Getopt::Long - Extended processing of command line options
use Getopt::Long;
$result = GetOptions (...option-descriptions...);The Getopt::Long module implements an extended getopt function called GetOptions(). This function adheres to the POSIX syntax for command line options, with GNU extensions. In general, this means that options have long names instead of single letters, and are introduced with a double dash "--". Support for bundling of command line options, as was the case with the more traditional single-letter approach, is provided but not enabled by default.
Command line operated programs traditionally take their arguments from the command line, for example filenames or other information that the program needs to know. Besides arguments, these programs often take command line options as well. Options are not necessary for the program to work, hence the name 'option', but are used to modify its default behaviour. For example, a program could do its job quietly, but with a suitable option it could provide verbose information about what it did.
Command line options come in several flavours. Historically, they are preceded by a single dash -, and consist of a single letter.
-l -a -cUsually, these single-character options can be bundled:
-lacOptions can have values, the value is placed after the option character. Sometimes with whitespace in between, sometimes not:
-s 24 -s24Due to the very cryptic nature of these options, another style was developed that used long names. So instead of a cryptic -l one could use the more descriptive --long. To distinguish between a bundle of single-character options and a long one, two dashes are used to precede the option name. Early implementations of long options used a plus + instead. Also, option values could be specified either like
--size=24or
--size 24The + form is now obsolete and strongly deprecated.
Getopt::Long is the Perl5 successor of newgetopt.pl. This was the firs Perl module that provided support for handling the new style of command line options, hence the name Getopt::Long. This module also supports single-character options and bundling. In this case, the options are restricted to alphabetic characters only, and the characters ? and -.
To use Getopt::Long from a Perl program, you must include the following line in your Perl program:
use Getopt::Long;This will load the core of the Getopt::Long module and prepare your program for using it. Most of the actual Getopt::Long code is not loaded until you really call one of its functions.
In the default configuration, options names may be abbreviated to uniqueness, case does not matter, and a single dash is sufficient, even for long option names. Also, options may be placed between non-option arguments. See "Configuring Getopt::Long" for more details on how to configure Getopt::Long.
The most simple options are the ones that take no values. Their mere presence on the command line enables the option. Popular examples are:
--all --verbose --quiet --debugHandling simple options is straightforward:
my $verbose = ''; # option variable with default value (false)
my $all = ''; # option variable with default value (false)
GetOptions ('verbose' => \$verbose, 'all' => \$all);The call to GetOptions() parses the command line arguments that are present in @ARGV and sets the option variable to the value 1 if the option did occur on the command line. Otherwise, the option variable is not touched. Setting the option value to true is often called enabling the option.
The option name as specified to the GetOptions() function is called the option specification. Later we'll see that this specification can contain more than just the option name. The reference to the variable is called the option destination.
GetOptions() will return a true value if the command line could be processed successfully. Otherwise, it will write error messages to STDERR, and return a false result.
Getopt::Long supports two useful variants of simple options: negatable options and incremental options.
A negatable option is specified with a exclamation mark ! after the option name:
my $verbose = ''; # option variable with default value (false)
GetOptions ('verbose!' => \$verbose);Now, using --verbose on the command line will enable $verbose, as expected. But it is also allowed to use --noverbose, which will disable $verbose by setting its value to 0. Using a suitable default value, the program can find out whether $verbose is false by default, or disabled by using --noverbose.
An incremental option is specified with a plus + after the option name:
my $verbose = ''; # option variable with default value (false)
GetOptions ('verbose+' => \$verbose);Using --verbose on the command line will increment the value of $verbose. This way the program can keep track of how many times the option occurred on the command line. For example, each occurrence of --verbose could increase the verbosity level of the program.
Usually programs take command line options as well as other arguments, for example, file names. It is good practice to always specify the options first, and the other arguments last. Getopt::Long will, however, allow the options and arguments to be mixed and 'filter out' all the options before passing the rest of the arguments to the program. To stop Getopt::Long from processing further arguments, insert a double dash -- on the command line:
--size 24 -- --allIn this example, --all will not be treated as an option, but passed to the program unharmed, in @ARGV.
For options that take values it must be specified whether the option value is required or not, and what kind of value the option expects.
Three kinds of values are supported: integer numbers, floating point numbers, and strings.
If the option value is required, Getopt::Long will take the command line argument that follows the option and assign this to the option variable. If, however, the option value is specified as optional, this will only be done if that value does not look like a valid command line option itself.
my $tag = ''; # option variable with default value
GetOptions ('tag=s' => \$tag);In the option specification, the option name is followed by an equals sign = and the letter s. The equals sign indicates that this option requires a value. The letter s indicates that this value is an arbitrary string. Other possible value types are i for integer values, and f for floating point values. Using a colon : instead of the equals sign indicates that the option value is optional. In this case, if no suitable value is supplied, string valued options get an empty string '' assigned, while numeric options are set to 0.
