fix them.
Once all the deprecation warnings are gone, you can upgrade with a lot more confidence.
Deprecations in PHPUnit
When you run your tests using PHPUnit, no deprecation notices are shown. To help you here, Symfony
provides a PHPUnit bridge. This bridge will show you a nice summary of all deprecation notices at the
end of the test report.
All you need to do is install the PHPUnit bridge:
Listing 145-1
$ composer require --dev symfony/phpunit-bridge
Now, you can start fixing the notices:

Listing 145-2
1
2
3
4
5
6
7
8
9
10
11
12
13
$ phpunit
...
OK (10 tests, 20 assertions)
Remaining deprecation notices (6)
The "request" service is deprecated and will be removed in 3.0. Add a typehint for
Symfony\Component\HttpFoundation\Request to your controller parameters to retrieve the
request instead: 6x
3x in PageAdminTest::testPageShow from Symfony\Cmf\SimpleCmsBundle\Tests\WebTest\Admin
2x in PageAdminTest::testPageList from Symfony\Cmf\SimpleCmsBundle\Tests\WebTest\Admin
1x in PageAdminTest::testPageEdit from Symfony\Cmf\SimpleCmsBundle\Tests\WebTest\Admin
Once you fixed them all, the command ends with 0 (success) and you're done!

Using the Weak Deprecations Mode

Sometimes, you can't fix all deprecations (e.g. something was deprecated in 2.8 and you still need
to support 2.7). In these cases, you can still use the bridge to fix as many deprecations as possible
and then switch to the weak test mode to make your tests pass again. You can do this by using the
SYMFONY_DEPRECATIONS_HELPER env variable:
Listing 145-3
1
2
3
4
5
6
7
8


<phpunit>

<php>
<env name="SYMFONY_DEPRECATIONS_HELPER" value="weak"/>
</php>
</phpunit>
(you can also execute the command like SYMFONY_DEPRECATIONS_HELPER=weak phpunit).
2) Update to the New Major Version via Composer

Once your code is deprecation free, you can update the Symfony library via Composer by modifying your
composer.json file:
Listing 145-4
1
2
3
4
5
6
7
8
{
"...": "...",
"require": {
"symfony/symfony": "3.0.*",
},
"...": "..."
}
Next, use Composer to download new versions of the libraries:

Listing 145-5
Dependency Errors
If you get a dependency error, it may simply mean that you need to upgrade other Symfony dependencies
too. In that case, try the following command:
Listing 145-6
$ composer update symfony/symfony --with-dependencies
This updates symfony/symfony and all packages that it depends on, which will include several other
packages. By using tight version constraints in composer.json, you can control what versions each
library upgrades to.
If this still doesn't work, your composer.json file may specify a version for a library that is not
compatible with the newer Symfony version. In that case, updating that library to a newer version in
composer.json may solve the issue.
Or, you may have deeper issues where different libraries depend on conflicting versions of other libraries.
Check your error message to debug.


Listing 145-7
$ composer update
3) Update your Code to Work with the New Version

There is a good chance that you're done now! However, the next major version may also contain new BC
breaks as a BC layer is not always a possibility. Make sure you read the UPGRADE-X.0.md (where X is
the new major version) included in the Symfony repository for any BC break that you need to be aware
of.

Chapter 146
Upgrading a Third-Party Bundle for a Major

Symfony Version
Symfony 3 was released on November 2015. Although this version doesn't contain any new feature, it
removes all the backwards compatibility layers included in the previous 2.8 version. If your bundle uses
any deprecated feature and it's published as a third-party bundle, applications upgrading to Symfony 3
will no longer be able to use it.
Allowing to Install Symfony 3 Components

Most third-party bundles define their Symfony dependencies using the ~2.N or ^2.N constraints in the
composer.json file. For example:
Listing 146-1
1
2
3
4
5
6
7
{
"require": {
"symfony/framework-bundle": "~2.7",
"symfony/finder": "~2.7",
"symfony/validator": "~2.7"
}
}
These constraints prevent the bundle from using Symfony 3 components, so it makes it impossible to
install it in a Symfony 3 based application. This issue is very easy to solve thanks to the flexibility of
Composer dependencies constraints. Just replace ~2.N by ~2.N|~3.0 (or ^2.N by ^2.N|~3.0).
The above example can be updated to work with Symfony 3 as follows:
Listing 146-2
1
2
3
4
5
6
7
{
"require": {
"symfony/framework-bundle": "~2.7|~3.0",
"symfony/finder": "~2.7|~3.0",
"symfony/validator": "~2.7|~3.0"
}
}

Chapter 146: Upgrading a Third-Party Bundle for a Major Symfony Version | 474
Another common version constraint found on third-party bundles is >=2.N. You should avoid using
that constraint because it's too generic (it means that your bundle is compatible with any future
Symfony version). Use instead ~2.N|~3.0 or ^2.N|~3.0 to make your bundle future-proof.
Looking for Deprecations and Fix Them

Besides allowing users to use your bundle with Symfony 3, your bundle must stop using any feature
deprecated by the 2.8 version because they are removed in 3.0 (you'll get exceptions or PHP errors). The
easiest way to detect deprecations is to install the symfony/phpunit-bridge package1 and then run the test
suite.
First, install the component as a dev dependency of your bundle:
Listing 146-3
$ composer require --dev symfony/phpunit-bridge
Then, run your test suite and look for the deprecation list displayed after the PHPUnit test report:
Listing 146-4
1
2
3
4
5
6
7
8
9
10
11
12
13
$ phpunit
# ... PHPUnit output

