NAME

    Getopt::Euclid - Executable Uniform Command-Line Interface Descriptions

VERSION

    This document describes Getopt::Euclid version 0.4.5

SYNOPSIS

        use Getopt::Euclid;
    
        if ($ARGV{-i}) {
            print "Interactive mode...\n";
        }
    
        for my $x (0..$ARGV{-size}{h}-1) {
            for my $y (0..$ARGV{-size}{w}-1) {
                do_something_with($x, $y);
            }
        }
    
        __END__
    
        =head1 NAME
    
        yourprog - Your program here
    
        =head1 VERSION
    
        This documentation refers to yourprog version 1.9.4
    
        =head1 USAGE
    
            yourprog [options]  -s[ize]=<h>x<w>  -o[ut][file] <file>
    
        =head1 REQUIRED ARGUMENTS
    
        =over
    
        =item  -s[ize]=<h>x<w>    
    
        Specify size of simulation
    
        =for Euclid:
            h.type:    int > 0
            h.default: 24
            w.type:    int >= 10
            w.default: 80
    
        =item  -o[ut][file] <file>    
    
        Specify output file
    
        =for Euclid:
            file.type:    writable
            file.default: '-'
    
        =back
    
        =head1 OPTIONS
    
        =over
    
        =item  -i
    
        Specify interactive simulation
    
        =item  -l[[en][gth]] <l>
    
        Length of simulation. The default is l.default
    
        =for Euclid:
            l.type:    int > 0
            l.default: 99
    
        =item --debug [<log_level>]
    
        Set the log level. Default is log_level.default but if you provide --debug,
        then it is log_level.opt_default.
    
        =for Euclid:
            log_level.type:        int
            log_level.default:     0
            log_level.opt_default: 1
    
        =item --version
    
        =item --usage
    
        =item --help
    
        =item --man
    
        Print the usual program information
    
        =back
    
        Remainder of documentation starts here...
    
        =head1 AUTHOR
    
        Damian Conway (DCONWAY@CPAN.org)
    
        =head1 BUGS
    
        There are undoubtedly serious bugs lurking somewhere in this code.
        Bug reports and other feedback are most welcome.
    
        =head1 COPYRIGHT
    
        Copyright (c) 2005, Damian Conway. All Rights Reserved.
        This module is free software. It may be used, redistributed
        and/or modified under the terms of the Perl Artistic License
        (see http://www.perl.com/perl/misc/Artistic.html)

DESCRIPTION

    Getopt::Euclid uses your program's own POD documentation to create a
    powerful command-line argument parser. This ensures that your program's
    documented interface and its actual interface always agree.

    The created command-line argument parser includes many features such as
    argument type checking, required arguments, exclusive arguments,
    optional arguments with default values, automatic usage message, ...

    To use the module, simply write the following at the top of your
    program:

        use Getopt::Euclid;

    This will cause Getopt::Euclid to be require'd and its import method
    will be called. It is important that the import method be allowed to
    run, so do not invoke Getopt::Euclid in the following manner:

        # Will not work
        use Getopt::Euclid ();

    When the module is loaded within a regular Perl program, it will:

      1. locate any POD in the same *.pl file or its associated *.pod file.

      2. extract information from that POD, most especially from the =head1
      REQUIRED ARGUMENTS and =head1 OPTIONS sections,

      3. build a parser that parses the arguments and options the POD
      specifies,

      4. remove the command-line arguments from @ARGV and parse them, and

      5. put the results in the global %ARGV variable (or into specifically
      named optional variables, if you request that -- see "Exporting
      option variables").

    As a special case, if the module is loaded within some other module
    (i.e. from within a .pm file), it still locates and extracts POD
    information, but instead of parsing @ARGV immediately, it caches that
    information and installs an import() subroutine in the caller module.
    This new import() acts just like Getopt::Euclid's own import, except
    that it adds the POD from the caller module to the POD of the callee.

    All of which just means you can put some or all of your CLI
    specification in a module, rather than in the application's source
    file. See "Module interface" for more details.

INTERFACE

 Program interface

    You write:

        use Getopt::Euclid;

    and your command-line is parsed automagically.

 Module interface

    import()

      You write:

          use Getopt::Euclid;

      and your module will then act just like Getopt::Euclid (i.e. you can
      use your module instead of Getopt::Euclid>, except that your module's
      POD will also be prepended to the POD of any module that loads yours.
      In other words, you can use Getopt::Euclid in a module to create a
      standard set of CLI arguments, which can then be added to any
      application simply by loading your module.

      To accomplish this trick Getopt::Euclid installs an import()
      subroutine in your module. If your module already has an import()
      subroutine defined, terrible things happen. So do not do that.