Options sometimes take several values. For example, a program could use multiple directories to search for library files:
--library lib/stdlib --library lib/extlibTo accomplish this behaviour, simply specify an array reference as the destination for the option:
my @libfiles = ();
GetOptions ("library=s" => \@libfiles);Used with the example above, @libfiles would contain two strings upon completion: "lib/srdlib" and "lib/extlib", in that order. It is also possible to specify that only integer or floating point numbers are acceptible values.
Often it is useful to allow comma-separated lists of values as well as multiple occurrences of the options. This is easy using Perl's split() and join() operators:
my @libfiles = ();
GetOptions ("library=s" => \@libfiles);
@libfiles = split(/,/,join(',',@libfiles));Of course, it is important to choose the right separator string for each purpose.
If the option destination is a reference to a hash, the option will take, as value, strings of the form key=value. The value will be stored with the specified key in the hash.
my %defines = ();
GetOptions ("define=s" => \%defines);When used with command line options:
--define os=linux --define vendor=redhatthe hash %defines will contain two keys, "os" with value "linux and "vendor" with value "redhat". It is also possible to specify that only integer or floating point numbers are acceptible values. The keys are always taken to be strings.
Ultimate control over what should be done when (actually: each time) an option is encountered on the command line can be achieved by designating a reference to a subroutine (or an anonymous subroutine) as the option destination. When GetOptions() encounters the option, it will call the subroutine with two arguments: the name of the option, and the value to be assigned. It is up to the subroutine to store the value, or do whatever it thinks is appropriate.
A trivial application of this mechanism is to implement options that are related to each other. For example:
my $verbose = ''; # option variable with default value (false)
GetOptions ('verbose' => \$verbose,
'quiet' => sub { $verbose = 0 });Here --verbose and --quiet control the same variable $verbose, but with opposite values.
If the subroutine needs to signal an error, it should call die() with the desired error message as its argument. GetOptions() will catch the die(), issue the error message, and record that an error result must be returned upon completion.
If the text of the error message starts with an exclamantion mark ! it is interpreted specially by GetOptions(). There is currently one special command implemented: die("!FINISH") will cause GetOptions() to stop processing options, as if it encountered a double dash --.
Often it is user friendly to supply alternate mnemonic names for options. For example --height could be an alternate name for --length. Alternate names can be included in the option specification, separated by vertical bar | characters. To implement the above example:
GetOptions ('length|height=f' => \$length);The first name is called the primary name, the other names are called aliases.
Multiple alternate names are possible.
Without additional configuration, GetOptions() will ignore the case of option names, and allow the options to be abbreviated to uniqueness.
GetOptions ('length|height=f' => \$length, "head" => \$head);This call will allow --l and --L for the length option, but requires a least --hea and --hei for the head and height options.
Each option specifier consists of two parts: the name specification and the argument specification.
The name specification contains the name of the option, optionally followed by a list of alternative names separated by vertical bar characters.
length option name is "length"
length|size|l name is "length", aliases are "size" and "l"The argument specification is optional. If omitted, the option is considered boolean, a value of 1 will be assigned when the option is used on the command line.
The argument specification can be
The option does not take an argument and may be negated, i.e. prefixed by "no". E.g. "foo!" will allow --foo (a value of 1 will be assigned) and --nofoo (a value of 0 will be assigned). If the option has aliases, this applies to the aliases as well.
Using negation on a single letter option when bundling is in effect is pointless and will result in a warning.
The option does not take an argument and will be incremented by 1 every time it appears on the command line. E.g. "more+", when used with --more --more --more, will increment the value three times, resulting in a value of 3 (provided it was 0 or undefined at first).
The + specifier is ignored if the option destination is not a scalar.
The option requires an argument of the given type. Supported types are:
String. An arbitrary sequence of characters. It is valid for the argument to start with - or --.
Integer. An optional leading plus or minus sign, followed by a sequence of digits.
Real number. For example 3.14, -6.23E24 and so on.
The desttype can be @ or % to specify that the option is list or a hash valued. This is only needed when the destination for the option value is not otherwise specified. It should be omitted when not needed.
Like =, but designates the argument as optional. If omitted, an empty string will be assigned to string values options, and the value zero to numeric options.
Note that if a string argument starts with - or --, it will be considered an option on itself.
