Open Source RAD with OpenObject

PREAMBLE OpenERP is a modern Enterprise Management Software, released

Control). You also need to install the required dependencies (PostgreSQL and a few Python libraries see documentation on doc.openerp.com).
Compilation tip: OpenERP being Python-based, no compilation step is needed
1 2 3

under the AGPL license, and featuring CRM, HR, Sales, Accounting, Manufacturing, Inventory, Project Management, ... It is based on OpenObject, a modular, scalable, and intuitive Rapid Application Development (RAD) framework written in Python.

Typical bazaar checkout procedure (on Debian-based Linux) $ sudo apt-get install bzr # install bazaar version control $ bzr branch lp:openerp # retrieve source installer $ cd openerp && python ./bzr_set.py # fetch code and perform setup

OpenObject features a complete and modular toolbox for quickly building applications: integrated Object-Relationship Mapping (ORM) support, template-based Model-View-Controller (MVC) interfaces, a report generation system, automated internationalization, and much more. Python is a high-level dynamic programming language, ideal for RAD, combining power with clear syntax, and a core kept small by design.
Tip: Useful links Main website, with OpenERP downloads: www.openerp.com Functional & technical documentation: doc.openerp.com Community resources: www.launchpad.net/open-object Integration server: test,openobject.com Learning Python: doc.python.org OpenERP E-Learning platform: edu.openerp.com

Database creation

After installation, run the server and the client. From the GTK client, use FileDatabasesNew Database to create a new database (default super admin password is admin). Each database has its own modules and config, and demo data can be included.

20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36

'version' : '1.0', 'author' : 'OpenERP', 'description' : 'Ideas management module', 'category': 'Enterprise Innovation', 'website': 'http://www.openerp.com', 'depends' : ['base'], # list of dependencies, conditioning startup order 'update_xml' : [ # data files to load at module init 'security/groups.xml', # always load groups first! 'security/ir.model.access.csv', # load access rights after groups 'workflow/workflow.xml', 'view/views.xml', 'wizard/wizard.xml', 'report/report.xml', ], 'demo_xml': ['demo/demo.xml'], # demo data (for unit tests) 'active': False, # whether to install automatically at new DB creation }

Object Service ORM

Building an OpenERP module: idea

CONTEXT

The code samples used in this memento are taken from a hypothetical idea module. The purpose of this module would be to help creative minds, who often come up with ideas that cannot be pursued immediately, and are too easily forgotten if not logged somewhere. It could be used to record these ideas, sort them and rate them.
Note: Modular development OpenObject uses modules as feature containers, to foster maintainable and robust development. Modules provide feature isolation, an appropriate level of abstraction, and obvious MVC patterns.

Key component of OpenObject, the Object Service (OSV) implements a complete Object-Relational mapping layer, freeing developers from having to write basic SQL plumbing. Business objects are declared as Python classes inheriting from the osv.osv class, which makes them part of the OpenObject Model, and magically persisted by the ORM layer. Predefined attributes are used in the Python class to specify a business object's characteristics for the ORM:
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

idea.py:

Installing OpenERP
OpenERP is distributed as packages/installers for most platforms, but can of course be installed from the source on any platform.

Composition of a module

OpenERP Architecture

A module may contain any of the following elements: business objects: declared as Python classes extending the osv.osv OpenObject class, the persistence of these resources is completely managed by OpenObject ; data: XML/CSV files with meta-data (views and workflows declaration), configuration data (modules parametrization) and demo data (optional but recommended for testing, e.g. sample ideas) ; wizards: stateful interactive forms used to assist users, often available as contextual actions on resources ; reports: RML (XML format), MAKO or OpenOffice report templates, to be merged with any kind of business data, and generate HTML, ODT or PDF reports.

Typical module structure

OpenERP uses the well-known client-server paradigm, with different pieces of software acting as client and server depending on the desired configuration.Client software OpenERP provides a thick desktop client (GTK+) on all platforms, and a web interface is also accessible using any modern browser.
Tip: Installation procedure The procedure for installing OpenERP is likely to evolve (dependencies and so on), so make sure to always check the specific documentation (packaged & on website) for the latest procedures. See http://doc.openerp.com/install

Each module is contained in its own directory within the server/bin/addons directory in the server installation.
Note: You can declare your own add-ons directory in the configuration file of OpenERP (passed to the server with the -c option) using the addons_path option.
4 5 6 7 8 9 10 11 12 13 14 15

Package installation
Windows Linux Mac

all-in-one installer, and separate installers for server, client, and webserver are on the website openerp-server and openerp-client packages are available via corresponding package manager (e.g. Synaptic on Ubuntu) look online for package installers for the GTK client, as well as tutorials for installing the server (e.g. devteam.taktik.be)

addons/ |- idea/ |- demo/ |- i18n/ |- report/ |- security/ |- view/ |- wizard/ |- workflow/ |- __init__.py |- __terp__.py |- idea.py

# # # # # # # # # # #

The module directory Demo and unit test population data Translation files Report definitions Declaration of groups and access rights Views (forms,lists), menus and actions Wizards definitions Workflow definitions Python package initialization (required) module declaration (required) Python classes, the module's objects

from osv import osv, fields class idea(osv.osv): idea _name = 'idea.idea' _columns = { 'name': fields.char('Title', size=64, required=True, translate=True), 'state': fields.selection([('draft','Draft'), ('confirmed','Confirmed')],'State',required=True,readonly=True), # Description is read-only when not draft! 'description': fields.text('Description', readonly=True, states={'draft': [('readonly', False)]} ), 'active': fields.boolean('Active'), 'invent_date': fields.date('Invent date'), # by convention, many2one fields end with '_id' 'inventor_id': fields.many2one('res.partner','Inventor'), 'inventor_country_id': fields.related('inventor_id','country', readonly=True, type='many2one', relation='res.country', string='Country'), # by convention, *2many fields end with '_ids' 'vote_ids': fields.one2many('idea.vote','idea_id','Votes'), 'sponsor_ids': fields.many2many('res.partner','idea_sponsor_rel', 'idea_id','sponsor_id','Sponsors'), 'score': fields.float('Score',digits=(2,1)), 'category_id' = many2one('idea.category', 'Category'), } _defaults = { 'active': lambda *a: 1, # ideas are active by default 'state': lambda *a: 'draft', # ideas are in draft state by default } def _check_name(self,cr,uid,ids): for idea in self.browse(cr, uid, ids): if 'spam' in idea.name: return False # Can't create ideas with spam! return True _sql_constraints = [('name_uniq','unique(name)', 'Idea must be unique!')] _constraints = [(_check_name,'Please avoid spam in ideas !', ['name'])] idea() # Instantiate the class

