package Getopt::Euclid;

use version; our $VERSION = version->declare('0.4.5');

use warnings;
use strict;
use 5.005000; # perl 5.5.0
use Carp;
use Symbol ();
use re 'eval'; # for matcher regex
use Pod::Select;
use Pod::PlainText;
use File::Basename;
use File::Spec::Functions qw(splitpath catpath catfile);
use List::Util qw( first );
use Text::Balanced qw(extract_multiple extract_bracketed extract_variable extract_delimited);


# Set some module variables
my $skip_keyword = 'Getopt::Euclid'; # Ignore files with a first line containing this keyword.
my $pod_file_msg = "# This file was generated dynamically by $skip_keyword. Do not edit it.";

my $has_run = 0;
my $has_processed_pod = 0;
my $export_lvl = 1;
my @pod_names;
my $minimal_keys;
my $vars_prefix;
my $defer = 0;
my $matcher;
my %requireds;
my %options;
my %longnames;
our $man;    # --man     message
my $help;    # --help    message
my $usage;   # --usage   message
my $version; # --version message

my $optional_re;
$optional_re = qr{ \[ [^[]* (?: (??{$optional_re}) [^[]* )* \] }xms;


# Global variables
our $SCRIPT_NAME;
our $SCRIPT_VERSION; # for ticket # 55259


# Convert arg specification syntax to Perl regex syntax

my %std_matcher_for = (
    integer => '[+-]?\\d+(?:[eE][+]?\d+)?',
    number  => '[+-]?(?:\\d+\\.?\\d*|\\.\\d+)(?:[eE][+-]?\d+)?',
    input   => '\S+',
    output  => '\S+',
    string  => '\S+',
    q{}     => '\S+',
);

_make_equivalent(
    \%std_matcher_for,
    integer => [qw( int i +int +i 0+int 0+i +integer 0+integer )],
    number  => [qw( num n +num +n 0+num 0+n +number 0+number   )],
    input   => [qw( readable in )],
    output  => [qw( writable writeable out )],
    string  => [qw( str s )],
);

my %std_constraint_for = (
    'string'    => sub { 1 },            # Always okay (matcher ensures this)
    'integer'   => sub { 1 },            # Always okay (matcher ensures this)
    '+integer'  => sub { $_[0] > 0 },
    '0+integer' => sub { $_[0] >= 0 },
    'number'    => sub { 1 },            # Always okay (matcher ensures this)
    '+number'   => sub { $_[0] > 0 },
    '0+number'  => sub { $_[0] >= 0 },
    'input'     => sub { $_[0] eq '-' || -r $_[0] },
    'output'    => sub {
        my ( $vol, $dir ) = splitpath( $_[0] );
        $dir = ($vol && $dir) ? catpath($vol, $dir) : '.';
        $_[0] eq '-' ? 1 : -e $_[0] ? -w $_[0] : -w $dir;
    },
);

_make_equivalent(
    \%std_constraint_for,
    'integer'   => [qw(   int   i )],
    '+integer'  => [qw(  +int  +i )],
    '0+integer' => [qw( 0+int 0+i )],
    'number'    => [qw(   num   n )],
    '+number'   => [qw(  +num  +n )],
    '0+number'  => [qw( 0+num 0+n )],
    'string'    => [qw( str s )],
    'input'     => [qw( in readable )],
    'output'    => [qw( out writable writeable )],
);


sub Getopt::Euclid::Importer::DESTROY {
    return if $has_run || $^C;    # No errors when only compiling
    croak '.pm file cannot define an explicit import() when using Getopt::Euclid';
}


sub import {
    shift @_;
    @_ = grep { !( /:minimal_keys/     and $minimal_keys = 1             ) } @_;
    @_ = grep { !( /:vars(?:<(\w+)>)?/ and $vars_prefix  = $1 || 'ARGV_' ) } @_;
    @_ = grep { !( /:defer/            and $defer        = 1             ) } @_;
    croak "Unknown mode ('$_')" for @_;
    $export_lvl++ if not $defer;

    # No POD parsing and argument processing in Perl compile mode (ticket 34195)
    return if $^C;

    # Get name of caller program and its modules in @pod_names
    return unless _get_pod_names();

    # Extract POD of given files
    __PACKAGE__->process_pods( [reverse @pod_names] );
    undef @pod_names;
    $has_run = 1;

    # Parse POD + parse and export arguments

    ######
    #use Data::Dumper; print "ARGV: ".Dumper(\@ARGV);
    ######

    __PACKAGE__->process_args( \@ARGV ) unless $defer;

    return 1;
}


sub process_pods {
    # Extract POD content from list of Perl scripts (.pl) and modules (.pm) and
    # their corresponding .pod file if available. When given the argument
    # {-strict => 1}, do not look for .pod files.
    my ($self, $perl_files, $args) = @_;

    my $pod_string = '';
    open my $pod_fh, '>', \$pod_string
      or croak "Could not open filehandle to variable because $!";
    for my $perl_file (@$perl_files) {

        my $got_pod_file = 0;

        if ( not $args->{-strict} ) {

            # Find corresponding .pod file
            my ($name_re, $path, $suffix) = fileparse($perl_file, qr/\.[^.]*/);
            my $pod_file = catfile( $path, $name_re.'.pod' );