      You may also short-circuit the import method within your calling
      program to have the POD from several modules included for argument
      parsing.

          use Module1::Getopt (); # No argument parsing
          use Module2::Getopt (); # No argument parsing
          use Getopt::Euclid;     # Arguments parsed

    process_args()

      Alternatively, to parse arguments from a source different from @ARGV,
      use the process_args() subroutine.

          use Getopt::Euclid qw(:defer);
          my @args = ( '-in', 'file.txt', '-out', 'results.txt' );
          Getopt::Euclid->process_args(\@args);

      If you want to use the :minimal or :vars mode in this type of
      scenario, you can pass extra options to process_args():

          use Getopt::Euclid qw(:defer);
          my @args = ( '-in', 'file.txt', '-out', 'results.txt' );
          Getopt::Euclid->process_args(\@args, {-minimal => 1, -vars => 'prefix_'});

      This is particularly when you plan on processing POD manually.

    process_pods()

      Similarly, to parse argument specifications from a source different
      than the current script (and its dependencies), use the
      process_pods() subroutine.

          use Getopt::Euclid ();
          my @pods = ( 'script.pl', 'Module.pm' );
          $Getopt::Euclid::MAN = Getopt::Euclid->process_pods(\@pods, {-strict => 1});
          my @args = ( '-in', 'file.txt', '-out', 'results.txt' );
          Getopt::Euclid->process_args(\@args);

      By default, this method will look for .pod files associated with the
      given .pl and .pm files and use these .pod files preferentially when
      available. Set -strict to 1 to only use the given files.

 POD interface

    This is where all the action is. POD markup can be placed in a .pod
    file that has the same prefix as the corresponding Perl file.
    Alternatively, POD can be inserted anywhere in the Perl code, but is
    typically added either after an __END__ statement (like in the
    SYNOPSIS), or interspersed in the code:

        use Getopt::Euclid;
    
        =head1 NAME
    
        yourprog - Your program here
    
        =head1 REQUIRED ARGUMENTS
    
        =over
    
        =item  -s[ize]=<h>x<w>    
    
        Specify size of simulation
    
        =for Euclid:
            h.type:    int > 0
            h.default: 24
            w.type:    int >= 10
            w.default: 80
    
        =back
    
        =head1 OPTIONS
    
        =over
    
        =item  -i
    
        Specify interactive simulation
    
        =back
    
        =cut
    
        # Getopt::Euclid has parsed commandline parameters and stored them in %ARGV
    
        if ($ARGV{-i}) {
            print "Interactive mode...\n";
        }
    
        for my $x (0..$ARGV{-size}{h}-1) {
            for my $y (0..$ARGV{-size}{w}-1) {
                do_something_with($x, $y);
            }
        }

    When Getopt::Euclid is loaded in a non-.pm file, it searches that file
    for the following POD documentation:

    =head1 NAME

      Getopt::Euclid ignores the name specified here. In fact, if you use
      the standard --help, --usage, --man, --podfile, or --version
      arguments (see "Standard arguments"), the module replaces the name
      specified in this POD section with the actual name by which the
      program was invoked (i.e. with $0).

    =head1 USAGE

      Getopt::Euclid ignores the usage line specified here. If you use the
      standard --help, --usage, --man or --podfile arguments, the module
      replaces the usage line specified in this POD section with a usage
      line that reflects the actual interface that the module has
      constructed.

    =head1 VERSION

      Getopt::Euclid extracts the current version number from this POD
      section. To do that it simply takes the first substring that matches
      <digit>.<digit> or <digit>_<digit>. It also accepts one or more
      additional trailing .<digit> or _<digit>, allowing for multi-level
      and "alpha" version numbers such as:

          =head1 VERSION
          
          This is version 1.2.3

      or:

          =head1 VERSION
          
          This is alpha release 1.2_34

      You may also specify the version number in your code. However, in
      order for Getopt::Euclid to properly read it, it must be in a BEGIN
      block:

          BEGIN { use version; our $VERSION = qv('1.2.3') }
          use Getopt::Euclid;

      Euclid stores the version as $Getopt::Euclid::SCRIPT_VERSION.

    =head1 REQUIRED ARGUMENTS

      Getopt::Euclid uses the specifications in this POD section to build a
      parser for command-line arguments. That parser requires that every
      one of the specified arguments is present in any command-line
      invocation. See "Specifying arguments" for details of the
      specification syntax.

      The actual headings that Getopt::Euclid can recognize here are:

          =head1 [STANDARD|STD|PROGRAM|SCRIPT|CLI|COMMAND[-| ]LINE] [REQUIRED|MANDATORY] [PARAM|PARAMETER|ARG|ARGUMENT][S]

      Caveat: Do not put additional subheadings (=headX) inside the
      REQUIRED ARGUMENTS section.