Remaining deprecation notices (3)
The "pattern" option in file ... is deprecated since version 2.2 and will be
removed in 3.0. Use the "path" option in the route definition instead ...
Twig Function "form_enctype" is deprecated. Use "form_start" instead in ...
The Symfony\Component\Security\Core\SecurityContext class is deprecated since
version 2.6 and will be removed in 3.0. Use ...
Fix the reported deprecations, run the test suite again and repeat the process until no deprecation usage
is reported.
Useful Resources
There are several resources that can help you detect, understand and fix the use of deprecated features:
Official Symfony Guide to Upgrade from 2.x to 3.02
The full list of changes required to upgrade to Symfony 3.0 and grouped by component.
SensioLabs DeprecationDetector3
It runs a static code analysis against your project's source code to find usages of deprecated methods,
classes and interfaces. It works for any PHP application, but it includes special detectors for
Symfony applications, where it can also detect usages of deprecated services.
Symfony Upgrade Fixer4
It analyzes Symfony projects to find deprecations. In addition it solves automatically some of them
thanks to the growing list of supported "fixers".
1. https://github.com/symfony/phpunit-bridge
2. https://github.com/symfony/symfony/blob/2.8/UPGRADE-3.0.md
3. https://github.com/sensiolabs-de/deprecation-detector
4. https://github.com/umpirsky/Symfony-Upgrade-Fixer

Testing your Bundle in Symfony 3

Now that your bundle has removed all deprecations, it's time to test it for real in a Symfony 3 application.
Assuming that you already have a Symfony 3 application, you can test the updated bundle locally without
having to install it through Composer.
If your operating system supports symbolic links, just point the appropriate vendor directory to your
local bundle root directory:
Listing 146-5
$ ln -s /path/to/your/local/bundle/ vendor/you-vendor-name/your-bundle-name
If your operating system doesn't support symbolic links, you'll need to copy your local bundle directory
into the appropriate directory inside vendor/.
Update the Travis CI Configuration

In addition to running tools locally, it's recommended to set-up Travis CI service to run the tests of your
bundle using different Symfony configurations. Use the following recommended configuration as the
starting point of your own configuration:
Listing 146-6
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
language: php
sudo: false
php:
- 5.3
- 5.6
- 7.0
matrix:
include:
- php:
env:
- php:
env:
- php:
env:
- php:
env:
- php:
env:
- php:
env:
5.3.3
COMPOSER_FLAGS='--prefer-lowest --prefer-stable' SYMFONY_DEPRECATIONS_HELPER=weak
5.6
SYMFONY_VERSION='2.7.*'
5.6
5.6
5.6
5.6
DEPENDENCES='dev' SYMFONY_VERSION='3.2.*@dev'
before_install:
- composer self-update
- if [ "$DEPENDENCIES" == "dev" ]; then perl -pi -e 's/^}$/,"minimum-stability":"dev"}/' composer.json;
fi;
- if [ "$SYMFONY_VERSION" != "" ]; then composer --no-update require symfony/symfony:${SYMFONY_VERSION};
fi;
install: composer update $COMPOSER_FLAGS
script: phpunit
Updating your Code to Support Symfony 2.x and 3.x at the Same Time
The real challenge of adding Symfony 3 support for your bundles is when you want to support both
Symfony 2.x and 3.x simultaneously using the same code. There are some edge cases where you'll need
to deal with the API differences.
Before diving into the specifics of the most common edge cases, the general recommendation is to not
rely on the Symfony Kernel version to decide which code to use:

Listing 146-7
1
2
3
4
5
if (Kernel::VERSION_ID <= 20800) {

// code for Symfony 2.x
} else {
// code for Symfony 3.x
}
Instead of checking the Symfony Kernel version, check the version of the specific component. For
example, the OptionsResolver API changed in its 2.6 version by adding a setDefined() method. The
recommended check in this case would be:
Listing 146-8
1
2
3
4
5
if (!method_exists('Symfony\Component\OptionsResolver\OptionsResolver', 'setDefined')) {
// code for the old OptionsResolver API
} else {
// code for the new OptionsResolver API
}

Chapter 147
How to Create a custom Validation Constraint

You can create a custom constraint by extending the base constraint class, Constraint1. As an example
you're going to create a simple validator that checks if a string contains only alphanumeric characters.
Creating the Constraint Class

First you need to create a Constraint class and extend Constraint2:
Listing 147-1
1
2
3
4
5
6
7
8
9
10
11
12
// src/AppBundle/Validator/Constraints/ContainsAlphanumeric.php
namespace AppBundle\Validator\Constraints;
/**
* @Annotation
*/
class ContainsAlphanumeric extends Constraint
{
public $message = 'The string "%string%" contains an illegal character: it can only contain letters or
numbers.';
}
The @Annotation annotation is necessary for this new constraint in order to make it available for
use in classes via annotations. Options for your constraint are represented as public properties on
the constraint class.
1. http://api.symfony.com/2.7/Symfony/Component/Validator/Constraint.html
2. http://api.symfony.com/2.7/Symfony/Component/Validator/Constraint.html

Chapter 147: How to Create a custom Validation Constraint | 478
Creating the Validator itself