Predefined osv.osv attributes for business objects

_name (required) _columns (required) _defaults _auto _inherit _inherits

business object name, in dot-notation (in module namespace) dictionary {field names object fields declarations } dictionary: { field names functions providing defaults }
_defaults['name'] = lambda self,cr,uid,context: 'eggs'

The __init__.py file is the Python module descriptor, because an OpenERP module is also a regular Python module.
16 17

__init__.py:

if True (default) the ORM will create the database table set to False to create your own table/view within the init() method
_name of the parent business object (for prototype inheritance)

# Import all files & directories containing python code import idea, wizard, report

Installing from source

There are two alternatives: using a tarball provided on the website, or directly getting the source using Bazaar (distributed Source Version

The __terp__.py (or __openerp__.py as of v5.2) is the OpenERP descriptor and contains a single Python dictionary with the actual declaration of the module: its name, dependencies, description, and composition.
18 19

for multiple / instance inheritance mechanism: dictionary mapping the _name of the parent business objects to the names of the corresponding foreign key fields to use list of tuples defining the Python constraints, in the form
(func_name, message, fields). (70)

__terp__.py:
{

_constraints

'name' : 'Idea',

Copyright 2010 Open Object Press - All rights reserved See license on page 7.

p.1/7

Predefined osv.osv attributes for business objects

_sql_constraints _log_access

ORM fields types

boolean(...) integer(...) date(...) datetime(...) time(...) char(string,size,translate=False,..) text(string, translate=False, )
'active': fields.boolean('Active'), 'priority': fields.integer('Priority'), 'start_date': fields.date('Start Date'),

ORM fields types

property(obj, type='float', view_load=None, group_name=None, )

list of tuples defining the SQL constraints, in the form

(name, sql_def, message). (69)

If True (default), 4 fields (create_uid, create_date, write_uid, write_date) will be used to log record-level operations, made accessible via osv's perm_read() function Name of the field used to sort the records in lists (default: 'id') Alternative field to use as name, used by osv's name_get() (default: 'name') SQL code to create the table/view for this object (if _auto is False) can be replaced by SQL execution in the init() method SQL table name to use (default: _name with dots '.' replaced by underscores '_')

Text-based fields
float(string, digits=None, ...)

_order _rec_name _sql _table

translate: True if field values can be translated by users size: maximum size for char fields (41,45) digits: tuple (precision, scale) (58) . If digits

Dynamic attribute with specific access rights obj: object (required) type: type of equivalent field
one2many many2one are symmetric

Tip: relational fields symmetry

Floating-point value with arbitrary precision and scale Field allowing selection among a set of predefined values

is not provided, it's a float, not a decimal type.

values: list of values (key-label tuples) or function returning such a list (required) (42) filters: optional filename filters

many2many many2many are symmetric when inversed (swap field1 and field2) one2many many2one + many2one one2many = many2many

selection(values, string, ...)

Special / Reserved field names

binary(string, filters=None, ...)

A few field names are reserved for pre-defined behavior in OpenObject. Some of them are created automatically by the system, and in that case any field with that name will be ignored. id name active sequence state parent_id

Inheritance mechanisms

Field for storing a file or binary content.

reference(string, selection, size,..)

'picture': fields.binary('Picture', filters='*.png,*.gif')

unique system identifier for the object (created by ORM, do not add it) defines the value used by default to display the record in lists, etc. if missing, set _rec_name to specify another field to use for this purpose defines visibility: records with active set to False are hidden by default defines order and allows drag&drop reordering if included in list views defines life-cycle stages for the object, used for workflows defines tree structure on records, and enables child_of operator

Field with dynamic relationship to any other object, associated with an assistant widget

selection: model _name of allowed objects types and corresponding label (same format as values for selection fields) (required) size: size of text column used to store it (as text: 'model_name,object_id' ) (required)
'contact': fields.reference('Contact',[ ('res.partner','Partner'), ('res.partner.contact','Contact')], 128)

Relational fields

Common attributes supported by relational fields

many2one(obj, ondelete='set null', ) (50)

domain: optional restriction in the form of arguments for search (see search()) obj: _name of destination object (required) ondelete: deletion handling, e.g. 'set null ', 'cascade', see PostgreSQL documentation obj: _name of destination object (required) field_id: field name of inverse many2one, i.e.

parent_left , used in conjunction with _parent_store flag on object, allows faster parent_right access to tree structures (see also Performance Optimization section) create_date, used to log creator, last updater, date of creation and last update date of create_uid , the record. disabled if _log_access flag is set to False write_date, (created by ORM, do not add them) write_uid

Relationship towards a parent object (using a foreign key) Virtual relationship towards multiple objects (inverse of many2one)

one2many(obj, field_id, ) (55)

Working with the ORM

corresponding foreign key (required)

obj: _name of destination object (required) rel: relationship table to use (required) field1: name of field in rel table storing the id

many2many(obj, rel, field1, field2, ) (56)

Inheriting from the osv.osv class makes all the ORM methods available on business objects. These methods may be invoked on the self object within the Python class itself (see examples in the table below), or from outside the class by first obtaining an instance via the ORM pool system.
72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87

Bidirectional multiple relationship between objects

ORM usage sample

of the current object (required) field2: name of field in rel table storing the id of the target object (required)
Functional fields