    =head1 OPTIONS

      Getopt::Euclid uses the specifications in this POD section to build a
      parser for command-line arguments. That parser does not require that
      any of the specified arguments is actually present in a command-line
      invocation. Again, see "Specifying arguments" for details of the
      specification syntax.

      Typically a program will specify both REQUIRED ARGUMENTS and OPTIONS,
      but there is no requirement that it supply both, or either.

      The actual headings that Getopt::Euclid recognizes here are:

          =head1 [STANDARD|STD|PROGRAM|SCRIPT|CLI|COMMAND[-| ]LINE] OPTION[AL|S] [PARAM|PARAMETER|ARG|ARGUMENT][S]

      Caveat: Do not put additional subheadings (=headX) inside the
      REQUIRED ARGUMENTS section.

    =head1 COPYRIGHT

      Getopt::Euclid prints this section whenever the standard --version
      option is specified on the command-line.

      The actual heading that Getopt::Euclid recognizes here is any heading
      containing any of the words "COPYRIGHT", "LICENCE", or "LICENSE".

 Specifying arguments

    Each required or optional argument is specified in the POD in the
    following format:

        =item ARGUMENT_STRUCTURE
    
        ARGUMENT_DESCRIPTION
    
        =for Euclid:
            ARGUMENT_OPTIONS
            PLACEHOLDER_CONSTRAINTS

  Argument structure

      * Each argument is specified as an =item.

      * Any part(s) of the specification that appear in square brackets are
      treated as optional.

      * Any parts that appear in angle brackets are placeholders for actual
      values that must be specified on the command-line.

      * Any placeholder that is immediately followed by ... may be repeated
      as many times as desired.

      * Any whitespace in the structure specifies that any amount of
      whitespace (including none) is allowed at the same position on the
      command-line.

      * A vertical bar indicates the start of an alternative variant of the
      argument.

    For example, the argument specification:

        =item -i[n] [=] <file> | --from <file>

    indicates that any of the following may appear on the command-line:

        -idata.txt    -i data.txt    -i=data.txt    -i = data.txt
                                         
        -indata.txt   -in data.txt   -in=data.txt   -in = data.txt
    
        --from data.text

    as well as any other combination of whitespacing.

    Any of the above variations would cause all three of:

        $ARGV{'-i'}
        $ARGV{'-in'}
        $ARGV{'--from'}

    to be set to the string 'data.txt'.

    You could allow the optional = to also be an optional colon by
    specifying:

        =item -i[n] [=|:] <file>

    Optional components may also be nested, so you could write:

        =item -i[n[put]] [=] <file>

    which would allow -i, -in, and -input as synonyms for this argument and
    would set all three of $ARGV{'-i'}, $ARGV{'-in'}, and $ARGV{'-input'}
    to the supplied file name.

    The point of setting every possible variant within %ARGV is that this
    allows you to use a single key (say $ARGV{'-input'}, regardless of how
    the argument is actually specified on the command-line.

 Repeatable arguments

    Normally Getopt::Euclid only accepts each specified argument once, the
    first time it appears in @ARGV. However, you can specify that an
    argument may appear more than once, using the repeatable option:

        =item file=<filename>
    
        =for Euclid:
            repeatable

    When an argument is marked repeatable the corresponding entry of %ARGV
    will not contain a single value, but rather an array reference. If the
    argument also has "Multiple placeholders", then the corresponding entry
    in %ARGV will be an array reference with each array entry being a hash
    reference.

 Boolean arguments

    If an argument has no placeholders it is treated as a boolean switch
    and its entry in %ARGV will be true if the argument appeared in @ARGV.

    For a boolean argument, you can also specify variations that are false,
    if they appear. For example, a common idiom is:

        =item --print
    
        Print results
    
        =item --noprint
    
        Do not print results

    These two arguments are effectively the same argument, just with
    opposite boolean values. However, as specified above, only one of
    $ARGV{'--print'} and $ARGV{'--noprint'} will be set.

    As an alternative you can specify a single argument that accepts either
    value and sets both appropriately:

        =item --[no]print
    
        [Do not] print results
    
        =for Euclid:
            false: --noprint

    With this specification, if --print appears in @ARGV, then
    $ARGV{'--print'} will be true and $ARGV{'--noprint'} will be false. On
    the other hand, if --noprint appears in @ARGV, then $ARGV{'--print'}
    will be false and $ARGV{'--noprint'} will be true.

    The specified false values can follow any convention you wish:

        =item [+|-]print
    
        =for Euclid:
            false: -print

    or:

        =item -report[_no[t]]
    
        =for Euclid:
            false: -report_no[t]

    et cetera.