            # Get POD either from .pod file (preferably) or from Perl file
            if ( -e $pod_file ) {
                # Get .pod file content
                open my $in, '<', $pod_file
                  or croak "Could not open file $pod_file because $!";
                my $first_line = <$in>;
                chomp $first_line;
                if ( not ($first_line =~ m/$skip_keyword/) ) {
                    # Skip G::E auto-generated files since they lack important data
                    print $pod_fh "$first_line\n";
                    print $pod_fh $_ while <$in>;
                    $got_pod_file = 1;
                }
                close $in;
            }
        }

        if (not $got_pod_file) {
            # Parse POD content of Perl file
            podselect( {-output => $pod_fh}, $perl_file );
        }
        print $pod_fh "\n" if $pod_string;

    }
    close $pod_fh;
    $man = $pod_string;
    return 1;
}


sub process_args {
    # First, parse the POD specifications. Then, parse the given array of
    # arguments (\@ARGV or other) and populate %ARGV (or export specific
    # variable names).
    my ($self, $args, $options) = @_;

    # Parse POD
    if (not $has_processed_pod) {
        _parse_pod();
        $has_processed_pod = 1;
    }

    # Set options for argument parsing
    if (defined $options) {
       if (exists $options->{-minimal_keys}) {
          $minimal_keys = 1;
       }
       if (exists $options->{-vars}) {
          $vars_prefix = $options->{-vars};
       }
    }

    %ARGV = ();

    # Handle standard args...
    if ( first { $_ eq '--man' } @$args ) {
        _print_pod( __PACKAGE__->man(), 'paged' );
        exit;
    } elsif ( first { $_ eq '--usage' } @$args ) {
        print __PACKAGE__->usage();
        exit;
    } elsif ( first { $_ eq '--help' } @$args ) {
        _print_pod( __PACKAGE__->help(), 'paged' );
        exit;
    } elsif ( first { $_ eq '--version' } @$args ) {
        print __PACKAGE__->version();
        exit;
    } elsif ( first { $_ eq '--podfile' } @$args ) {
        # Option meant for authors
        my $podfile = podfile( );
        print "Wrote POD manual in file $podfile\n";
        exit;
    }

    # Subroutine to report problems during parsing...
    *_bad_arglist = sub {
        my (@msg) = @_;
        my $msg = join q{}, @msg;
        $msg = _rectify_arg($msg);
        $msg =~ s/\n?\z/\n/xms;
        warn "$msg\nTry this for usage help: $SCRIPT_NAME --help\n".
                   "Or this for full manual: $SCRIPT_NAME --man\n\n";
        exit 2;    # Traditional "bad arg list" value
    };

    # Run matcher...
    my $argv = join( q{ }, map { $_ = _escape_arg($_) } @$args );
    my $all_args_ref = { %options, %requireds };
    if ( my $error = _doesnt_match( $matcher, $argv, $all_args_ref ) ) {
        _bad_arglist($error);
    }

    # Check that all requireds have been found...
    my @missing;
    while ( my ($req) = each %requireds ) {
        push @missing, "\t$req\n" if !exists $ARGV{$req};
    }
    _bad_arglist(
        'Missing required argument',
        ( @missing == 1 ? q{} : q{s} ),
        ":\n", @missing
    ) if @missing;

    # Back-translate \0-quoted spaces and \1-quoted tabs...
    _rectify_all_args();

    # Check exclusive variables, variable constraints and fill in defaults...
    _verify_args($all_args_ref);

    # Clean up @$args since everything must have been parsed
    @$args = ();

    # Clean up %ARGV
    for my $arg_name ( keys %ARGV ) {

        # Flatten non-repeatables...
        my $vals = delete $ARGV{$arg_name};
        my $repeatable = $all_args_ref->{$arg_name}{is_repeatable};
        if ($repeatable) {
            pop @{$vals};
        }

        for my $val ( @{$vals} ) {
            my $var_count = keys %{$val};
            $val = $var_count == 0
              ? 1    # Boolean -> true
              : $var_count == 1
              ? ( values %{$val} )[0]    # Single var -> var's val
              : $val                     # Otherwise keep hash
              ;
            my $false_vals = $all_args_ref->{$arg_name}{false_vals};
            my %vars_opt_vals;

            for my $arg_flag ( _get_variants($arg_name) ) {
                my $variant_val = $val;
                if ( $false_vals && $arg_flag =~ m{\A $false_vals \z}xms ) {
                    $variant_val = $variant_val ? 0 : 1;
                }

                if ($repeatable) {
                    push @{ $ARGV{$arg_flag} }, $variant_val;
                } else {
                    $ARGV{$arg_flag} = $variant_val;
                }
                $vars_opt_vals{$arg_flag} = $ARGV{$arg_flag} if $vars_prefix;
            }

            if ($vars_prefix) {
                _minimize_entries_of( \%vars_opt_vals );
                my $maximal = _longestname( keys %vars_opt_vals );
                _export_var( $vars_prefix, $maximal, $vars_opt_vals{$maximal} );
                delete $longnames{$maximal};
            }
        }
    }

    if ($vars_prefix) {

        # Export any unspecified options to keep use strict happy
        while ( my ($opt_name, $arg_name) = each %longnames ) {
            my $arg_info = $all_args_ref->{$arg_name};
            my $val;
            if ( $arg_info->{is_repeatable} or $arg_name =~ />\.\.\./ ) {
               # Empty arrayref for repeatable options
               $val = [];
            } else {               
               if (keys %{ $arg_info->{var} } > 1) {
                   # Empty hashref for non-repeatable options with multiple placeholders
                   $val = {};
               }
            }
            _export_var( $vars_prefix, $opt_name, $val );
        }
    }