Getopt::Long encourages the use of Pod::Usage to produce help messages. For example:
use Getopt::Long;
use Pod::Usage;
my $man = 0;
my $help = 0;
GetOptions('help|?' => \$help, man => \$man) or pod2usage(2);
pod2usage(1) if $help;
pod2usage(-exitstatus => 0, -verbose => 2) if $man;
__END__
=head1 NAME
sample - Using GetOpt::Long and Pod::Usage
=head1 SYNOPSIS
sample [options] [file ...]
Options:
-help brief help message
-man full documentation
=head1 OPTIONS
=over 8
=item B<-help>
Print a brief help message and exits.
=item B<-man>
Prints the manual page and exits.
=back
=head1 DESCRIPTION
B<This program> will read the given input file(s) and do someting
useful with the contents thereof.
=cutSee Pod::Usage for details.
Sometimes, for example when there are a lot of options, having a separate variable for each of them can be cumbersome. GetOptions() supports, as an alternative mechanism, storing options in a hash.
To obtain this, a reference to a hash must be passed as the first argument to GetOptions(). For each option that is specified on the command line, the option value will be stored in the hash with the option name as key. Options that are not actually used on the command line will not be put in the hash, on other words, exists($h{option}) (or defined()) can be used to test if an option was used. The drawback is that warnings will be issued if the program runs under use strict and uses $h{option} without testing with exists() or defined() first.
my %h = ();
GetOptions (\%h, 'length=i'); # will store in $h{length}For options that take list or hash values, it is necessary to indicate this by appending an @ or % sign after the type:
GetOptions (\%h, 'colours=s@'); # will push to @{$h{colours}}To make things more complicated, the hash may contain references to the actual destinations, for example:
my $len = 0;
my %h = ('length' => \$len);
GetOptions (\%h, 'length=i'); # will store in $lenThis example is fully equivalent with:
my $len = 0;
GetOptions ('length=i' => \$len); # will store in $lenAny mixture is possible. For example, the most frequently used options could be stored in variables while all other options get stored in the hash:
my $verbose = 0; # frequently referred
my $debug = 0; # frequently referred
my %h = ('verbose' => \$verbose, 'debug' => \$debug);
GetOptions (\%h, 'verbose', 'debug', 'filter', 'size=i');
if ( $verbose ) { ... }
if ( exists $h{filter} ) { ... option 'filter' was specified ... }With bundling it is possible to set several single-character options at once. For example if a, v and x are all valid options,
-vaxwould set all three.
Getopt::Long supports two levels of bundling. To enable bundling, a call to Getopt::Long::Configure is required.
The first level of bundling can be enabled with:
Getopt::Long::Configure ("bundling");Configured this way, single-character options can be bundled but long options must always start with a double dash -- to avoid abiguity. For example, when vax, a, v and x are all valid options,
-vaxwould set a, v and x, but
--vaxwould set vax.
The second level of bundling lifts this restriction. It can be enabled with:
Getopt::Long::Configure ("bundling_override");Now, -vax would set the option vax.
When any level of bundling is enabled, option values may be inserted in the bundle. For example:
-h24w80is equivalent to
-h 24 -w 80When configured for bundling, single-character options are matched case sensitive while long options are matched case insensitive. To have the single-character options matched case insensitive as well, use:
Getopt::Long::Configure ("bundling", "ignorecase_always");It goes without saying that bundling can be quite confusing.
Some applications require the option - (that's a lone dash). This can be achieved by adding an option specification with an empty name:
GetOptions ('' => \$stdio);A lone dash on the command line will now be legal, and set options variable $stdio.
A special option 'name' <> can be used to designate a subroutine to handle non-option arguments. When GetOptions() encounters an argument that does not look like an option, it will immediately call this subroutine and passes it the argument as a parameter.
For example:
my $width = 80;
sub process { ... }
GetOptions ('width=i' => \$width, '<>' => \&process);When applied to the following command line:
arg1 --width=72 arg2 --width=60 arg3This will call process("arg1") while $width is 80, process("arg2") while $width is 72, and process("arg3") while $width is 60.
This feature requires configuration option permute, see section "Configuring Getopt::Long".
Getopt::Long can be configured by calling subroutine Getopt::Long::Configure(). This subroutine takes a list of quoted strings, each specifying a configuration option to be set, e.g. ignore_case, or reset, e.g. no_ignore_case. Case does not matter. Multiple calls to Configure() are possible.
The following options are available:
This option causes all configuration options to be reset to their default values.
Allow option names to be abbreviated to uniqueness. Default is set unless environment variable POSIXLY_CORRECT has been set, in which case auto_abbrev is reset.
Allow + to start options. Default is set unless environment variable POSIXLY_CORRECT has been set, in which case getopt_compat is reset.