 Multiple placeholders

    An argument can have two or more placeholders:

        =item -size <h> <w>

    The corresponding command line argument would then have to provide two
    values:

        -size 24 80

    Multiple placeholders can optionally be separated by literal characters
    (which must then appear on the command-line). For example:

        =item -size <h>x<w>

    would then require a command-line of the form:

        -size 24x80

    If an argument has two or more placeholders, the corresponding entry in
    %ARGV becomes a hash reference, with each of the placeholder names as
    one key. That is, the above command-line would set both
    $ARGV{'-size'}{'h'} and $ARGV{'-size'}{'w'}.

 Optional placeholders

    Placeholders can be specified as optional as well:

        =item -size <h> [<w>]

    This specification then allows either:

        -size 24

    or:

        -size 24 80

    on the command-line. If the second placeholder value is not provided,
    the corresponding $ARGV{'-size'}{'w'} entry is set to undef. See also
    "Placeholder defaults".

 Unflagged placeholders

    If an argument consists of a single placeholder with no "flag" marking
    it:

        =item <filename>

    then the corresponding entry in %ARG will have a key the same as the
    placeholder (including the surrounding angle brackets):

        if ($ARGV{'<filename>'} eq '-') {
            $fh = \*STDIN;
        }

    The same is true for any more-complicated arguments that begin with a
    placeholder:

        =item <h> [x <w>]

    The only difference in the more-complex cases is that, if the argument
    has any additional placeholders, the entire entry in %ARGV becomes a
    hash:

        my $total_size
            = $ARGV{'<h>'}{'h'} * $ARGV{'<h>'}{'w'}

    Note that, as in earlier multi-placeholder examples, the individual
    second- level placeholder keys do not retain their angle-brackets.

 Repeated placeholders

    Any placeholder that is immediately followed by ..., like so:

        =item -lib <file>...
    
        =for Euclid:
            file.type: readable

    will match at least once, but as many times as possible before
    encountering the next argument on the command-line. This allows one to
    specify multiple values for an argument, for example:

        -lib file1.txt file2.txt

    An unconstrained repeated unflagged placeholder (see "Placeholder
    constraints" and "Unflagged placeholders") will consume the rest of the
    command-line, and so should be specified last in the POD

        =item -n <name>
    
        =item <offset>...
    
        =for Euclid:
            offset.type: 0+int

    and on the command-line:

        -n foobar 1 5 0 23

    If a placeholder is repeated, the corresponding entry in %ARGV will
    then be an array reference, with each individual placeholder match in a
    separate element. For example:

        for my $lib (@{ $ARGV{'-lib'} }) {
            add_lib($lib);
        }
    
        warn "First offset is: $ARGV{'<offsets>'}[0]";
        my $first_offset = shift @{ $ARGV{'<offsets>'} };

 Placeholder constraints

    You can specify that the value provided for a particular placeholder
    must satisfy a particular set of restrictions by using a =for Euclid
    block. For example:

        =item -size <h>x<w>
    
        =for Euclid:
            h.type: integer
            w.type: integer

    specifies that both the <h> and <w> must be given integers. You can
    also specify an operator expression after the type name:

        =for Euclid:
            h.type: integer > 0
            w.type: number <= 100

    specifies that <h> has to be given an integer that is greater than
    zero, and that <w> has to be given a number (not necessarily an
    integer) that is no more than 100.

    These type constraints have two alternative syntaxes:

        PLACEHOLDER.type: TYPE BINARY_OPERATOR EXPRESSION

    as shown above, and the more general:

        PLACEHOLDER.type: TYPE [, EXPRESSION_INVOLVING(PLACEHOLDER)]

    Using the second syntax, you could write the previous constraints as:

        =for Euclid:
            h.type: integer, h > 0
            w.type: number,  w <= 100

    In other words, the first syntax is just sugar for the most common case
    of the second syntax. The expression can be as complex as you wish and
    can refer to the placeholder as many times as necessary:

        =for Euclid:
            h.type: integer, h > 0 && h < 100
            w.type: number,  Math::is_prime(w) || w % 2 == 0

    Note that the expressions are evaluated in the package main namespace,
    so it is important to qualify any subroutines that are not in that
    namespace. Furthermore, any subroutines used must be defined (or loaded
    from a module) before the use Getopt::Euclid statement.