    if ($minimal_keys) {
        _minimize_entries_of( \%ARGV );
    }

    return 1;
}


sub podfile {
    # Write the given POD doc into a .pod file, overwriting any existing .pod file
    return if not -e $0;
    my ($name_re, $path, $suffix) = fileparse($0, qr/\.[^.]*/);
    my $pod_file = catfile( $path, $name_re.'.pod' );
    open my $out_fh, '>', $pod_file or croak "Could not write file $pod_file because $!";
    print $out_fh $pod_file_msg."\n\n".__PACKAGE__->man();
    close $out_fh;
    return $pod_file;
}


sub man {
    return $man;
}


sub usage {
    return $usage;
}


sub help {
    return $help;
}


sub version {
    return $version;
}


# # # # # # # # Utility subs # # # # # # # #

# Recursively remove decorations on %ARGV keys

sub AUTOLOAD {
    our $AUTOLOAD;
    $AUTOLOAD =~ s{.*::}{main::}xms;
    no strict 'refs';
    goto &$AUTOLOAD;
}


sub _parse_pod {
    # Set up parsing rules...
    my $space_re      = qr{ [^\S\n]*               }xms;
    my $head_start_re = qr{ ^=head1                }xms;
    my $head_end_re   = qr{ (?= $head_start_re | \z)  }xms;
    my $pod_cmd_re    = qr{            = [^\W\d]\w+ [^\n]* (?= \n\n )}xms;
    my $pod_cut_re    = qr{ (?! \n\n ) = cut $space_re        (?= \n\n )}xms;

    my $name_re  = qr{ $space_re NAME    $space_re \n }xms;
    my $vers_re  = qr{ $space_re VERSION $space_re \n }xms;
    my $usage_re = qr{ $space_re USAGE   $space_re \n }xms;

    my $std_re = qr{ STANDARD | STD | PROGRAM | SCRIPT | CLI  | COMMAND(?:-|\s)?LINE }xms;
    my $arg_re = qr{ $space_re (?:PARAM(?:ETER)?|ARG(?:UMENT)?)S? }xms;

    my $options_re  = qr{ $space_re $std_re? $space_re OPTION(?:AL|S)?        $arg_re? $space_re \n }xms;
    my $required_re = qr{ $space_re $std_re? $space_re (?:REQUIRED|MANDATORY) $arg_re? $space_re \n }xms;

    my $euclid_arg = qr{ ^=item \s* ([^\n]*?) \s* \n\s*\n
                        (
                        .*?
                        (?:
                            ^=for \s* (?i: Euclid) .*? \n\s*\n
                            | (?= ^=[^\W\d]\w* | \z)
                        )
                        )
                    }xms;

    # Clean up line delimiters
    $man =~ s{ [\n\r] }{\n}gx;

    # Clean up significant entities...
    $man =~ s{ E<lt> }{<}gxms;
    $man =~ s{ E<gt> }{>}gxms;

    # Put program name in man
    $SCRIPT_NAME = (-e $0) ? (splitpath $0)[-1] : 'one-liner';
    $man =~ s{ ($head_start_re $name_re \s*) .*? (- .*)? $head_end_re }
              {$1.$SCRIPT_NAME.($2 ? " $2" : "\n\n")}xems;

    # Put version number in man
    ($SCRIPT_VERSION) = 
        $man =~ m/$head_start_re $vers_re .*? (\d+(?:[._]\d+)+) .*? $head_end_re /xms;
    if ( !defined $SCRIPT_VERSION ) {
        $SCRIPT_VERSION = $main::VERSION;
    }
    if ( !defined $SCRIPT_VERSION ) {
        $SCRIPT_VERSION = (-e $0) ? localtime((stat $0)[9]) : 'one-liner';
    }
    $man =~ s{ ($head_start_re $vers_re    \s*) .*? (\s*) $head_end_re }
             {$1This document refers to $SCRIPT_NAME version $SCRIPT_VERSION $2}xms;

    # Extra info from PODs
    my ($options, $opt_name, $required, $req_name, $licence);
    while ($man =~ m/$head_start_re ($required_re) (.*?) $head_end_re /gxms) {
        # Required arguments
        my ( $more_req_name, $more_required ) = ($1, $2);
        $req_name = $more_req_name if not defined $req_name;
        $required = ( $more_required || q{} ) . ( $required || q{} );
    }
    while ($man =~ m/$head_start_re ($options_re)  (.*?) $head_end_re /gxms) {
        # Optional arguments
        my ( $more_opt_name, $more_options ) = ($1, $2);
        $opt_name = $more_opt_name if not defined $opt_name;
        $options = ( $more_options || q{} ) . ( $options || q{} );
    }
    while ($man =~ m/$head_start_re [^\n]+ (?i: licen[sc]e | copyright ) .*? \n \s* (.*?) \s* $head_end_re /gxms) {
        # License information
        my ($more_licence) = ($1, $2);
        $licence = ( $more_licence || q{} ) . ( $licence || q{} );
    }

    # Clean up interface titles...
    for my $name_re ( $opt_name, $req_name ) {
        next if !defined $name_re;
        $name_re =~ s{\A \s+ | \s+ \z}{}gxms;
    }