ORM field types

Objects may contain 3 types of fields: simple, relational, and functional. Simple types are integers, floats, booleans, strings, etc. Relational fields represent the relationships between objects (one2many, many2one, many2many). Functional fields are not stored in the database but calculated on-the-fly as Python functions. Relevant examples in the idea class above are indicated with the corresponding line numbers (XX,XX)

function(fnct, arg=None, fnct_inv=None, fnct_inv_arg=None, type='float', fnct_search=None, obj=None, method=False, store=False, multi=False,)

Functional field simulating a real field, computed rather than stored

ORM fields types

Common attributes supported by all fields (optional unless specified)

string: field label (required) required: True if mandatory readonly: True if not editable help: help tooltip select: 1 to include in search

views and optimize for list filtering (with database index)

context: dictionary with contextual parameters (for relational fields) change_default: True if field should be usable as condition for default values in clients states: dynamic changes to this field's common attributes based on the state field (42,46) Simple fields

fnct: function to compute the field value (required) def fnct(self, cr, uid, ids, field_name, arg, context) returns a dictionary { idsvalues } with values of type type fnct_inv: function used to write a value in the field instead def fnct_inv(obj, cr, uid, id, name, value, fnct_inv_arg, context) type: type of simulated field (any other type besides 'function') fnct_search: function used to search on this field def fnct_search(obj, cr, uid, obj, name, args) returns a list of tuples arguments for search(), e.g. [('id','in',[1,3,5])] obj: model _name of simulated field if it is a relational field store, multi: optimization mechanisms (see usage in Performance Section)

class idea2(osv.osv): idea2 _name = 'idea.idea' _inherit = 'idea.idea' def _score_calc(self,cr,uid,ids,field,arg,context=None): res = {} # This loop generates only 2 queries thanks to browse()! for idea in self.browse(cr,uid,ids,context=context): sum_vote = sum([v.vote for v in idea.vote_ids]) avg_vote = sum_vote/len(idea.vote_ids) res[idea.id] = avg_vote return res _columns = { # Replace static score with average of votes 'score':fields.function(_score_calc,type='float',method=True) } idea2()

OSV generic accessor

ORM Methods on osv.osv objects self.pool.get('object_name') may be used to

obtain a model class from anywhere

Common parameters, used by multiple methods

related(f1, f2, , type='float', ) Shortcut field equivalent to browsing chained fields f1,f2,...: chained fields to reach target (f1 required) (51) type: type of target field

cr: database connection (cursor) uid: id of user performing the operation ids: list of record ids, or single integer when

there is only one id context: optional dictionary of contextual parameters, such as user language e.g. { 'lang': 'en_US', ... }

Copyright 2010 Open Object Press - All rights reserved See license on page 7.

p.2/7

context=None)

ORM Methods on osv.osv objects create(cr, uid, values, values: dictionary of field values for the
record Creates a new record with the specified value Returns: id of the new record
search(cr, uid, args, offset=0, limit=None, order=None, context=None, count=False)
idea_id = self.create(cr, uid, { 'name': 'Spam recipe', 'description' : 'spam & eggs', 'inventor_id': 45, })

ORM Methods on osv.osv objects fields_get(cr, uid, fields=None, fields: list of field names
context=None)

Returns a dictionary of field dictionaries, each one describing a field of the business object
fields_view_get(cr, uid, view_id=None, view_type='form', context=None, toolbar=False)

class idea(osv.osv): idea (...) _columns = { 'name' : fields.char('Name',size=64) (...) def test_fields_get(self,cr,uid): assert(self.fields_get('name')['size'] == 64)

96 97 98 99 100 101

<record model="object_model_name2" id="object_xml_id2"> <field name="field1" ref="module.object_xml_id"/> <field name="field2" eval="ref('module.object_xml_id')"/> </record> </data> </openerp>

Each type of record (view, menu, action) support a specific set of child entities and attributes, but all share the following special attributes:
id ref the unique (per module) XML identifier of this record (xml_id) used instead of element content to reference another record (works cross-module by prepending the module name)

Returns: list of ids of records matching the given criteria

args: list of tuples specifying search criteria offset: optional number of records to skip limit : optional max number of records to

return
order: optional columns to sort by (default: self._order ) count: if True, returns only the number of

Returns a dictionary describing the composition of the requested view (including inherited views and extensions)
name_get(cr, uid, ids, context={})

view_id: id of the view or None view_type: type of view to return if view_id is None ('form','tree', ...) toolbar: True to include contextual actions
def test_fields_view_get(self,cr,uid): idea_obj = self.pool.get('idea.idea') form_view = idea_obj.fields_view_get(cr,uid) # Ideas should be shown with invention date def name_get(self,cr,uid,ids): res = [] for r in self.read(cr,uid,ids['name','create_date']) res.append((r['id'], '%s (%s)' (r['name'],year)) return res

eval used instead of element content to provide value as a Python expression, that can use the ref() method to find the database id for a given xml_id Tip: XML RelaxNG validation OpenObject validates the syntax and structure of XML files, according to a RelaxNG grammar, found in server/bin/import_xml.rng. For manual check use xmllint: xmllint relaxng /path/to/import_xml.rng <file>

records matching the criteria, not their ids

#Operators: =, !=, >, >=, <, <=, like, ilike, #in, not in, child_of, parent_left, parent_right #Prefix operators: '&' (default), '|', '!' #Fetch non-spam partner shops + partner 34 ids = self.search(cr, uid, [ '|', ('partner_id', '!=', 34), '!', ('name', 'ilike', 'spam'),], order='partner_id' )

Returns tuples with the text representation of requested objects for to-many relationships

name_search(cr, uid, name='', args=None, operator='ilike', context=None, limit=80)

read(cr, user, ids, fields=None, context=None)

fields: optional list of field names to return (default: all fields)

results = self.read(cr, uid, [42,43], ['name', 'inventor_id']) print 'Inventor:', results[0]['inventor_id']

name: object name to search for operator : operator for name criterion args, limit: same as for search() )