As you can see, a constraint class is fairly minimal. The actual validation is performed by another
"constraint validator" class. The constraint validator class is specified by the constraint's
validatedBy() method, which includes some simple default logic:
Listing 147-2
// in the base Symfony\Component\Validator\Constraint class

public function validatedBy()
{
return get_class($this).'Validator';
}
1
2
3
4
5
In other words, if you create a custom Constraint (e.g. MyConstraint), Symfony will automatically
look for another class, MyConstraintValidator when actually performing the validation.
The validator class is also simple, and only has one required method validate():
Listing 147-3
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
// src/AppBundle/Validator/Constraints/ContainsAlphanumericValidator.php
namespace AppBundle\Validator\Constraints;
use Symfony\Component\Validator\ConstraintValidator;
class ContainsAlphanumericValidator extends ConstraintValidator
{
public function validate($value, Constraint $constraint)
{
if (!preg_match('/^[a-zA-Z0-9]+$/', $value, $matches)) {
// If you're using the new 2.5 validation API (you probably are!)
->setParameter('%string%', $value)
->addViolation();
// If you're using the old 2.4 validation API

/*
$this->context->addViolation(
array('%string%' => $value)
);
*/
}
}
}
Inside validate, you don't need to return a value. Instead, you add violations to the validator's
context property and a value will be considered valid if it causes no violations. The buildViolation
method takes the error message as its argument and returns an instance of
ConstraintViolationBuilderInterface3. The addViolation method call finally adds the
violation to the context.
Using the new Validator

Using custom validators is very easy, just as the ones provided by Symfony itself:
Listing 147-4
1
2
3
4
5
// src/AppBundle/Entity/AcmeEntity.php
use AppBundle\Validator\Constraints as AcmeAssert;
class AcmeEntity
3. http://api.symfony.com/2.7/Symfony/Component/Validator/Violation/ConstraintViolationBuilderInterface.html

6
7
8
9
10
11
12
13
14
15
16
// ...
/**
* @Assert\NotBlank
* @AcmeAssert\ContainsAlphanumeric
*/
protected $name;
// ...
}
If your constraint contains options, then they should be public properties on the custom Constraint class
you created earlier. These options can be configured like options on core Symfony constraints.
Constraint Validators with Dependencies

If your constraint validator has dependencies, such as a database connection, it will need to be configured
as a service in the Dependency Injection Container. This service must include the
validator.constraint_validator tag and should include an alias attribute to be used in the
validatedBy method of your validator class:
Listing 147-5
1
2
3
4
5
6
services:
validator.unique.your_validator_name:
class: Fully\Qualified\Validator\Class\Name
tags:
- { name: validator.constraint_validator, alias: alias_name }
As mentioned above, Symfony will automatically look for a class named after the constraint, with
Validator appended. You can override this in your constraint class:
Listing 147-6
public function validatedBy()

{
return 'Fully\Qualified\ConstraintValidator\Class\Name'; // or 'alias_name' if provided
}
Make sure to use the 'alias_name' when you have configured your validator as a service. Otherwise your
validator class will be simply instantiated without your dependencies.
Class Constraint Validator

Beside validating a class property, a constraint can have a class scope by providing a target in its
Constraint class:
Listing 147-7
public function getTargets()

{
return self::CLASS_CONSTRAINT;
}
With this, the validator validate() method gets an object as its first argument:
Listing 147-8
1
2
3
4
5
6
7
8
9
10
class ProtocolClassValidator extends ConstraintValidator

{
public function validate($protocol, Constraint $constraint)
{
if ($protocol->getFoo() != $protocol->getBar()) {
// If you're using the new 2.5 validation API (you probably are!)
->atPath('foo')
->addViolation();

11
12
13
14
15
16
17
18
19
20
21
22
// If you're using the old 2.4 validation API

/*
$this->context->addViolationAt(
'foo',
array(),
null
);
*/
}
}
}
Note that a class constraint validator is applied to the class itself, and not to the property:
Listing 147-9
1
2
3
4
5
6
7
/**
* @AcmeAssert\ContainsAlphanumeric
*/
class AcmeEntity
{
// ...
}

Chapter 148
How to Handle Different Error Levels

Sometimes, you may want to display constraint validation error messages differently based on some rules.
For example, you have a registration form for new users where they enter some personal information and
choose their authentication credentials. They would have to choose a username and a secure password,
but providing bank account information would be optional. Nonetheless, you want to make sure that
these optional fields, if entered, are still valid, but display their errors differently.
The process to achieve this behavior consists of two steps:
1. Apply different error levels to the validation constraints;
2. Customize your error messages depending on the configured error level.
1. Assigning the Error Level

New in version 2.6: The payload option was introduced in Symfony 2.6.
Use the payload option to configure the error level for each constraint:
Listing 148-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
class User
{
/**
* @Assert\NotBlank(payload = {"severity" = "error"})
*/
protected $username;
/**
* @Assert\NotBlank(payload = {"severity" = "error"})
*/
protected $password;
/**
* @Assert\Iban(payload = {"severity" = "warning"})
*/
protected $bankAccountNumber;
}

Chapter 148: How to Handle Different Error Levels | 482
2. Customize the Error Message Template