    # Extract the actual interface and store each arg entry into a hash of specifications...
    my $seq  = 0;
    my $seen = {};
    while ( ( $required || q{} ) =~ m{ $euclid_arg }gxms ) {
        $seen = _register_specs( $1, $2, $seq, \%requireds, \%longnames, $seen );
        $seq++;
    }
    while ( ( $options  || q{} ) =~ m{ $euclid_arg }gxms ) {
        $seen = _register_specs( $1, $2, $seq, \%options, \%longnames, $seen );
        $seq++;
    }
    undef $seen;
    _minimize_entries_of( \%longnames );

    # Extract Euclid information...
    my $all_specs = {%requireds, %options};
    _process_euclid_specs( $all_specs );

    # Insert default values (if any) in the program's documentation
    $required = _insert_default_values(\%requireds);
    $options  = _insert_default_values(\%options  );

    # One-line representation of interface...
    my $arg_summary = join ' ', (sort
       { $requireds{$a}{'seq'} <=> $requireds{$b}{'seq'} }
       (keys %requireds));

    1 while $arg_summary =~ s/\[ [^][]* \]//gxms;

    if ($opt_name) {
      $arg_summary .= ' ' if $arg_summary;
      $arg_summary .= lc "[$opt_name]";
    }
    $arg_summary =~ s/\s+/ /gxms;

    # Manual message
    $man =~ s{ ($head_start_re $usage_re    \s*) .*? (\s*) $head_end_re } {$1$SCRIPT_NAME $arg_summary$2}xms;
    $man =~ s{ ($head_start_re $required_re \s*) .*? (\s*) $head_end_re } {$1$required$2}xms;
    $man =~ s{ ($head_start_re $options_re  \s*) .*? (\s*) $head_end_re } {$1$options$2}xms;

    # Usage message
    $usage  = "       $SCRIPT_NAME $arg_summary\n";
    $usage .= "       $SCRIPT_NAME --help\n";
    $usage .= "       $SCRIPT_NAME --man\n";
    $usage .= "       $SCRIPT_NAME --usage\n";
    $usage .= "       $SCRIPT_NAME --version\n";

    # Help message
    $help  = "=head1 \L\uUsage:\E\n\n$usage\n";
    $help .= "=head1 \L\u$req_name:\E\n\n$required\n\n"
      if ( $req_name || q{} ) =~ /\S/;
    $help .= "=head1 \L\u$opt_name:\E\n\n$options\n\n"
      if ( $opt_name || q{} ) =~ /\S/;

    $usage  = "Usage:\n".$usage;

    # Version message
    $version  = "This is $SCRIPT_NAME version $SCRIPT_VERSION\n";
    $version .= "\n$licence\n" if $licence;

    # Convert arg specifications to regexes...
    _convert_to_regex( $all_specs );

    # Build matcher...
    my @arg_list = ( values(%requireds), values(%options) );
    $matcher = join '|', map { $_->{matcher} }
      sort( { $b->{name} cmp $a->{name} } grep { $_->{name} =~ /^[^<]/ } @arg_list ),
      sort( { $a->{seq}  <=> $b->{seq}  } grep { $_->{name} =~ /^[<]/  } @arg_list );
    $matcher .= '|(?> (.+)) (?{ push @errors, $^N }) (?!)';
    $matcher = '(?:' . $matcher . ')';

    return 1;
}


sub _register_specs {
    my ($name_re, $spec, $seq, $storage, $longnames, $seen) = @_;
    my @variants = _get_variants($name_re);
    $storage->{$name_re} = {
        seq      => $seq,
        src      => $spec,
        name     => $name_re,
        variants => \@variants,
    };
    if ($minimal_keys) {
        my $minimal = _minimize_name($name_re);
        croak "Internal error: minimalist mode caused arguments ".
           "'$name_re' and '".$seen->{$minimal}."' to clash"
           if $seen->{$minimal};
        $seen->{$minimal} = $name_re;
    }
    $longnames->{ _longestname(@variants) } = $name_re;
    return $seen;
}


sub _process_euclid_specs {
    my ($args) = @_;
    my %all_var_list;
    my %excluded_by_def;

  ARG:
    while ( (undef, my $arg) = each %$args ) {

        # Validate and record variable names seen here...
        my $var_list = _validate_name( $arg->{name} );
        while (my ($var_name, undef) = each %$var_list) {
            $all_var_list{$var_name} = undef;
        }

        # Process arguments with a Euclid specification further
        $arg->{src} =~ s{^ =for \s+ Euclid\b [^\n]* \s* (.*) \z}{}ixms
            or next ARG;
        my $info = $1;

        $arg->{is_repeatable} = $info =~ s{^ \s* repeatable \s*? $}{}xms;

        my @false_vals;
        while ( $info =~ s{^ \s* false \s*[:=] \s* ([^\n]*)}{}xms ) {
            my $regex = $1;
            1 while $regex =~ s/ \[ ([^]]*) \] /(?:$1)?/gxms;
            $regex =~ s/ (\s+) /$1.'[\\s\\0\\1]*'/egxms;
            push @false_vals, $regex;
        }
        if (@false_vals) {
            $arg->{false_vals} = '(?:' . join( '|', @false_vals ) . ')';
        }

        while (
            $info =~ m{\G \s* (([^.]+)\.([^:=\s]+) \s*[:=]\s* ([^\n]*)) }gcxms )
        {
            my ( $spec, $var, $field, $val ) = ( $1, $2, $3, $4 );