Common CSV syntax CSV files can also be added in update_xml, and the records will be inserted by the OSV's import_data() method, using the CSV filename to determine the target object model. The ORM automatically reconnects relationships based on the following special column names:
id (xml_id) many2one_field many2one_field:id many2one_field.id many2many_field many2many_field:id many2many_field.id column containing identifiers for relationships reconnect many2one using name_search() reconnect many2one based on object's xml_id reconnect many2one based on object's database id reconnects via name_search(), repeat for multiple values reconnects with object's xml_id, repeat for multiple values reconnects with object's database id, repeat for multiple values

Returns: list of dictionaries with requested field values

write(cr, uid, ids, values, context=None)

values: dictionary of field values to update

Returns list of object names matching the criteria, used to provide completion for to-many relationships. Equivalent of search() on name + name_get()

Updates records with given ids with the given values. Returns: True
copy(cr, uid, id, defaults,context=None)

self.write(cr, uid, [42,43], { 'name': 'spam & eggs', 'partner_id': 24, })

defaults: dictionary of field values to change export_data(cr, uid, ids, fields, context=None)

Duplicates record with given id updating it with defaults values. Returns: True
unlink(cr, uid, ids, context=None)

before saving the duplicated object

# Countries can be searched by code or name def name_search(self,cr,uid,name='', args=[],operator='ilike',context={}, limit=80): ids = [] if name and len(name) == 2: ids = self.search(cr, user, [('code', '=', name)] + args, limit=limit, context=context) if not ids: ids = self.search(cr, user, [('name', operator, name)] + args, limit=limit, context=context) return self.name_get(cr,uid,ids)

102 103 104

one2many_field/field creates one2many destination record and sets field value ir.model.access.csv
"id","name","model_id:id","group_id:id","perm_read","perm_write","perm_create","perm_unlink" "access_idea_idea","idea.idea","model_idea_idea","base.group_user",1,0,0,0 "access_idea_vote","idea.vote","model_idea_vote","base.group_user",1,0,0,0

fields: list of field names context may contain import_comp (default:

Menus and actions

self.unlink(cr, uid, [42,43])

Deletes records with the given ids Returns: True

browse(cr, uid, ids, context=None)

Exports fields for selected objects, returning a dictionary with a datas matrix. Used when exporting data via client menu.
import_data(cr, uid, fields, data, mode='init', current_module='', noupdate=False, context=None, filename=None)

False) to make exported data compatible with import_data() (may prevent exporting some fields)

Actions are declared as regular records and can be triggered in 3 ways: by clicking on menu items linked to a specific action by clicking on buttons in views, if these are connected to actions as contextual actions on an object
105 106 107 108 109 110 111 112 113 114 115

Fetches records as objects, allowing to use dot-notation to browse fields and relations Returns: object or list of objects requested
default_get(cr, uid, fields, context=None)

idea = self.browse(cr, uid, 42) print 'Idea description:', idea.description print 'Inventor country code:', idea.inventor_id.address[0].country_id.code for vote in idea.vote_ids: print 'Vote %2.2f' % vote.vote

Imports given data in the given module Used when exporting data via client menu

fields: list of field names data: data to import (see export_data()) mode: 'init' or 'update' for record creation current_module : module name noupdate: flag for record creation filename: optional file to store partial import

Action declaration
<record model="ir.actions.act_window" id="action_id"> <field name="name">action.name</field> <field name="view_id" ref="view_id"/> <field name="domain">[list of 3-tuples (max 250 characters)]</field> <field name="context">{context dictionary (max 250 characters)}</field> <field name="res_model">object.model.name</field> <field name="view_type">form|tree</field> <field name="view_mode">form,tree,calendar,graph</field> <field name="target">new</field> <field name="search_view_id" ref="search_view_id"/> </record>

fields: list of field names

state for recovery

Returns: a dictionary of the default values for fields (set on the object class, by the user preferences, or via the context)
perm_read(cr, uid, ids, details=True)

defs = self.default_get(cr,uid, ['name','active']) # active should be True by default assert defs['active']

Tip: use read() through webservice calls, but always browse() internally

Building the module interface

Common XML structure
88 89 90 91 92 93 94 95

details : if True, *_uid fields are replaced

Returns: a list of ownership dictionaries for each requested record

with the name of the user returned dictionaries contain: object id (id), creator user id (create_uid ), creation date (create_date), updater user id (write_uid ), update date (write_date)
perms = self.perm_read(cr,uid,[42,43]) print 'creator:', perms[0].get('create_uid', 'n/a')

To construct a module, the main mechanism is to insert data records declaring the module interface components. Each module element is a regular data record: menus, views, actions, roles, access rights, etc. XML files declared in a module's update_xml attribute contain record declarations in the following form:
<?xml version="1.0" encoding="utf-8"?> <openerp> <data> <record model="object_model_name" id="object_xml_id"> <field name="field1">value1</field> <field name="field2">value2</field> </record>

id name view_id domain context res_model view_type view_mode target search_view_id

identifier of the action in table ir.actions.act_window, must be unique action name (required) specific view to open (if missing, highest priority view of given type is used) tuple (see search() arguments) for filtering the content of the view context dictionary to pass to the view object model on which the view to open is defined set to form to open records in edit mode, set to tree for a tree view only if view_type is form, list allowed modes for viewing records (form, tree, ...) set to new to open the view in a new window identifier of the search view to replace default search form (new in version 5.2)

116 117

Menu declaration The menuitem entity is a shortcut for declaring an ir.ui.menu record and connect it with a corresponding action via an ir.model.data record.
<menuitem id="menu_id" parent="parent_menu_id" name="label" icon="icon-code" action="action_id" groups="groupname1,groupname2" sequence="10"/>

id

identifier of the menuitem, must be unique

Copyright 2010 Open Object Press - All rights reserved See license on page 7.

p.3/7

parent name action icon groups sequence

id of the parent menu in the hierarchy Optional menu label (default: action name) identifier of action to execute, if any icon to use for this menu (e.g. terp-graph, STOCK_OPEN, see doc.opernerp.com) list of groups that can see this menu item (if missing, all groups can see it) integer index for ordering sibling menuitems (10,20,30..)