    You can also use constraints that involve variables. You must use the
    :defer mode and the variables must be globally accessible:

        use Getopt::Euclid qw(:defer);
        our $MIN_VAL = 100;
        Getopt::Euclid->process_args(\@ARGV);
    
        __END__
    
        =head1 OPTIONS
    
        =over
    
        =item --magnitude <magnitude>
    
        =for Euclid
           magnitude.type: number, magnitude > $MIN_VAL
    
        =back

 Standard placeholder types

    Getopt::Euclid recognizes the following standard placeholder types:

        Name            Placeholder value...        Synonyms
        ============    ====================        ================
    
        integer         ...must be an integer       int    i
    
        +integer        ...must be a positive       +int   +i
                        integer
                        (same as: integer > 0)
    
        0+integer       ...must be a positive       0+int  0+i
                        integer or zero
                        (same as: integer >= 0)
    
        number          ...must be an number        num    n
    
        +number         ...must be a positive       +num   +n
                        number
                        (same as: number > 0)
    
        0+number        ...must be a positive       0+num  0+n
                        number or zero
                        (same as: number >= 0)
    
        string          ...may be any string        str    s
                        (default type)
    
        readable        ...must be the name         input  in
                        of a readable file
    
        writeable       ...must be the name         writable output out
                        of a writeable file
                        (or of a non-existent
                        file in a writeable
                        directory)
                        
        /<regex>/       ...must be a string
                        matching the specified
                        pattern

    Since regular expressions are supported, you can easily match many more
    type of strings for placeholders by using the regular expressions
    available in Regexp::Common. If you do that, you may want to also use
    custom placeholder error messages (see "Placeholder type errors") since
    the messages would otherwise not be very informative to users.

        use Regexp::Common qw /zip/;
        use Getopt::Euclid;
    
        ...
    
        =item -p <postcode>
    
        Enter your postcode here
    
        =for Euclid:
            postcode.type:  /$RE{zip}{France}/
            postcode.type.error: <postcode> must be a valid ZIP code

 Placeholder type errors

    If a command-line argument's placeholder value does not satisify the
    specified type, an error message is automatically generated. However,
    you can provide your own message instead, using the .type.error
    specifier:

        =for Euclid:
            h.type:        integer, h > 0 && h < 100
            h.type.error:  <h> must be between 0 and 100 (not h)
    
            w.type:        number,  Math::is_prime(w) || w % 2 == 0
            w.type.error:  Cannot use w for <w> (must be an even prime number)

    Whenever an explicit error message is provided, any occurrence within
    the message of the placeholder's unbracketed name is replaced by the
    placeholder's value (just as in the type test itself).

 Placeholder defaults

    You can also specify a default value for any placeholders that are not
    given values on the command-line (either because their argument is not
    provided at all, or because the placeholder is optional within the
    argument). For example:

        =item -size <h>[x<w>]
    
        Set the size of the simulation
    
        =for Euclid:
            h.default: 24
            w.default: 80

    This ensures that if no <w> value is supplied:

        -size 20

    then $ARGV{'-size'}{'w'} is set to 80. Likewise, of the -size argument
    is omitted entirely, both $ARGV{'-size'}{'h'} and $ARGV{'-size'}{'w'}
    are set to their respective default values

    However, Getopt::Euclid also supports a second type of default,
    optional defaults, that apply only to flagged, optional placeholders.

    For example:

        =item --debug [<log_level>]
    
        Set the log level
    
        =for Euclid:
            log_level.type:        int
            log_level.default:     0
            log_level.opt_default: 1

    This ensures that if the option --debug is not specified, then
    $ARGV{'--debug'} is set to 0, the regular default. But if no
    <log_level> value is supplied:

        --debug
        

    then $ARGV{'--debug'} is set to 1, the optional default.

    The default value can be any valid Perl compile-time expression:

        =item -pi=<pi value>
    
        =for Euclid:
            pi value.default: atan2(0,-1)

    You can refer to an argument default or optional default value in its
    POD entry as shown below:

        =item -size <h>[x<w>]
    
        Set the size of the simulation [default: h.default x w.default]
    
        =for Euclid:
            h.default: 24
            w.default: 80
    
        =item --debug <level>
       
        Set the debug level. The default is level.default if you supply --debug but
        omit a <level> value.
    
        =for Euclid:
            level.opt_default: 3

    Just like for "Placeholder constraints", you can also use variables to
    define default values. You must use the :defer mode and the variables
    must be globally accessible:

        use Getopt::Euclid qw(:defer);
        Getopt::Euclid->process_args(\@ARGV);
    
        __END__
    
        =head1 OPTIONS
    
        =over
    
        =item --home <home>
    
        Your project home. When omitted, this defaults to the location stored in
        the HOME environment variable.
    
        =for Euclid
           home.default: $ENV{'HOME'}
    
        =back

 Exclusive placeholders

    Some arguments can be mutually exclusive. In this case, it is possible
    to specify that a placeholder excludes a list of other placeholders,
    for example:

        =item -height <h>
    
        Set the desired height
    
        =item -width <w>
    
        Set the desired width
    
        =item -volume <v>
    
        Set the desired volume
    
        =for Euclid:
            v.excludes: h, w
            v.excludes.error: Either set the volume or the height and weight

    Specifying both placeholders at the same time on the command-line will
    generate an error. Note that the error message can be customized, as
    illustrated above.