            # Check for misplaced fields...
            if ( $arg->{name} !~ m{\Q<$var>}xms ) {
                _fail( "Invalid constraint: $spec\n(No <$var> placeholder in ".
                    "argument: $arg->{name})" );
            }

            # Decode...
            if ( $field eq 'type.error' ) {
                $arg->{var}{$var}{type_error} = $val;
            } elsif ( $field eq 'type' ) {
                $val = _qualify_variables_fully( $val );
                my ( $matchtype, $comma, $constraint ) =
                  $val =~ m{(/(?:\.|.)+/ | [^,\s]+)\s*(?:(,))?\s*(.*)}xms;
                $arg->{var}{$var}{type} = $matchtype;
                if ( $comma && length $constraint ) {
                    ( $arg->{var}{$var}{constraint_desc} = $constraint ) =~
                      s/\s*\b\Q$var\E\b\s*//g;
                    $constraint =~ s/\b\Q$var\E\b/\$_[0]/g;
                    $arg->{var}{$var}{constraint} = eval "sub{ $constraint }"
                      or _fail("Invalid .type constraint: $spec\n($@)");
                } elsif ( length $constraint ) {
                    $arg->{var}{$var}{constraint_desc} = $constraint;
                    $arg->{var}{$var}{constraint} = 
                      eval "sub{ \$_[0] $constraint }"
                      or _fail("Invalid .type constraint: $spec\n($@)");
                } else {
                    $arg->{var}{$var}{constraint_desc} = $matchtype;
                    $arg->{var}{$var}{constraint} =
                      $matchtype =~ m{\A\s*/.*/\s*\z}xms
                      ? sub { 1 }
                      : $std_constraint_for{$matchtype}
                      or _fail("Unknown .type constraint: $spec");
                }

            } elsif ( ($field eq 'default') || ($field eq 'opt_default') ) {
                $val = _qualify_variables_fully( $val );
                eval "\$val = $val; 1"
                  or _fail("Invalid .$field value: $spec\n($@)");
                $arg->{var}{$var}{$field} = $val;
                my $has_field = 'has_'.$field;
                $arg->{$has_field} = exists $arg->{$has_field} ?
                                      $arg->{$has_field}++ :
                                      1;

                if ($field eq 'opt_default') {
                    # Check that placeholders with optional defaults have a flagged argument
                    if ( $arg->{name} =~ m{^<}xms ) {
                       _fail( "Invalid .$field constraint: $spec\nParameter ".
                           "$arg->{name} must have a flag" );
                    }
                    # Check that placeholders with optional defaults is optional
                    if ( $arg->{name} !~ m{\Q[<$var>]}xms ) {
                       _fail( "Invalid .$field constraint: $spec\nPlaceholder".
                           " <$var> must be optional, i.e. [<$var>], to have ".
                           "an optional default in argument: $arg->{name}" );
                    }
                }

            } elsif ( $field eq 'excludes.error' ) {
                $arg->{var}{$var}{excludes_error} = $val;
            } elsif ( $field eq 'excludes' ) {
                $arg->{var}{$var}{excludes} = [ split '\s*,\s*', $val ];
                for my $excl_var (@{$arg->{var}{$var}{excludes}}) {
                    if ($var eq $excl_var) {
                        _fail( "Invalid .excludes value for variable <$var>: ".
                            "<$excl_var> cannot exclude itself." );
                    }
                }
            } else {
                _fail("Unknown specification: $spec");
            }
        }
        # Record variables excluded by another that has a default
        while (my ($var_name, $var_data) = each %{$arg->{var}}) {
            for my $excl_var (@{$arg->{var}{$var_name}{excludes}}) {
                $excluded_by_def{$excl_var}{default}{$var_name}     = 1 if $arg->{has_default};
                $excluded_by_def{$excl_var}{opt_default}{$var_name} = 1 if $arg->{has_opt_default};
            }
        }
        if ( $info =~ m{\G \s* ([^\s\0\1] [^\n]*) }gcxms ) {
            _fail("Unknown specification: $1");
        }
    }

    # Validate and complete .excludes specs

    while ( (undef, my $arg) = each %$args ) {
        while ( my ($var, $var_specs) = each %{$arg->{var}} ) {
            # Check for invalid placeholder name in .excludes specifications
            for my $excl_var (@{$var_specs->{excludes}}) {
                if (not exists $all_var_list{$excl_var}) {
                    _fail( "Invalid .excludes value for variable <$var>: ".
                        "<$excl_var> does not exist\n" );
                }
            }
            # Remove default for placeholders excluded by others that have a default
            for my $type ( 'default', 'opt_default' ) {
                if ( (exists $arg->{var}->{$var}->{$type}) && (exists $excluded_by_def{$var}{$type}) ) {
                    delete $arg->{var}->{$var}->{$type};
                    $arg->{"has_$type"}--;
                    if ($arg->{"has_$type"} == 0) {
                        delete $arg->{"has_$type"};
                    }
                }
            }
        }
    }