Form Elements

Calendars
Attributes

Views and inheritance

Generic view declaration

Views form a hierarchy. Several views of the same type can be declared on the same object, and will be used depending on their priorities. By declaring an inherited view it is possible to add/remove features in a view.
118 119 120 121 122 123 124 125 126

<record model="ir.ui.view" id="view_id"> <field name="name">view.name</field> <field name="model">object_name</field> <field name="type">form</field> # tree,form,calendar,search,graph,gantt <field name="priority" eval="16"/> <field name="arch" type="xml"> <!-- view content: <form>, <tree>, <graph>, --> </field> </record>

id name model type priority arch

unique view identifier view name object model on which the view is defined (same as res_model in actions) view type: form, tree, graph, calendar, search, gantt (search is new in 5.2) view priority, smaller is higher (default: 16) architecture of the view, see various view types below

Forms
127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158

(to view/edit records) Forms allow creation/edition or resources, and correspond to <form> elements.

Allowed elements

all (see form elements below)

<form string="Idea form"> <group col="6" colspan="4"> <group colspan="5" col="6"> <field name="name" select="1" colspan="6"/> <field name="inventor_id" select="1"/> <field name="inventor_country_id" /> <field name="score" select="2"/> </group> <group colspan="1" col="2"> <field name="active"/><field name="invent_date"/> </group> </group> <notebook colspan="4"> <page string="General"> <separator string="Description"/> <field colspan="4" name="description" nolabel="1"/> </page> <page string="Votes"> <field colspan="4" name="vote_ids" nolabel="1" select="1"> <tree> <field name="partner_id"/> <field name="vote"/> </tree> </field> </page> <page string="Sponsors"> <field colspan="4" name="sponsor_ids" nolabel="1" select="1"/> </page> </notebook> <field name="state"/> <button name="do_confirm" string="Confirm" icon="gtk-ok" type="object"/> </form>

Common attributes for all elements: string: label of the element nolabel: 1 to hide the field label colspan: number of column on which the field must span rowspan: number of rows on which the field must span col: number of column this element must allocate to its child elements invisible: 1 to hide this element completely eval: evaluate this Python code as element content (content is string by default) attrs: Python map defining dynamic conditions on these attributes: readonly, invisible , required based on search tuples on other field values field automatic widgets depending on the corresponding field type. Attributes: string: label of the field, also for search (overrides field name) select: 1 to show the field in normal search, 2 for advanced only nolabel: 1 to hide the field label required: override required field attribute readonly: override readonly field attribute password: True to hide characters typed in this field context: Python code declaring a context dictionary domain: Python code declaring list of tuples for restricting values on_change: Python method call to trigger when value is changed completion: 1 to enable auto-completion of values when possible groups: comma-separated list of group (id) allowed to see this field widget: select alternative widget (url, email, image, float_time, reference, text_wiki, text_html, progressbar) properties dynamic widget showing all available properties (no attribute) button clickable widget associated with actions. Specific attributes: type: type of button: workflow (default), object, or action name: workflow signal, function name (without parentheses) or action to call (depending on type) confirm: text of confirmation message when clicked states: comma-separated list of states in which this button is shown icon: optional icon (all GTK STOCK icons e.g. gtk-ok) separator horizontal separator line for structuring views, with optional label newline place-holder for completing the current line of the view label free-text caption or legend in the form group used to organise fields in groups with optional label (adds frame) notebook, notebook elements are tab containers for page elements. Attributes: page name: label for the tab/page position : tabs position in notebook (inside, top, bottom, left, right)

Views used to display date fields as calendar events (<calendar> parent)

color: name of field for color segmentation date_start: name of field containing event start date/time day_length: length of a calendar day in hours (default: 8) date_stop: name of field containing event stop date/time

or
date_delay: name of field containing event duration

Allowed elements
163 164 165

field (to define the label for each calendar event)

<calendar string="Ideas" date_start="invent_date" color="inventor_id"> <field name="name"/> </calendar>

Gantt Charts

Bar chart typically used to show project schedule (<gantt> parent element) same as <calendar> Attributes

Allowed elements

field, level

level elements are used to define the Gantt chart levels, with the enclosed field used as label for that drill-down level
166 167 168 169 170

<gantt string="Ideas" date_start="invent_date" color="inventor_id"> <level object="idea.idea" link="id" domain="[]"> <field name="inventor_id"/> </level> </gantt>

Charts (Graphs)
Attributes

Views used to display statistical charts (<graph> parent element)

Tip: charts are most useful with custom views extracting ready-to-use statistics

type: type of chart: bar, pie (default) orientation : horizontal, vertical

Allowed elements field, with specific behavior: first field in view is X axis, 2nd one is Y, 3rd one is Z 2 fields required, 3rd one is optional group attribute defines the GROUP BY field (set to 1) operator attribute sets the aggregation operator to use for other fields when one field is grouped (+,*,**,min,max)
171 172 173 174

<graph string="Total idea score by Inventor" type="bar"> <field name="inventor_id" /> <field name="score" operator="+"/> </graph>

Search views (new in v5.2)

Dynamic views

In addition to what can be done with states and attrs attributes, functions may be called by view elements (via buttons of type object, or on_change attributes on fields) to obtain dynamic behavior. These functions may alter the view interface by returning a Python map with the following entries:

Search views are used to customize the search panel on top of list views, and are declared with the search type, and a top-level <search> element. After defining a search view with a unique id, add it to the action opening the list view using the search_view_id field in its declaration. Allowed elements field, group, separator, label, search, filter, newline, properties

175 176 177 178 179 180 181 182 183 184 185 186 187

filter elements allow defining button for domain filters adding a context attribute to fields makes widgets that alter the

value domain

a dictionary of field names and their updated values a dictionary of field names and their updated domains of value

search context (useful for context-sensitive fields, e.g. pricelistdependent prices)

<search string="Search Ideas"> <group col="6" colspan="4"> <filter string="My Ideas" icon="terp-partner" domain="[('inventor_id','=',uid)]" help="My own ideas"/> <field name="name" select="1"/> <field name="description" select="1"/> <field name="inventor_id" select="1"/> <!-- following context field is for illustration only --> <field name="inventor_country_id" select="1" widget="selection" context="{'inventor_country': self}"/> </group> </search>

warning a dictionary with a title and message to show a warning dialog

Lists/Trees
Attributes

Lists include field elements, are created with type tree, and have a <tree> parent element.