    When using exclusive arguments that have default values, the default
    value of the placeholder with the .excludes statement has precedence
    over any other placeholders.

 Argument cuddling

    Getopt::Euclid allows any "flag" argument to be "cuddled". A flag
    argument consists of a single non- alphanumeric character, followed by
    a single alpha-numeric character:

        =item -v
    
        =item -x
    
        =item +1
    
        =item =z

    Cuddling means that two or more such arguments can be concatenated
    after a single common non-alphanumeric. For example:

        -vx

    Note, however, that only flags with the same leading non-alphanumeric
    can be cuddled together. Getopt::Euclid would not allow:

        -vxz

    This is because cuddling is recognized by progressively removing the
    second character of the cuddle. In other words:

        -vxz

    becomes:

        -v -xz

    which becomes:

        -v -x z

    which will fail, unless a z argument has also been specified.

    On the other hand, if the argument:

        =item -e <cmd>

    had been specified, the module would accept:

        -vxe'print time'

    as a cuddled version of:

        -v -x -e'print time'

 Exporting option variables

    By default, the module only stores arguments into the global %ARGV
    hash. You can request that options are exported as variables into the
    calling package using the special ':vars' specifier:

        use Getopt::Euclid qw( :vars );

    That is, if your program accepts the following arguments:

        -v
        --mode <modename>
        <infile>
        <outfile>
        --auto-fudge <factor>      (repeatable)
        --also <a>...
        --size <w>x<h>
        --multiply <num1>x<num2>   (repeatable)

    Then these variables will be exported

        $ARGV_v
        $ARGV_mode
        $ARGV_infile
        $ARGV_outfile
        @ARGV_auto_fudge
        @ARGV_also
        %ARGV_size          # With entries $ARGV_size{w} and $ARGV_size{h}
        @ARGV_multiply      # With entries that are hashref similar to \%ARGV_size

    For options that have multiple variants, only the longest variant is
    exported.

    The type of variable exported (scalar, hash, or array) is determined by
    the type of the corresponding value in %ARGV. Command-line flags and
    arguments that take single values will produce scalars, arguments that
    take multiple values will produce hashes, and repeatable arguments will
    produce arrays.

    If you do not like the default prefix of "ARGV_", you can specify your
    own, such as "opt_", like this:

        use Getopt::Euclid qw( :vars<opt_> );

    The major advantage of using exported variables is that any misspelling
    of argument variables in your code will be caught at compile-time by
    use strict.

 Standard arguments

    Getopt::Euclid automatically provides four standard arguments to any
    program that uses the module. The behaviours of these arguments are
    "hard- wired" and cannot be changed, not even by defining your own
    arguments of the same name.

    The standard arguments are:

    --usage usage()

      The --usage argument causes the program to print a short usage
      summary and exit. The Getopt::Euclid-usage()> subroutine provides
      access to the string of this message.

    --help help()

      The --help argument causes the program to take a longer usage summary
      (with a full list of required and optional arguments) provided in POD
      format by help(), convert it to plaintext, display it and exit. The
      message is paged using IO::Pager::Page (or IO::Page) if possible.

    --man man()

      The --man argument causes the program to take the POD documentation
      for the program, provided by man(), convert it to plaintext, display
      it and exit. The message is paged using IO::Pager::Page (or IO::Page)
      if possible.

    --podfile podfile()

      The --podfile argument is provided for authors. It causes the program
      to take the POD manual from man(), write it in a .pod file with the
      same base name as the program, display the name of the output file
      and exit. These actions can also be executed by calling the podfile()
      subroutine.This argument is not really a standard argument, but it is
      useful if the program's POD is to be passed to a POD converter
      because, among other things, any default value specified is
      interpolated and replaced by its value in the .pod file, contrary to
      in the program's .pl file.

      If you want to automate the creation of a POD file during the build
      process, you can edit you Makefile.PL or Build.PL file and add these
      lines:

         my @args = ($^X, '-Ilib', '/path/to/script', '--podfile');
         system(@args) == 0 or die "System call to '@args' failed:\n$?\n";

      If you use Module::Install to bundle your script, you might be
      interested in using Module::Install::PodFromEuclid to include the
      --podfile step into the installation process.

    --version version()

      The --version argument causes the program to print the version number
      of the program (as specified in the =head1 VERSION section of the
      POD) and any copyright information (as specified in the =head1
      COPYRIGHT POD section) and then exit. The Getopt::Euclid-version()>
      subroutine provides access to the string of this message.