    return 1;
}


sub _qualify_variables_fully {
    # Restore fully-qualified name to variables:
    #    $x          becomes  $main::x
    #    $::x        becomes  $main::x
    #    $Package::x stays as $Package::x
    #    /^asdf$/    stays as /^asdf$/
    #    '$10'       stays as '$10'
    # Note: perlvar indicates that ' can also be used instead of ::
    my ($val) = @_;
    if ($val =~ m/[\$\@\%]/) { # Avoid expensive Text::Balanced operations when there are no variables
        my $new_val;
        for my $s (extract_multiple($val,[{Quoted=>sub{extract_delimited($_[0])}}],undef,0)) {
            if (not ref $s) {
                # A non-quoted section... may contain variables to fix
                for my $var_name ( @{_get_variable_names($s)} ) {
                    # Skip fully qualified names, such as '$Package::x'
                    next if $var_name =~ m/main(?:'|::)/;
                    # Remove sigils from beginning of variable name: $ @ % {
                    $var_name =~ s/^[\$\@\%\{]+//;
                    # Substitute non-fully qualified vars, e.g. '$x' or '$::x', by '$main::x'
                    my $new_name = Symbol::qualify($var_name, 'main');
                    next if $new_name eq $var_name;
                    $var_name = quotemeta( $var_name );
                    $s =~ s/$var_name/$new_name/;
                }
                $new_val .= $s;
            } else {
                # A quoted section, to keep as-is
                $new_val .= $$s;
            }
        }
        return $new_val;
    } else {
        return $val;
    }
}


sub _get_variable_names {
    # Get an arrayref of the variables names found in the provided string.
    # This function is a hack, needed only because of Text::Balanced ticket #78855:
    #    https://rt.cpan.org/Public/Bug/Display.html?id=78855
    my ($str) = @_;
    my $vars = [];
    for my $var (extract_multiple($str,[sub{extract_variable($_[0],'')}],undef,1)) {
        # Name must start with underscore or a letter, e.g. $t $$h{a} ${$h}{a} $h->{a} @_
        # Skip special or invalid names, e.g. $/ $1
        my $tmp = $var;
        $tmp =~ s/(?:{|})//g;
        next if not $tmp =~ m/^[\$\@\%]+[_a-z]/i;
        push @$vars, $var;
    }
    return $vars;
}


sub _minimize_name {
    my ($name_re) = @_;
    $name_re =~ s{[][]}{}gxms;                      # remove all square brackets
    $name_re =~ s{\A \W+ ([\w-]*) .* \z}{$1}gxms;
    $name_re =~ s{-}{_}gxms;
    return $name_re;
}


sub _minimize_entries_of {
    my ($arg_ref) = @_;
    return if ref $arg_ref ne 'HASH';

    for my $old_key (keys %$arg_ref) {
        my $new_key = _minimize_name($old_key);
        $arg_ref->{$new_key} = delete $arg_ref->{$old_key};
    }

    return 1;
}


# Do match, recursively trying to expand cuddles...
sub _doesnt_match {
    my ( $matcher, $argv, $arg_specs_ref ) = @_;

    our @errors; # 'our' instead of 'my' because it is needed for the re pragma
    local @errors = ();
    %ARGV = ();

    # Match arguments, populate %ARGV and @errors
    # Note that the matcher needs the pragma: use re 'eval';
    $argv =~ m{\A (?: \s* $matcher )* \s* \z}xms;

    # Report errors in passed arguments
    for my $error (@errors) {
        if ( $error =~ m/\A ((\W) (\w) (\w+))/xms ) {
            my ( $bundle, $marker, $firstchar, $chars ) = ( $1, $2, $3, $4 );
            $argv =~ s{\Q$bundle\E}{$marker$firstchar $marker$chars}xms;
            return if !_doesnt_match( $matcher, $argv, $arg_specs_ref );
        }
      ARG:
        for my $arg_spec_ref ( values %{$arg_specs_ref} ) {
            our $bad_type;
            local $bad_type;
            next ARG
              if $error !~ m/\A [\s\0\1]* ($arg_spec_ref->{generic_matcher})/xms
                  || !$bad_type;
            
            my $msg = _type_error( $bad_type->{arg}, $bad_type->{var},
                $bad_type->{val}, $bad_type->{type}, $bad_type->{type_error} );
            return $msg;
        }
        return "Unknown argument: $error";
    }

    return 0;    # No error
}


sub _escape_arg {
    my $arg = shift;
    my ($num_replaced) = ($arg =~ tr/ \t/\0\1/);
    return $arg;
}


sub _rectify_arg {
    my $arg = shift;
    my ($num_replaced) = ($arg =~ tr/\0\1/ \t/);
    return $arg;
}


sub _rectify_all_args {
    while ( my (undef, $arg_list) = each %ARGV ) {
        for my $arg ( @{$arg_list} ) {
            if ( ref $arg eq 'HASH' ) {
                for my $var ( values %{$arg} ) {
                    if ( ref $var eq 'ARRAY' ) {
                        $var = [ map { _rectify_arg($_) } @{$var} ];
                    } else {
                        $var = _rectify_arg($var);
                    }
                }
            } else {
                if ( ref $arg eq 'ARRAY' ) {
                    $arg = [ map { _rectify_arg($_) } @{$arg} ];
                } else {
                    $arg = _rectify_arg($arg);
                }
            }
        }
    }
    return 1;
}


sub _verify_args {
    my ($arg_specs_ref) = @_;
    # Check exclusive variables, variable constraints and fill in defaults...
    # Handle mutually exclusive arguments
    my %seen_vars; 
    while ( my ($arg_name, $arg_elems) = each %ARGV ) {
        for my $elem (@{$arg_elems}) {
            while ( my ($var_name) = each (%{$elem}) ) {
                $seen_vars{$var_name} = $arg_name if $var_name;
            }
        }
    }