New in version 2.6: The getConstraint() method in the ConstraintViolation class was
introduced in Symfony 2.6.
When validation of the User object fails, you can retrieve the constraint that caused a particular
failure using the getConstraint()1 method. Each constraint exposes the attached payload as a public
property:
Listing 148-2
1
2
3
4
5
// a constraint validation failure, instance of

// Symfony\Component\Validator\ConstraintViolation
$constraintViolation = ...;
$constraint = $constraintViolation->getConstraint();
$severity = isset($constraint->payload['severity']) ? $constraint->payload['severity'] : null;
For example, you can leverage this to customize the form_errors block so that the severity is added as
an additional HTML class:
Listing 148-3
1
2
3
4
5
6
7
8
9
10
11
12
{%- block form_errors -%}

{%- if errors|length > 0 -%}
<ul>
{%- for error in errors -%}
{% if error.cause.constraint.payload.severity is defined %}
{% set severity = error.cause.constraint.payload.severity %}
{% endif %}
<li{% if severity is defined %} class="{{ severity }}"{% endif %}>{{ error.message }}</li>
{%- endfor -%}
</ul>
{%- endif -%}
{%- endblock form_errors -%}
For more information on customizing form rendering, see How to Customize Form Rendering.
1. http://api.symfony.com/2.7/Symfony/Component/Validator/ConstraintViolation.html#method_getConstraint

Chapter 148: How to Handle Different Error Levels | 483
Chapter 149
How to Dynamically Configure Validation

Groups
Sometimes you need advanced logic to determine the validation groups. If they can't be determined by
a simple callback, you can use a service. Create a service that implements __invoke which accepts a
FormInterface as a parameter.
Listing 149-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
// src/AppBundle/Validation/ValidationGroupResolver.php
namespace AppBundle\Validation;
class ValidationGroupResolver
{
private $service1;
private $service2;
public function __construct($service1, $service2)
{
$this->service1 = $service1;
$this->service2 = $service2;
}
/**
* @param FormInterface $form
* @return array
*/
public function __invoke(FormInterface $form)
{
$groups = array();
// ... determine which groups to apply and return an array
return $groups;
}
}
Then in your form, inject the resolver and set it as the validation_groups.
Listing 149-2

Chapter 149: How to Dynamically Configure Validation Groups | 484
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
// src/AppBundle/Form/MyClassType.php;
use AppBundle\Validator\ValidationGroupResolver;
use Symfony\Component\Form\AbstractType
class MyClassType extends AbstractType
{
private $groupResolver;
public function __construct(ValidationGroupResolver $groupResolver)
{
$this->groupResolver = $groupResolver;
}
// ...
{
'validation_groups' => $this->groupResolver,
));
}
}
This will result in the form validator invoking your group resolver to set the validation groups returned
when validating.

Chapter 149: How to Dynamically Configure Validation Groups | 485
Chapter 150
How to Use PHP's built-in Web Server

New in version 2.6: The ability to run the server as a background process was introduced in Symfony 2.6.
Since PHP 5.4 the CLI SAPI comes with a built-in web server1. It can be used to run your PHP applications
locally during development, for testing or for application demonstrations. This way, you don't have to
bother configuring a full-featured web server such as Apache or Nginx.
The built-in web server is meant to be run in a controlled environment. It is not designed to be used
on public networks.
Starting the Web Server

Running a Symfony application using PHP's built-in web server is as easy as executing the
server:start command:
Listing 150-1
$ php app/console server:start
This starts the web server at localhost:8000 in the background that serves your Symfony application.
By default, the web server listens on port 8000 on the loopback device. You can change the socket passing
an IP address and a port as a command-line argument:
Listing 150-2
$ php app/console server:start 192.168.0.1:8080
1. http://www.php.net/manual/en/features.commandline.webserver.php

Chapter 150: How to Use PHP's built-in Web Server | 486
You can use the server:status command to check if a web server is listening on a certain socket:
Listing 150-3
1
2
3
$ php app/console server:status

$ php app/console server:status 192.168.0.1:8080
The first command shows if your Symfony application will be server through localhost:8000,
the second one does the same for 192.168.0.1:8080.
Before Symfony 2.6, the server:run command was used to start the built-in web server. This
command is still available and behaves slightly different. Instead of starting the server in the
background, it will block the current terminal until you terminate it (this is usually done by pressing
Ctrl and C).
Using the built-in Web Server from inside a Virtual Machine

If you want to use the built-in web server from inside a virtual machine and then load the site from
a browser on your host machine, you'll need to listen on the 0.0.0.0:8000 address (i.e. on all IP
addresses that are assigned to the virtual machine):
Listing 150-4
$ php app/console server:start 0.0.0.0:8000
You should NEVER listen to all interfaces on a computer that is directly accessible from the
Internet. The built-in web server is not designed to be used on public networks.
Command Options
The built-in web server expects a "router" script (read about the "router" script on php.net2) as an
argument. Symfony already passes such a router script when the command is executed in the prod or
in the dev environment. Use the --router option in any other environment or to use another router
script:
Listing 150-5
$ php app/console server:start --env=test --router=app/config/router_test.php
If your application's document root differs from the standard directory layout, you have to pass the
correct location using the --docroot option:
Listing 150-6
$ php app/console server:start --docroot=public_html
Stopping the Server

When you are finished, you can simply stop the web server using the server:stop command:
Listing 150-7
$ php app/console server:stop
2. http://php.net/manual/en/features.commandline.webserver.php#example-411