hierarchies as a side toolbar (example: the menu) Allowed elements field, group, separator, tree, button, filter, newline
159 160 161 162

colors: list of colors mapped to Python conditions editable: top or bottom to allow in-place edit toolbar : set to True to display the top level of object

<tree string="Idea Categories" toolbar="1" colors="blue:state==draft"> <field name="name"/> <field name="state"/> </tree>

Copyright 2010 Open Object Press - All rights reserved See license on page 7.

p.4/7

View Inheritance

203

to generate the RML template file from a .sxw template -->

split_mode

Existing views should be modifying through inherited views, never directly. An inherited view references its parent view using the inherit_id field, and may add or modify existing elements in the view by referencing them through XPath expressions, specifying the appropriate position .
Tip: XPath reference can be found at www.w3.org/TR/xpath position inside: placed inside match (default) before: placed before match

id name string model rml, sxw, xml, xsl auto header groups menu keywords

188 189 190 191 192 193 194 195 196 197 198

replace: replace match

after: placed after match

unique report identifier name for the report (required) report title (required) object model on which the report is defined (required) path to report template sources (starting from addons), depending on report set to False to use a custom parser, by subclassing report_sxw.rml_parse and declaring the report as follows:
report_sxw.report_sxw(report_name, object_model,rml_path,parser=customClass)

kind

<!-- improved idea categories list --> <record id="idea_category_list2" model="ir.ui.view"> <field name="name">id.category.list2</field> <field name="model">ir.ui.view</field> <field name="inherit_id" ref="id_category_list"/> <field name="arch" type="xml"> <xpath expr="/tree/field[@name='description']" position="after"> <field name="idea_ids" string="Number of ideas"/> </xpath> </field> </record>

set to False to suppress report header (default: True) comma-separated list of groups allowed to view this report set to True to link the report with the Print icon (default: True) specify report type keyword (default: client_print_multi)

Reports

There are several report engines in OpenERP, to produce reports from different sources and in many formats.

204 205 206 207 208 209 210 211 212 213 214 215

Tip: RML User Guide: www.reportlab.com/docs/rml2pdf-userguide.pdf Example RML report extract:

<story> <blockTable style="Table"> <tr> <td><para style="Title">Idea name</para> </td> <td><para style="Title">Score</para> </td> </tr> <tr> <td><para>[[ repeatIn(objects,'o','tr') ]] [[ o.name ]]</para></td> <td><para>[[ o.score ]]</para></td> </tr> </blockTable> </story>

subflow_id action

logical behavior of this node regarding outgoing transitions: XOR: one valid transition necessary, send workitem on it (default) OR: send workitems on all valid transitions (0 or more), sequentially AND: send a workitem on all valid transitions at once (fork) type of action to perform when node is activated by a transition: dummy to perform no operation when activated (default) function to invoke a function determined by action subflow to execute the subflow with subflow_id, invoking action to determine the record id of the record for which the subflow should be instantiated. If action returns no result, the workitem is deleted. stopall to terminate the workflow upon activation if kind subflow, id of the subflow to execute (use ref attribute or search with a tuple) object method call, used if kind is function or subflow. This function should also update the state field of the object, e.g. for a function kind:
def action_confirmed(self, cr, uid, ids): self.write(cr, uid, ids, { 'state' : 'confirmed' }) # perform other tasks return True

Workflow Transitions (edges)

227 228 229 230 231 232 233

Conditions are evaluated in this order: role_id, signal, condition expression

<record id="trans_idea_draft_confirmed" model="workflow.transition"> <field name="act_from" ref="act_draft"/> <field name="act_to" ref="act_confirmed"/> <field name="signal">button_confirm</field> <field name="role_id" ref="idea_manager"/> <field name="condition">1 == 1</field> </record>

Workflows
Workflows may be associated with any object in OpenERP, and are entirely customizable. Workflows are used to structure and manage the lifecycles of business objects and documents, and define transitions, triggers, etc. with graphical tools. Workflows, activities (nodes or actions) and transitions (conditions) are declared as XML records, as usual. The tokens that navigate in workflows are called workitems.

act_from, act_to signal role_id condition

identifiers of the source and destination activities name of a button of type workflow that triggers this transition reference to the role that user must have to trigger the transition (see Roles) Python expression that must evaluate to True for transition to be triggered

Special expressions used inside report templates produce dynamic data and/or modify the report structure at rendering time. Custom report parsers may be written to support additional expressions.

Tip: The Web client features a graphical workflow editor, via the CustomiseManage Workflows link in lists and forms.

Alternative Report Formats (see doc.openerp.com)

sxw2rml rml xml,xsl:rml odt2odt mako OpenOffice 1.0 templates (.sxw) converted to RML with sxw2rml tool, and the RML rendered in HTML or PDF RML templates rendered directly as HTML or PDF XML data + XSL:RML stylesheets to generate RML OpenOffice templates (.odt) used to produce directly OpenOffice documents (.odt) (As of OpenERP 5.2) Mako template library used to produce HTML output, by embedding Python code and OpenERP expressions within any text file (As of OpenERP 5.2) double brackets content is evaluated as a Python expression based on the following expressions

Security

Access control mechanisms must be combined to achieve a coherent security policy.