 Minimalist keys

    By default, the keys of %ARGV will match the program's interface
    exactly. That is, if your program accepts the following arguments:

        -v
        --mode <modename>
        <infile>
        <outfile>
        --auto-fudge

    Then the keys that appear in %ARGV will be:

        '-v'
        '--mode'
        '<infile>'
        '<outfile>'
        '--auto-fudge'

    In some cases, however, it may be preferable to have Getopt::Euclid set
    up those hash keys without "decorations". That is, to have the keys of
    %ARGV be simply:

        'v'
        'mode'
        'infile'
        'outfile'
        'auto_fudge'

    You can arrange this by loading the module with the special
    ':minimal_keys' specifier:

        use Getopt::Euclid qw( :minimal_keys );

    Note that, in rare cases, using this mode may cause you to lose data
    (for example, if the interface specifies both a --step and a <step>
    option). The module throws an exception if this happens.

 Deferring argument parsing

    In some instances, you may want to avoid the parsing of arguments to
    take place as soon as your program is executed and Getopt::Euclid is
    loaded. For example, you may need to examine @ARGV before it is
    processed (and emptied) by Getopt::Euclid. Or you may intend to pass
    your own arguments manually only using process_args().

    To defer the parsing of arguments, use the specifier ':defer':

        use Getopt::Euclid qw( :defer );
        # Do something...
        Getopt::Euclid->process_args(\@ARGV);

DIAGNOSTICS

 Compile-time diagnostics

    The following diagnostics are mainly caused by problems in the POD
    specification of the command-line interface:

    Getopt::Euclid was unable to access POD

      Something is horribly wrong. Getopt::Euclid was unable to read your
      program to extract the POD from it. Check your program's permissions,
      though it is a mystery how perl was able to run the program in the
      first place, if it is not readable.

    .pm file cannot define an explicit import() when using Getopt::Euclid

      You tried to define an import() subroutine in a module that was also
      using Getopt::Euclid. Since the whole point of using Getopt::Euclid
      in a module is to have it build an import() for you, supplying your
      own import() as well defeats the purpose.

    Unknown specification: %s

      You specified something in a =for Euclid section that Getopt::Euclid
      did not understand. This is often caused by typos, or by reversing a
      placeholder.type or placeholder.default specification (that is,
      writing type.placeholder or default.placeholder instead).

    Unknown type (%s) in specification: %s

    Unknown .type constraint: %s

      Both these errors mean that you specified a type constraint that
      Getopt::Euclid did not recognize. This may have been a typo:

          =for Euclid
              count.type: inetger

      or else the module simply does not know about the type you specified:

          =for Euclid
              count.type: complex

      See "Standard placeholder types" for a list of types that
      Getopt::Euclid does recognize.

    Invalid .type constraint: %s

      You specified a type constraint that is not valid Perl. For example:

          =for Euclid
              max.type: integer not equals 0

      instead of:

          =for Euclid
              max.type: integer != 0

    Invalid .default value: %s

      You specified a default value that is not valid Perl. For example:

          =for Euclid
              curse.default: *$@!&

      instead of:

          =for Euclid
              curse.default: '*$@!&'

    Invalid .opt_default value: %s

      Same as previous diagnostic, but for optional defaults.

    Invalid reference to field %s.default in argument description: %s

      You referred to a default value in the description of an argument,
      but there is no such default. It may be a typo, or you may be
      referring to the default value for a different argument, e.g.:

          =item -a <age>
      
          An optional age. Default: years.default
      
          =for Euclid
              age.default: 21

      instead of:

          =item -a <age>
      
          An optional age. Default: age.default
      
          =for Euclid
              age.default: 21

    Invalid reference to field %s.opt_default in argument description: %s

      Same as previous diagnostic, but for optional defaults.

    Invalid .opt_default constraint: Placeholder <%s> must be optional

      You specified an optional default but the placeholder that it affects
      is not an optional placeholder. For example:

          =item  -l[[en][gth]] <l>
      
          =for Euclid:
              l.opt_default: 123

      instead of:

          =item  -l[[en][gth]] [<l>]
      
          =for Euclid:
              l.opt_default: 123

    Invalid .opt_default constraint: Parameter %s must have a flag

      You specified an optional default but the parameter that it affects
      is unflagged. For example:

          =item  <length>
      
          =for Euclid:
              l.opt_default: 123

      instead of:

          =item  -l [<length>]
      
          =for Euclid:
              l.opt_default: 123

    Invalid .excludes value for variable %s: <%s> does not exist

      You specified to exclude a variable that was not seen in the POD.
      Make sure that this is not a typo.