Like with the start command, if you omit the socket information, Symfony will stop the web server bound
to localhost:8000. Just pass the socket information when the web server listens to another IP address
or to another port:
Listing 150-8
$ php app/console server:stop 192.168.0.1:8080

Chapter 151
How to Create a SOAP Web Service in a

Symfony Controller
Setting up a controller to act as a SOAP server is simple with a couple tools. You must, of course, have
the PHP SOAP1 extension installed. As the PHP SOAP extension can not currently generate a WSDL, you
must either create one from scratch or use a 3rd party generator.
There are several SOAP server implementations available for use with PHP. Zend SOAP2 and
NuSOAP3 are two examples. Although the PHP SOAP extension is used in these examples, the
general idea should still be applicable to other implementations.
SOAP works by exposing the methods of a PHP object to an external entity (i.e. the person using the
SOAP service). To start, create a class - HelloService - which represents the functionality that you'll
expose in your SOAP service. In this case, the SOAP service will allow the client to call a method called
hello, which happens to send an email:
Listing 151-1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
// src/Acme/SoapBundle/Services/HelloService.php
namespace Acme\SoapBundle\Services;
class HelloService
{
private $mailer;
public function __construct(\Swift_Mailer $mailer)
{
$this->mailer = $mailer;
}
public function hello($name)
{
1. http://php.net/manual/en/book.soap.php
2. http://framework.zend.com/manual/current/en/modules/zend.soap.server.html
3. http://sourceforge.net/projects/nusoap

Chapter 151: How to Create a SOAP Web Service in a Symfony Controller | 489
18
19
20
21
22
23
24
25
->setSubject('Hello Service')
->setBody($name . ' says hi!');
$this->mailer->send($message);
return 'Hello, '.$name;
}
}
Next, you can train Symfony to be able to create an instance of this class. Since the class sends an email,
it's been designed to accept a Swift_Mailer instance. Using the Service Container, you can configure
Symfony to construct a HelloService object properly:
Listing 151-2
1
2
3
4
5
services:
hello_service:
class: Acme\SoapBundle\Services\HelloService
arguments: ['@mailer']
Below is an example of a controller that is capable of handling a SOAP request. If indexAction() is

accessible via the route /soap, then the WSDL document can be retrieved via /soap?wsdl.
Listing 151-3
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
namespace Acme\SoapBundle\Controller;
class HelloServiceController extends Controller
{
{
$server = new \SoapServer('/path/to/hello.wsdl');
$server->setObject($this->get('hello_service'));
$response->headers->set('Content-Type', 'text/xml; charset=ISO-8859-1');
ob_start();
$server->handle();
$response->setContent(ob_get_clean());
return $response;
}
}
Take note of the calls to ob_start() and ob_get_clean(). These methods control output buffering4
which allows you to "trap" the echoed output of $server->handle(). This is necessary because
Symfony expects your controller to return a Response object with the output as its "content". You must
also remember to set the "Content-Type" header to "text/xml", as this is what the client will expect. So,
you use ob_start() to start buffering the STDOUT and use ob_get_clean() to dump the echoed
output into the content of the Response and clear the output buffer. Finally, you're ready to return the
Response.
Below is an example calling the service using a NuSOAP5 client. This example assumes that the
indexAction in the controller above is accessible via the route /soap:
Listing 151-4
$client = new \Soapclient('http://example.com/app.php/soap?wsdl');

$result = $client->call('hello', array('name' => 'Scott'));
4. http://php.net/manual/en/book.outcontrol.php
5. http://sourceforge.net/projects/nusoap

An example WSDL is below.

Listing 151-5
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
<?xml version="1.0" encoding="ISO-8859-1"?>

<definitions xmlns:SOAP-ENV="http://schemas.xmlsoap.org/soap/envelope/"
xmlns:xsd="http://www.w3.org/2001/XMLSchema"
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xmlns:SOAP-ENC="http://schemas.xmlsoap.org/soap/encoding/"
xmlns:tns="urn:arnleadservicewsdl"
xmlns:soap="http://schemas.xmlsoap.org/wsdl/soap/"
xmlns:wsdl="http://schemas.xmlsoap.org/wsdl/"
xmlns="http://schemas.xmlsoap.org/wsdl/"
targetNamespace="urn:helloservicewsdl">
<types>
<xsd:schema targetNamespace="urn:hellowsdl">
<xsd:import namespace="http://schemas.xmlsoap.org/soap/encoding/" />
<xsd:import namespace="http://schemas.xmlsoap.org/wsdl/" />
</xsd:schema>
</types>
<message name="helloRequest">
<part name="name" type="xsd:string" />
</message>
<message name="helloResponse">
<part name="return" type="xsd:string" />
</message>
<portType name="hellowsdlPortType">
<operation name="hello">
<documentation>Hello World</documentation>
<input message="tns:helloRequest"/>
<output message="tns:helloResponse"/>
</operation>
</portType>
<binding name="hellowsdlBinding" type="tns:hellowsdlPortType">
<soap:binding style="rpc" transport="http://schemas.xmlsoap.org/soap/http"/>
<operation name="hello">
<soap:operation soapAction="urn:arnleadservicewsdl#hello" style="rpc"/>
<input>
<soap:body use="encoded" namespace="urn:hellowsdl"
encodingStyle="http://schemas.xmlsoap.org/soap/encoding/"/>
</input>
<output>
<soap:body use="encoded" namespace="urn:hellowsdl"
encodingStyle="http://schemas.xmlsoap.org/soap/encoding/"/>
</output>
</operation>
</binding>
<service name="hellowsdl">
<port name="hellowsdlPort" binding="tns:hellowsdlBinding">
<soap:address location="http://example.com/app.php/soap" />
</port>
</service>
</definitions>