Workflow declaration
216 217 218 219 220

Workflows are declared on objects that possess a state field (see the example idea class in the ORM section)
<record id="wkf_idea" model="workflow"> <field name="name">idea.basic</field> <field name="osv">idea.idea</field> <field name="on_create" eval="1"/> </record>

Group-based access control mechanisms

Groups are created as normal records on the res.groups model, and granted menu access via menu definitions. However even without a menu, objects may still be accessible indirectly, so actual object-level permissions (create,read,write,unlink) must be defined for groups. They are usually inserted via CSV files inside modules. It is also possible to restrict access to specific fields on a view or object using the field's groups attribute.
234 235 236

Expressions used in OpenERP report templates

[[ <content> ]] Predefined expressions: objects contains the list of records to print data comes from the wizard launching the report user contains the current user (as per browse()) time gives access to Python time module repeatIn(list,'var','tag') repeats the current parent element named tag for each object in list , making the object available as var during each loop setTag('tag1','tag2') replaces the parent RML tag1 with tag2 removeParentNode('tag') removes parent RML element tag formatLang(value, digits=2, date=False, date_time=False, grouping=True, monetary=False) can be used to format a date, time or amount according

id name osv on_create

unique workflow record identifier name for the workflow (required) object model on which the workflow is defined (required) if True, a workitem is instantiated automatically for each new osv record

ir.model.access.csv
"id","name","model_id:id","group_id:id","perm_read","perm_write","perm_create","perm_unlink" "access_idea_idea","idea.idea","model_idea_idea","base.group_user",1,1,1,0 "access_idea_vote","idea.vote","model_idea_vote","base.group_user",1,1,1,0

Roles

221 222 223 224 225 226

Workflow Activities (nodes)

<record id="act_confirmed" model="workflow.activity"> <field name="name">confirmed</field> <field name="wkf_id" ref="wkf_idea"/> <field name="kind">function</field> <field name="action">action_confirmed()</field> </record>

Roles are created as normal records on the res.roles model and used only to condition workflow transitions through transitions' role_id attribute.

Wizards
Wizards describe stateful interactive sessions with the user through dynamic forms. As of OpenERP v5.0, wizards make use of the osv_memory in-memory persistence to allow constructing wizards from regular business objects and views.

to the locale
setLang('lang_code') sets the current language and locale for translations Report declaration
<!-- The following creates records in ir.actions.report.xml model --> <report id="idea_report" string="Print Ideas" model="idea.idea" name="idea.report" rml="idea/report/idea.rml" > <!-- Use addons/base_report_designer/wizard/tiny_sxw2rml/tiny_sxw2rml.py

id wkf_id name flow_start flow_stop join_mode

unique activity identifier parent workflow identifier activity node label True to make it a 'begin' node, receiving a workitem for each workflow instance True to make it an 'end' node, terminating the workflow when all items reach it logical behavior of this node regarding incoming transitions: XOR: activate on the first incoming transition (default) AND: waits for all incoming transitions to become valid

Wizard objects (osv_memory)

237 238 239 240 241 242 243 244

In-memory objects are created by extending osv.osv_memory:

from osv import fields,osv import datetime class cleanup_wizard(osv.osv_memory): cleanup_wizard _name = 'idea.cleanup.wizard' _columns = { 'idea_age': fields.integer('Age (in days)'), } def cleanup(self,cr,uid,ids,context={}):

199 200 201 202

Copyright 2010 Open Object Press - All rights reserved See license on page 7.

p.5/7

245 246 247 248 249 250 251 252 253 254

idea_obj = self.pool.get('idea.idea') for wiz in self.browse(cr,uid,ids): if wiz.idea_age <= 3: raise osv.except_osv('UserError','Please select a larger age') limit = datetime.date.today()-datetime.timedelta(days=wiz.idea_age) ids_to_del = idea_obj.search(cr,uid, [('create_date', '<' , limit.strftime('%Y-%m-%d 00:00:00'))],context=context) idea_obj.unlink(cr,uid,ids_to_del) return {} cleanup_wizard()

317 318 319 320

$msg->addParam(new xmlrpcval("create", "string")); $msg->addParam(new xmlrpcval($arrayVal, "struct")); $resp = $client->send($msg); ?>

Performance Optimization

Internationalization

Views
255 256 257 258 259 260 261 262 263 264 265 266 267 268 269

Wizards use regular views and their buttons may use a special cancel attribute to close the wizard window when clicked.
<record id="wizard_idea_cleanup" model="ir.ui.view"> <field name="name">idea.cleanup.wizard.form</field> <field name="model">idea.cleanup.wizard</field> <field name="type">form</field> <field name="arch" type="xml"> <form string="Idea Cleanup Wizard"> <label colspan="4" string="Select the age of ideas to cleanup"/> <field name="idea_age" string="Age (days)"/> <group colspan="4"> <button string="Cancel" special="cancel" icon="gtk-cancel"/> <button string="Cleanup" name="cleanup" type="object" icon="gtk-ok"/> </group> </form> </field> </record>

Each module can provide its own translations within the i18n directory, by having files named LANG.po where LANG is the locale code for the language, or the language and country combination when they differ (e.g. pt.po or pt_BR.po). Translations will be loaded automatically by OpenERP for all enabled languages. Developers always use English when creating a module, then export the module terms using OpenERP's gettext POT export feature (Administration>Translations>Export a Translation File without specifying a language), to create the module template POT file, and then derive the translated PO files. Many IDE's have plugins or modes for editing and merging PO/POT files.
Tip: The GNU gettext format (Portable Object) used by OpenERP is integrated into LaunchPad, making it an online collaborative translation platform.
321 322 323 324 325 326

Wizard execution
270 271 272 273 274 275 276 277

Such wizards are launched via regular action records, with a special target field used to open the wizard view in a new window.
<record id="action_idea_cleanup_wizard" model="ir.actions.act_window"> <field name="name">Cleanup</field> <field name="type">ir.actions.act_window</field> <field name="res_model">idea.cleanup.wizard</field> <field name="view_type">form</field> <field name="view_mode">form</field> <field name="target">new</field> </record>

|- idea/ |- i18n/ | - idea.pot | - fr.po | - pt_BR.po | (...)