    Invalid constraint: %s (No <%s> placeholder in argument: %s)

      You attempted to define a .type constraint for a placeholder that did
      not exist. Typically this is the result of the misspelling of a
      placeholder name:

          =item -foo <bar>
      
          =for Euclid:
              baz.type: integer

      or a =for Euclid: that has drifted away from its argument:

          =item -foo <bar>
      
          =item -verbose
      
          =for Euclid:
              bar.type: integer

    Getopt::Euclid loaded a second time

      You tried to load the module twice in the same program.
      Getopt::Euclid does not work that way. Load it only once.

    Unknown mode ('%s')

      The only argument that a use Getopt::Euclid command accepts is
      ':minimal_keys' (see "Minimalist keys"). You specified something else
      instead (or possibly forgot to put a semicolon after use
      Getopt::Euclid).

    Internal error: minimalist mode caused arguments '%s' and '%s' to clash

      Minimalist mode removes certain characters from the keys hat are
      returned in %ARGV. This can mean that two command-line options (such
      as --step and <step>) map to the same key (i.e. 'step'). This in turn
      means that one of the two options has overwritten the other within
      the %ARGV hash. The program developer should either turn off
      ':minimal_keys' mode within the program, or else change the name of
      one of the options so that the two no longer clash.

 Run-time diagnostics

    The following diagnostics are caused by problems in parsing the
    command-line

    Missing required argument(s): %s

      At least one argument specified in the REQUIRED ARGUMENTS POD section
      was not present on the command-line.

    Invalid %s argument. %s must be %s but the supplied value (%s) is not.

      Getopt::Euclid recognized the argument you were trying to specify on
      the command-line, but the value you gave to one of that argument's
      placeholders was of the wrong type.

    Unknown argument: %s

      Getopt::Euclid did not recognize an argument you were trying to
      specify on the command-line. This is often caused by command-line
      typos or an incomplete interface specification.

CONFIGURATION AND ENVIRONMENT

    Getopt::Euclid requires no configuration files or environment
    variables.

DEPENDENCIES

      * version

      * Pod::Select

      * Pod::PlainText

      * File::Basename

      * File::Spec::Functions

      * List::Util

      * Text::Balanced

      * IO::Pager::Page (recommended)

INCOMPATIBILITIES

    Getopt::Euclid may not work properly with POD in Perl files that have
    been converted into an executable with PerlApp or similar software. A
    possible workaround may be to move the POD to a __DATA__ section or a
    separate .pod file.

BUGS AND LIMITATIONS

    Please report any bugs or feature requests via GitHub issues
    <https://github.com/bigpresh/Getopt-Euclid/issues>

    Pull requests to fix bugs / add features welcome!

    GitHub repo <https://github.com/bigpresh/Getopt-Euclid>

    git clone git@github.com:bigpresh/Getopt-Euclid.git

    The RT queue
    <https://rt.cpan.org/Public/Dist/Display.html?Name=Getopt-Euclid> still
    exists but GitHub issues are preferred due to their integration with
    pull requests etc.

AUTHOR

    Damian Conway <DCONWAY@cpan.org>

    Florent Angly <florent.angly@gmail.com>

    David Precious (BIGPRESH) <davidp@preshweb.co.uk>

LICENCE AND COPYRIGHT

    Copyright (c) 2005, Damian Conway <DCONWAY@cpan.org>. All rights
    reserved.

    This module is free software; you can redistribute it and/or modify it
    under the same terms as Perl itself.

DISCLAIMER OF WARRANTY

    BECAUSE THIS SOFTWARE IS LICENSED FREE OF CHARGE, THERE IS NO WARRANTY
    FOR THE SOFTWARE, TO THE EXTENT PERMITTED BY APPLICABLE LAW. EXCEPT
    WHEN OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR OTHER
    PARTIES PROVIDE THE SOFTWARE "AS IS" WITHOUT WARRANTY OF ANY KIND,
    EITHER EXPRESSED OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
    WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE
    ENTIRE RISK AS TO THE QUALITY AND PERFORMANCE OF THE SOFTWARE IS WITH
    YOU. SHOULD THE SOFTWARE PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL
    NECESSARY SERVICING, REPAIR, OR CORRECTION.

    IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING
    WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY AND/OR
    REDISTRIBUTE THE SOFTWARE AS PERMITTED BY THE ABOVE LICENCE, BE LIABLE
    TO YOU FOR DAMAGES, INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL, OR
    CONSEQUENTIAL DAMAGES ARISING OUT OF THE USE OR INABILITY TO USE THE
    SOFTWARE (INCLUDING BUT NOT LIMITED TO LOSS OF DATA OR DATA BEING
    RENDERED INACCURATE OR LOSSES SUSTAINED BY YOU OR THIRD PARTIES OR A
    FAILURE OF THE SOFTWARE TO OPERATE WITH ANY OTHER SOFTWARE), EVEN IF
    SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH
    DAMAGES.