    while ( my ($arg_name, $arg) = each %{$arg_specs_ref} ) {
        while ( my ($var_name, $var) = each %{$arg->{var}} ) {
            # Enforce placeholders that cannot be specified with others
            for my $excluded_var ( @{$var->{excludes}} ) {
                if (exists $seen_vars{$var_name} && 
                    exists $seen_vars{$excluded_var}) {
                    my $excl_arg = $seen_vars{$excluded_var};
                    my $msg;
                    if (exists $var->{excludes_error}) {
                        $msg = $var->{excludes_error};
                    } else {
                        $msg =
                            qq{Invalid "$excl_arg" argument.\n<$excluded_var> }.
                            qq{cannot be specified with <$var_name> because }.
                            qq{argument "$arg_name" excludes <$excluded_var>};
                    }
                    _bad_arglist($msg);                
                }
            }
        }
    }

    # Enforce constraints and fill in defaults...
  ARG:
    while (my ($arg_name, $arg_specs) = each %{$arg_specs_ref} ) {

        # Skip non-existent/non-defaulting/non-optional-defaulting arguments
        next ARG
          if !exists $ARGV{$arg_name}
              && !(   $arg_specs->{has_default}
                   || $arg_specs->{has_opt_default} );

        # Ensure all vars exist within arg...
        my @vars = keys %{$arg_specs->{placeholders}};
        for my $index ( 0 .. $#{ $ARGV{$arg_name} } ) {
            my $entry = $ARGV{$arg_name}[$index];
            @{$entry}{@vars} = @{$entry}{@vars};

            # Get arg specs...
          VAR:
            for my $var (@vars) {

                my $arg_vars = $arg_specs->{var}->{$var};

                # Check constraints on vars...
                if ( exists $ARGV{$arg_name} ) {

                    if ( ref $entry eq 'HASH' && defined $entry->{$var} ) {
                        # Named vars...
                        for my $val (
                            ref $entry->{$var} eq 'ARRAY'
                            ? @{ $entry->{$var} }
                            : $entry->{$var}
                          )
                        {
                            if ( $arg_vars->{constraint} &&
                                !$arg_vars->{constraint}->($val) ) {
                                _bad_arglist( _type_error($arg_name, $var, $val,
                                    $arg_vars->{constraint_desc},
                                    $arg_vars->{type_error}) );
                            }
                        }
                        next VAR;
                    } elsif ( ref $entry ne 'HASH' && defined $entry ) {
                        # Unnamed vars...
                        for my $val (
                            ref $entry eq 'ARRAY'
                            ? @{$entry}
                            : $entry
                          )
                        {
                            if ( $arg_vars->{constraint} &&
                                !$arg_vars->{constraint}->($val) ) {
                                _bad_arglist( _type_error( $arg_name, $var, $val,
                                    $arg_vars->{constraint_desc},
                                    $arg_vars->{type_error}) );
                            }
                            $entry->{$var} = ''
                              unless defined( $ARGV{$arg_name} );
                        }
                        next VAR;
                    }
                }

                # Assign placeholder defaults (if necessary)...
                next ARG
                  if   !exists $arg_vars->{default}
                    && !exists $arg_vars->{opt_default};

                $entry->{$var} = exists $arg_vars->{opt_default} ?
                                 $arg_vars->{opt_default} :
                                 $arg_vars->{default};
            }
        }

        # Handle defaults for missing args...
        if ( !@{ $ARGV{$arg_name} } ) {
            for my $var (@vars) {
                # Assign defaults (if necessary)...
                my $arg_vars = $arg_specs->{var}->{$var};
                next ARG
                  if !exists $arg_vars->{default}; # no default specified

                # Omit default if it conflicts with a specified parameter
                for my $excl_var ( @{$arg_specs->{var}->{$var}->{excludes}} ) {
                    if (exists $seen_vars{$excl_var}) {
                        next ARG;
                    }
                }

                $ARGV{$arg_name}[0]{$var} = $arg_vars->{default};
            }
        }
    }
    return 1;
}


sub _type_error {
    my ($arg_name, $var_name, $var_val, $var_constraint, $var_error) = @_;
    my $msg = qq{Invalid "$arg_name" argument.\n};
    $var_name =~ s{\W+}{}gxms;
    if ( $var_error ) {
        $msg = $var_error;
        $msg =~ s{(?<!<)\b$var_name\b|\b$var_name\b(?!>)}{$var_val}gxms;
    } else {
        $msg = qq{<$var_name> must be $var_constraint but the supplied value }.
               qq{("$var_val") is not.};
    }
    return $msg;
}


sub _convert_to_regex {
    my ($args_ref) = @_;

    # Regexp to capture the start of a new argument
    my $no_esc_ws = '(?!\0)'; # no escaped whitespaces

    my @arg_variants;
    while ( my ($arg_name, $arg_specs) = each %{$args_ref} ) {
        push @arg_variants, @{$arg_specs->{variants}};
    }

    my $no_match = join('|',@arg_variants);
    $no_match = _escape_specials($no_match);
    $no_match = '(?!(?:'.$no_match.')'.$no_esc_ws.')';

    while ( my ($arg_name, $arg) = each %{$args_ref} ) {
        my $regex = $arg_name;