# # # # #

The module directory Translation files Translation Template (exported from OpenERP) French translation Brazilian Portuguese translation

As Enterprise Management Software typically has to deal with large amounts of records, you may want to pay attention to the following antipatterns, to obtain consistent performance: Do not place browse() calls inside loops, put them before and access only the browsed objects inside the loop. The ORM will optimize the number of database queries based on the browsed attributes. Avoid recursion on object hierarchies (objects with a parent_id relationship), by adding parent_left and parent_right integer fields on your object, and setting _parent_store to True in your object class. The ORM will use a modified preorder tree traversal to be able to perform recursive operations (e.g. child_of ) with database queries in O(1) instead of O(n) Do not use function fields lightly, especially if you include them in tree views. To optimize function fields, two mechanisms are available: multi: all fields sharing the same multi attribute value will be computed with one single call to the function, which should then return a dictionary of values in its values map store: function fields with a store attribute will be stored in the database, and recomputed on demand when the relevant trigger objects are modified. The format for the trigger specification is as follows: store = {'model': (_ref_fnct, fields, priority)} (see example below)
327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349

Tip: By default OpenERP's POT export only extracts labels inside XML files or inside field definitions in Python code, but any Python string can be translated this way by surrounding it with the tools.translate._ method (e.g. _('Label') )

Rapid Application Development

Module recorder
The base_module_record module can be used to export a set of changes in the form of a new module. It should be used for all customizations that should be carried on through migrations and updates. It has 2 modes: Start/Pause/Stop mode, where all operations (on business objects or user interface) are recorded until the recorder is stopped or paused. Date- and model-based mode where all changes performed after a given date on the given models (object types) are exported. .

WebServices XML-RPC
OpenERP is accessible through XML-RPC interfaces, for which libraries exist in many languages.
278 279 280 281 282 283 285 284 286 287 288 289 290 291 292 293

Python example
import xmlrpclib # ... define HOST, PORT, DB, USER, PASS url = 'http://%s:%d/xmlrpc/common' % (HOST,PORT) sock = xmlrpclib.ServerProxy(url) uid = sock.login(DB,USER,PASS) print "Logged in as %s (uid:%d)" % (USER,uid) # Create a new idea url = 'http://%s:%d/xmlrpc/object' % (HOST,PORT) sock = xmlrpclib.ServerProxy(url) args = { 'name' : 'Another idea', 'description' : 'This is another idea of mine', 'inventor_id': uid, } idea_id = sock.execute(DB,uid,PASS,'idea.idea','create',args)

def _get_idea_from_vote(self,cr,uid,ids,context={}): res = {} vote_ids = self.pool.get('idea.vote').browse(cr,uid,ids,context=context) for v in vote_ids: res[v.idea_id.id] = True # Store the idea identifiers in a set return res.keys() def _compute(self,cr,uid,ids,field_name,arg,context={}): res = {} for idea in self.browse(cr,uid,ids,context=context): vote_num = len(idea.vote_ids) vote_sum = sum([v.vote for v in idea.vote_ids]) res[idea.id] = { 'vote_sum': vote_sum, 'vote_avg': (vote_sum/vote_num) if vote_num else 0.0, } return res _columns = { # These fields are recomputed whenever one of the votes changes 'vote_avg': fields.function(_compute, method=True, string='Votes Average', store = {'idea.vote': (_get_idea_from_vote,['vote'],10)},multi='votes'), 'vote_sum': fields.function(_compute, method=True, string='Votes Sum', store = {'idea.vote': (_get_idea_from_vote,['vote'],10)},multi='votes'), }

Report Creator (view) and Report Designer (print) modules

Community / Contributing
OpenERP projects are hosted on LaunchPad(LP), where all project resources may be found: Bazaar branches, bug tracking, blueprints, roadmap, FAQs, etc. Create a free account on launchpad.net to be able to contribute.

The base_report_creator module can be used to automate the creation of custom statistics views, e.g. to construct dashboards. The resulting dashboards can then be exported using the base_module_record module.

Launchpad groups
Group* OpenERP Quality Team (~openerp) OpenERP Commiters (~openerp-commiter) OpenERP Drivers (~openerp-drivers) OpenERP Community (~openerp-community) Members OpenERP Core Team Selected active community members Selected active community members Open group, anyone can join Bazaar/LP restrictions Can merge and commit on official branches. Can mark branches to be merged into official branch. Can commit on extra-addons branch Can confirm bugs and set milestones on bugs / blueprints Can create community branches where everyone can contribute

294 295 296 297 298 299 300 301 302 303 304 306 305 307 308 309 310 311 312 313 314 315 316

PHP example
<? include('xmlrpc.inc'); // Use phpxmlrpc library, available on sourceforge // ... define $HOST, $PORT, $DB, $USER, $PASS $client = new xmlrpc_client("http://$HOST:$PORT/xmlrpc/common"); $msg = new xmlrpcmsg("login"); $msg->addParam(new xmlrpcval($DB, "string")); $msg->addParam(new xmlrpcval($USER, "string")); $msg->addParam(new xmlrpcval($PASS, "string")); resp = $client->send($msg); uid = $resp->value()->scalarval() echo "Logged in as $USER (uid:$uid)" // Create a new idea $arrayVal = array( 'name'=>new xmlrpcval("Another Idea", "string") , 'description'=>new xmlrpcval("This is another idea of mine" , "string"), 'inventor_id'=>new xmlrpcval($uid, "int"), ); $msg = new xmlrpcmsg('execute'); $msg->addParam(new xmlrpcval($DB, "string")); $msg->addParam(new xmlrpcval($uid, "int")); $msg->addParam(new xmlrpcval($PASS, "string")); $msg->addParam(new xmlrpcval("idea.idea", "string"));

*Members of upper groups are also members of lower groups

Copyright 2010 Open Object Press - All rights reserved See license on page 7.

p.6/7

License
Copyright 2010 Open Object Press. All rights reserved. You may take electronic copy of this work and distribute it if you don't change the content. You can also print a copy to be read by yourself only. We have contracts with different publishers in different countries to sell and distribute paper or electronic based versions of this work (translated or not) in bookstores. This helps to distribute and promote the Open ERP product. It also helps us to create incentives to pay contributors and authors with the royalties. Due to this, grants to translate, modify or sell this work are strictly forbidden, unless OpenERP s.a. (representing Open Object Press) gives you a written authorization for this. While every precaution has been taken in the preparation of this work, the publisher and the authors assume no responsibility for errors or omissions, or for damages resulting from the use of the information contained herein. Published by Open Object Press, Grand Rosire, Belgium

Copyright 2010 Open Object Press - All rights reserved See license on page 7.

p.7/7