Chapter 152
How to Create and Store a Symfony Project in

Git
Though this entry is specifically about Git, the same generic principles will apply if you're storing
your project in Subversion.
Once you've read through Create your First Page in Symfony and become familiar with using Symfony,
you'll no-doubt be ready to start your own project. In this cookbook article, you'll learn the best way to
start a new Symfony project that's stored using the Git1 source control management system.
Initial Project Setup

To get started, you'll need to download Symfony and get things running. See the Installing and
Configuring Symfony chapter for details.
Once your project is running, just follow these simple steps:
1. Initialize your Git repository:
Listing 152-1
$ git init
2. Add all of the initial files to Git:

Listing 152-2
$ git add .
1. http://git-scm.com/

Chapter 152: How to Create and Store a Symfony Project in Git | 492
As you might have noticed, not all files that were downloaded by Composer in step
1, have been staged for commit by Git. Certain files and folders, such as the project's
dependencies (which are managed by Composer), parameters.yml (which contains
sensitive information such as database credentials), log and cache files and dumped assets
(which are created automatically by your project), should not be committed in Git. To
help you prevent committing those files and folders by accident, the Standard Distribution
comes with a file called .gitignore, which contains a list of files and folders that Git
should ignore.
You may also want to create a .gitignore file that can be used system-wide. This allows
you to exclude files/folders for all your projects that are created by your IDE or operating
system. For details, see GitHub .gitignore2.
3. Create an initial commit with your started project:

Listing 152-3
$ git commit -m "Initial commit"
At this point, you have a fully-functional Symfony project that's correctly committed to Git. You can
immediately begin development, committing the new changes to your Git repository.
You can continue to follow along with the Create your First Page in Symfony chapter to learn more about
how to configure and develop inside your application.
The Symfony Standard Edition comes with some example functionality. To remove the sample code,
follow the instructions in the "How to Remove the AcmeDemoBundle" article.
Managing Vendor Libraries with composer.json

How Does it Work?
Every Symfony project uses a group of third-party "vendor" libraries. One way or another the goal is to
download these files into your vendor/ directory and, ideally, to give you some sane way to manage the
exact version you need for each.
By default, these libraries are downloaded by running a composer install "downloader" binary.
This composer file is from a library called Composer3 and you can read more about installing it in the
Installation chapter.
The composer command reads from the composer.json file at the root of your project. This is an
JSON-formatted file, which holds a list of each of the external packages you need, the version to be
downloaded and more. composer also reads from a composer.lock file, which allows you to pin each
library to an exact version. In fact, if a composer.lock file exists, the versions inside will override those
in composer.json. To upgrade your libraries to new versions, run composer update.
If you want to add a new package to your application, run the composer require command:
Listing 152-4
$ composer require doctrine/doctrine-fixtures-bundle
2. https://help.github.com/articles/ignoring-files

To learn more about Composer, see GetComposer.org4:

It's important to realize that these vendor libraries are not actually part of your repository. Instead, they're
simply un-tracked files that are downloaded into the vendor/. But since all the information needed
to download these files is saved in composer.json and composer.lock (which are stored in the
repository), any other developer can use the project, run composer install, and download the exact
same set of vendor libraries. This means that you're controlling exactly what each vendor library looks
like, without needing to actually commit them to your repository.
So, whenever a developer uses your project, they should run the composer install script to ensure
that all of the needed vendor libraries are downloaded.
Upgrading Symfony
Since Symfony is just a group of third-party libraries and third-party libraries are entirely controlled
through composer.json and composer.lock, upgrading Symfony means simply upgrading
each of these files to match their state in the latest Symfony Standard Edition.
Of course, if you've added new entries to composer.json, be sure to replace only the original parts
(i.e. be sure not to also delete any of your custom entries).
Storing your Project on a remote Server

You now have a fully-functional Symfony project stored in Git. However, in most cases, you'll also want
to store your project on a remote server both for backup purposes, and so that other developers can
collaborate on the project.
The easiest way to store your project on a remote server is via a web-based hosting service like GitHub5
or Bitbucket6. Of course, there are more services out there, you can start your research with a comparison
of hosting services7.
Alternatively, you can store your Git repository on any server by creating a barebones repository8 and then
pushing to it. One library that helps manage this is Gitolite9.
5. https://github.com/
6. https://bitbucket.org/
7. https://en.wikipedia.org/wiki/Comparison_of_open-source_software_hosting_facilities
8. http://git-scm.com/book/en/Git-Basics-Getting-a-Git-Repository
9. https://github.com/sitaramc/gitolite

Chapter 153
How to Create and Store a Symfony Project in

Subversion
This entry is specifically about Subversion, and based on principles found in How to Create and Store
a Symfony Project in Git.
Once you've read through Create your First Page in Symfony and become familiar with using Symfony,
you'll no-doubt be ready to start your own project. The preferred method to manage Symfony projects is
using Git1 but some prefer to use Subversion2 which is totally fine!. In this cookbook article, you'll learn
how to manage your project using SVN3 in a similar manner you would do with Git4.
This is a method to tracking your Symfony project in a Subversion repository. There are several ways
to do and this one is simply one that works.
The Subversion Repository