Whether command line arguments are allowed to be mixed with options. Default is set unless environment variable POSIXLY_CORRECT has been set, in which case require_order is reset.
See also permute, which is the opposite of require_order.
Whether command line arguments are allowed to be mixed with options. Default is set unless environment variable POSIXLY_CORRECT has been set, in which case permute is reset. Note that permute is the opposite of require_order.
If permute is set, this means that
--foo arg1 --bar arg2 arg3is equivalent to
--foo --bar arg1 arg2 arg3If an argument call-back routine is specified, @ARGV will always be empty upon succesful return of GetOptions() since all options have been processed. The only exception is when -- is used:
--foo arg1 --bar arg2 -- arg3will call the call-back routine for arg1 and arg2, and terminate GetOptions() leaving "arg2" in @ARGV.
If require_order is set, options processing terminates when the first non-option is encountered.
--foo arg1 --bar arg2 arg3is equivalent to
--foo -- arg1 --bar arg2 arg3Setting this option will allow single-character options to be bundled. To distinguish bundles from long option names, long options must be introduced with -- and single-character options (and bundles) with -.
Note: resetting bundling also resets bundling_override.
If bundling_override is set, bundling is enabled as with bundling but now long option names override option bundles.
Note: resetting bundling_override also resets bundling.
Note: Using option bundling can easily lead to unexpected results, especially when mixing long options and bundles. Caveat emptor.
If set, case is ignored when matching long option names. Single character options will be treated case-sensitive.
Note: resetting ignore_case also resets ignore_case_always.
When bundling is in effect, case is ignored on single-character options also.
Note: resetting ignore_case_always also resets ignore_case.
Options that are unknown, ambiguous or supplied with an invalid option value are passed through in @ARGV instead of being flagged as errors. This makes it possible to write wrapper scripts that process only part of the user supplied command line arguments, and pass the remaining options to some other program.
This can be very confusing, especially when permute is also set.
The string that starts options. If a constant string is not sufficient, see prefix_pattern.
A Perl pattern that identifies the strings that introduce options. Default is (--|-|\+) unless environment variable POSIXLY_CORRECT has been set, in which case it is (--|-).
Enable copious debugging output.
Configuration errors and errors in the option definitions are signalled using die() and will terminate the calling program unless the call to Getopt::Long::GetOptions() was embedded in eval { ... }, or die() was trapped using $SIG{__DIE__}.
A return value of 1 (true) indicates success.
A return status of 0 (false) indicates that the function detected one or more errors during option parsing. These errors are signalled using warn() and can be trapped with $SIG{__WARN__}.
Errors that can't happen are signalled using Carp::croak().
The earliest development of newgetopt.pl started in 1990, with Perl version 4. As a result, its development, and the development of Getopt::Long, has gone through several stages. Since backward compatibility has always been extremely important, the current version of Getopt::Long still supports a lot of constructs that nowadays are no longer necessary or otherwise unwanted. This section describes briefly some of these 'features'.
When no destination is specified for an option, GetOptions will store the resultant value in a global variable named opt_XXX, where XXX is the primary name of this option. When a progam executes under use strict (recommended), these variables must be pre-declared with our() or use vars.
our $opt_length = 0;
GetOptions ('length=i'); # will store in $opt_lengthTo yield a usable Perl variable, characters that are not part of the syntax for variables are translated to underscores. For example, --fpp-struct-return will set the variable $opt_fpp_struct_return. Note that this variable resides in the namespace of the calling program, not necessarily main. For example:
GetOptions ("size=i", "sizes=i@");with command line "-size 10 -sizes 24 -sizes 48" will perform the equivalent of the assignments
$opt_size = 10;
@opt_sizes = (24, 48);A string of alternative option starter characters may be passed as the first argument (or the first argument after a leading hash reference argument).
my $len = 0;
GetOptions ('/', 'length=i' => $len);Now the command line may look like:
/length 24 -- argNote that to terminate options processing still requires a double dash --.
GetOptions() will not interpret a leading "<"> as option starters if the next argument is a reference. To force "<" and ""> as option starters, use "<">. Confusing? Well, using a starter argument is strongly deprecated anyway.
Previous versions of Getopt::Long used variables for the purpose of configuring. Although manipulating these variables still work, it is strongly encouraged to use the new config routine. Besides, it is much easier.
Johan Vromans <[email protected]>
This program is Copyright 2000,1990 by Johan Vromans. This program is free software; you can redistribute it and/or modify it under the terms of the Perl Artistic License or the GNU General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version.
This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
If you do not have a copy of the GNU General Public License write to the Free Software Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.