For this article it's assumed that your repository layout follows the widespread standard structure:
Listing 153-1
1
2
3
4
myproject/
branches/
tags/
trunk/
2. http://subversion.apache.org/

Chapter 153: How to Create and Store a Symfony Project in Subversion | 495
Most Subversion hosting should follow this standard practice. This is the recommended layout in
Version Control with Subversion5 and the layout used by most free hosting (see Subversion Hosting
Solutions).
Initial Project Setup

To get started, you'll need to download Symfony and get the basic Subversion setup. First, download and
get your Symfony project running by following the Installation chapter.
Once you have your new project directory and things are working, follow along with these steps:
1. Checkout the Subversion repository that will host this project. Suppose it is hosted on Google
code6 and called myproject:
Listing 153-2
$ svn checkout http://myproject.googlecode.com/svn/trunk myproject
2. Copy the Symfony project files in the Subversion folder:

Listing 153-3
$ mv Symfony/* myproject/
3. Now, set the ignore rules. Not everything should be stored in your Subversion repository. Some
files (like the cache) are generated and others (like the database configuration) are meant to be
customized on each machine. This makes use of the svn:ignore property, so that specific files
can be ignored.
Listing 153-4
1
2
3
4
5
6
7
8
9
10
11
12
$ cd myproject/
$ svn add --depth=empty app app/cache app/logs app/config web
$
$
$
$
$
svn
svn
svn
svn
svn
propset
propset
propset
propset
propset
svn:ignore
svn:ignore
svn:ignore
svn:ignore
svn:ignore
"vendor" .
"bootstrap*" app/
"parameters.yml" app/config/
"*" app/cache/
"*" app/logs/
$ svn propset svn:ignore "bundles" web

$ svn ci -m "commit basic Symfony ignore list (vendor, app/bootstrap*, app/config/parameters.yml,
app/cache/*, app/logs/*, web/bundles)"
4. The rest of the files can now be added and committed to the project:
Listing 153-5
1
2
$ svn add --force .

$ svn ci -m "add basic Symfony Standard 2.X.Y"
That's it! Since the app/config/parameters.yml file is ignored, you can store machine-specific
settings like database passwords here without committing them. The parameters.yml.dist file is
committed, but is not read by Symfony. And by adding any new keys you need to both files, new
developers can quickly clone the project, copy this file to parameters.yml, customize it, and start
developing.
At this point, you have a fully-functional Symfony project stored in your Subversion repository. The
development can start with commits in the Subversion repository.
You can continue to follow along with the Create your First Page in Symfony chapter to learn more about
how to configure and develop inside your application.
5. http://svnbook.red-bean.com/
6. http://code.google.com/hosting/

The Symfony Standard Edition comes with some example functionality. To remove the sample code,
follow the instructions in the "How to Remove the AcmeDemoBundle" article.
Managing Vendor Libraries with composer.json

How Does it Work?
Every Symfony project uses a group of third-party "vendor" libraries. One way or another the goal is to
download these files into your vendor/ directory and, ideally, to give you some sane way to manage the
exact version you need for each.
By default, these libraries are downloaded by running a composer install "downloader" binary.
This composer file is from a library called Composer7 and you can read more about installing it in the
Installation chapter.
The composer command reads from the composer.json file at the root of your project. This is an
JSON-formatted file, which holds a list of each of the external packages you need, the version to be
downloaded and more. composer also reads from a composer.lock file, which allows you to pin each
library to an exact version. In fact, if a composer.lock file exists, the versions inside will override those
in composer.json. To upgrade your libraries to new versions, run composer update.
If you want to add a new package to your application, run the composer require command:
Listing 153-6
$ composer require doctrine/doctrine-fixtures-bundle
To learn more about Composer, see GetComposer.org8:

It's important to realize that these vendor libraries are not actually part of your repository. Instead, they're
simply un-tracked files that are downloaded into the vendor/. But since all the information needed
to download these files is saved in composer.json and composer.lock (which are stored in the
repository), any other developer can use the project, run composer install, and download the exact
same set of vendor libraries. This means that you're controlling exactly what each vendor library looks
like, without needing to actually commit them to your repository.
So, whenever a developer uses your project, they should run the composer install script to ensure
that all of the needed vendor libraries are downloaded.
Upgrading Symfony
Since Symfony is just a group of third-party libraries and third-party libraries are entirely controlled
through composer.json and composer.lock, upgrading Symfony means simply upgrading
each of these files to match their state in the latest Symfony Standard Edition.
Of course, if you've added new entries to composer.json, be sure to replace only the original parts
(i.e. be sure not to also delete any of your custom entries).

Subversion Hosting Solutions

The biggest difference between Git9 and SVN10 is that Subversion needs a central repository to work. You
then have several solutions:
Self hosting: create your own repository and access it either through the filesystem or the network.
To help in this task you can read Version Control with Subversion.
Third party hosting: there are a lot of serious free hosting solutions available like GitHub11, Google
code12, SourceForge13 or Gna14. Some of them offer Git hosting as well.
11. https://github.com/
12. http://code.google.com/hosting/
13. http://sourceforge.net/
14. http://gna.org/

Chapter 154
Using Symfony with Homestead/Vagrant

In order to develop a Symfony application, you might want to use a virtual development environment
instead of the built-in server or WAMP/LAMP. Homestead1 is an easy-to-use Vagrant2 box to get a virtual
environment up and running quickly.
Due to the amount of filesystem operations in Symfony (e.g. updating cache files and writing to log
files), Symfony can slow down significantly. To improve the speed, consider overriding the cache and
log directories to a location outside the NFS share (for instance, by using sys_get_temp_dir3).
You can read this blog post4 for more tips to speed up Symfony on Vagrant.
Install Vagrant and Homestead

Before you can use Homestead, you need to install and configure Vagrant and Homestead as explained
in the Homestead documentation5.
Setting Up a Symfony Application

Imagine you've installed your Symfony application in ~/projects/symfony_demo on your local
system. You first need Homestead to sync your files in this project. Execute homestead edit to edit
the Homestead configuration and configure the ~/projects directory:
Listing 154-1
1
2
3
4
# ...
folders:
- map: ~/projects
to: /home/vagrant/projects
1. http://laravel.com/docs/homestead
2. https://www.vagrantup.com/
3. http://php.net/manual/en/function.sys-get-temp-dir.php
4. http://www.whitewashing.de/2013/08/19/speedup_symfony2_on_vagrant_boxes.html
5. http://laravel.com/docs/homestead#installation-and-setup

Chapter 154: Using Symfony with Homestead/Vagrant | 499
The projects/ directory on your PC is now accessible at /home/vagrant/projects in the

Homestead environment.
After you've done this, configure the Symfony application in the Homestead configuration:
Listing 154-2
1
2
3
4
5
# ...
sites:
- map: symfony-demo.dev
to: /home/vagrant/projects/symfony_demo/web
type: symfony
The type option tells Homestead to use the Symfony nginx configuration.
At last, edit the hosts file on your local machine to map symfony-demo.dev to 192.168.10.10
(which is the IP used by Homestead):
Listing 154-3
# /etc/hosts (unix) or C:\Windows\System32\drivers\etc\hosts (Windows)

192.168.10.10 symfony-demo.dev
Now, navigate to http://symfony-demo.dev in your web browser and enjoy developing your
Symfony application!
To learn more features of Homestead, including Blackfire Profiler integration, automatic creation of MySQL
databases and more, read the Daily Usage6 section of the Homestead documentation.
6. http://laravel.com/docs/5.1/homestead#daily-usage

Chapter 154: Using Symfony with Homestead/Vagrant | 500

Symfony Cookbook 2.7

Uploaded by

Copyright:

Available Formats

Symfony Cookbook 2.7

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Symfony Cookbook 2.7

Uploaded by

Copyright:

Available Formats

The Cookbook

The Cookbook (2.7)

Contents at a Glance | iii

How to Deploy a Symfony Application ............................................................................................ 130

How to Use Matchers to Enable the Profiler Conditionally ............................................................... 311

PDF brought to you by

How to Use Assetic for Asset Management

<script src="{{ asset('js/script.js') }}"></script>

Including JavaScript Files

PDF brought to you by

Chapter 1: How to Use Assetic for Asset Management | 7

Including CSS Stylesheets

{% stylesheets 'bundles/app/css/*' filter='cssrewrite' %}

PDF brought to you by

Chapter 1: How to Use Assetic for Asset Management | 8

Fixing CSS Paths with the cssrewrite Filter

PDF brought to you by

Chapter 1: How to Use Assetic for Asset Management | 9

Using Named Assets

{% javascripts '@AppBundle/Resources/public/js/*' filter='uglifyjs2' %}

PDF brought to you by

Chapter 1: How to Use Assetic for Asset Management | 10

Controlling the URL Used

{% javascripts '@AppBundle/Resources/public/js/*' output='js/compiled/main.js' %}

Dumping Asset Files

Dumping Asset Files in the prod Environment

$ php app/console assetic:dump --env=prod --no-debug

PDF brought to you by

Chapter 1: How to Use Assetic for Asset Management | 11

Dumping Asset Files in the dev Environment

$ php app/console assetic:dump

$ php app/console assetic:watch

{% javascripts '@AppBundle/Resources/public/js/*' output='js/compiled/main.js' %}

PDF brought to you by

Chapter 1: How to Use Assetic for Asset Management | 12

Combining, Compiling and Minimizing Web

Installing the Third-Party Compression Libraries

$ composer require leafo/scssphp

Organizing your Web Asset Files

PDF brought to you by

Combining and Minimizing CSS Files and Compiling SCSS Files

PDF brought to you by

Combining and Minimizing JavaScript Files

PDF brought to you by

How to Minify CSS/JS Files (Using UglifyJS and

$ npm install -g uglify-js

PDF brought to you by

Chapter 3: How to Minify CSS/JS Files (Using UglifyJS and UglifyCSS) | 16

Configure the uglifyjs2 Filter

You now have access to the uglifyjs2 filter in your application.

Configure the node Binary

PDF brought to you by

Chapter 3: How to Minify CSS/JS Files (Using UglifyJS and UglifyCSS) | 17

# the path to the uglifyjs executable

Minify your Assets

{% javascripts '@AppBundle/Resources/public/js/*' filter='uglifyjs2' %}

Disable Minification in Debug Mode

{% javascripts '@AppBundle/Resources/public/js/*' filter='?uglifyjs2' %}