        # Quotemeta specials...
        $regex = _escape_specials($regex);
        $regex = "(?:$regex)";

        # Convert optionals...
        1 while $regex =~ s/ \[ ([^]]*) \] /(?:$1)?/gxms;
        $regex =~ s/ (\s+) /$1.'\s*'.$no_esc_ws/egxms;
        my $generic = $regex;

        # Set the matcher
        $regex =~
            s{ < (.*?) >(\.\.\.|) }
             { my ($var_name, $var_rep) = ($1, $2);
               $var_name =~ s/(\s+)\[\\s\\0\\1]\*/$1/gxms;
               my $type = $arg->{var}{$var_name}{type} || q{};
               $arg->{placeholders}->{$var_name} = undef;
               my $matcher =
                   $type =~ m{\A\s*/.*/\s*\z}xms
                   ? eval "qr$type"
                   : $std_matcher_for{ $type }
                   or _fail("Unknown type ($type) in specification: $arg_name");
               $var_rep ?
                 "(?:[\\s\\0\\1]*$no_match($matcher)(?{push \@{(\$ARGV{q{$arg_name}}||=[{}])->[-1]{q{$var_name}}}, \$^N}))+"
                 :
                 "(?:$no_match($matcher)(?{(\$ARGV{q{$arg_name}}||=[{}])->[-1]{q{$var_name}} = \$^N}))";
             }gexms
             or do {
                 $regex .= "(?{(\$ARGV{q{$arg_name}}||=[{}])->[-1]{q{}} = 1})";
             };

        if ( $arg->{is_repeatable} ) {
            $arg->{matcher} = "$regex (?:(?<!\\w)|(?!\\w)) (?{push \@{\$ARGV{q{$arg_name}}}, {} })";
        } else {
            $arg->{matcher} = "(??{exists\$ARGV{q{$arg_name}}?'(?!)':''}) "
              . (
                $arg->{false_vals}
                ? "(?:$arg->{false_vals} (?:(?<!\\w)|(?!\\w)) (?{\$ARGV{q{$arg_name}} ||= [{ q{} => 0 }] }) | $regex (?:(?<!\\w)|(?!\\w)) (?{\$ARGV{q{$arg_name}} ||= [{ q{} => 1}] }))"
                : "$regex (?:(?<!\\w)|(?!\\w)) (?{\$ARGV{q{$arg_name}} ||= [{}] })"
              );
        }

        # Set the generic matcher
        $generic =~
            s{ < (.*?) > }
             { my $var_name = $1;
               $var_name =~ s/(\s+)\[\\s\\0\\1]\*/$1/gxms;
               my $type = $arg->{var}{$var_name}{type} || q{};
               my $type_error = $arg->{var}{$var_name}{type_error} || q{};
               my $matcher = $type =~ m{\A\s*/.*/\s*\z}xms
                                ? eval "qr$type"
                                : $std_matcher_for{ $type };
               "(?:($matcher|([^\\s\\0\\1]+)"
               . "(?{\$bad_type ||= "
               . "{arg=>q{$arg_name},type=>q{$type},type_error=>q{$type_error}, var=>q{<$var_name>},val=>\$^N};})))"
             }gexms;
        $arg->{generic_matcher} = $generic;
    }
    return 1;
}


sub _escape_specials {
    # Escape quotemeta special characters
    my $arg = shift;
    $arg =~ s{([@#\$^*()+{}?])}{\\$1}gxms;
    return $arg;
}


sub _print_pod {
    my ( $pod, $paged ) = @_;

    if ($paged) {
        # Page output
        eval { require IO::Pager::Page } or eval { require IO::Page };
    }
  
    # Convert POD to plaintext, wrapping the lines at 76 chars and print to STDOUT
    open my $parser_in, '<', \$pod or croak "Could not read from variable because $!";
    Pod::PlainText->new()->parse_from_filehandle($parser_in);
    close $parser_in;

    return 1;
}


sub _validate_name {
    # Check that the argument name only has pairs of < > brackets (ticket 34199)
    # Return the name of the variables that this argument specifies
    my ($name) = @_;
    if ($name =~ m/[<>]/) { # skip expensive Text::Balance functions if possible
        my %var_names;
        my $pos = 0;
        for my $s (extract_multiple($name,[sub{extract_bracketed($_[0],'<>')}],undef,0)) {
            next if not $s =~ m/[<>]/;
            $s =~ s/^<(.*)>$/$1/;
            if ( $s =~ m/[<>]/ ) {
                _fail( 'Invalid argument specification: '.$name );
            }
            $pos++;
            $var_names{$s} = $pos if not exists $var_names{$s};
        }
        return \%var_names;
    } else {    
        return {};
    }
}


sub _get_variants {
    my @arg_desc = shift =~ m{ [^[|]+ (?: $optional_re [^[|]* )* }gmxs;

    for (@arg_desc) {
        s{^ \s+ | \s+ $}{}gxms;
    }

    # Only consider first "word"...
    return $1 if $arg_desc[0] =~ m/\A (< [^>]+ >)/xms;

    $arg_desc[0] =~ s/\A ([^\s<]+) \s* (?: < .*)? \z/$1/xms;

    # Variants are all those with and without each optional component...
    my %variants;
    while (@arg_desc) {
        my $arg_desc_with    = shift @arg_desc;
        my $arg_desc_without = $arg_desc_with;

        if ( $arg_desc_without =~ s/ \[ [^][]* \] //xms ) {
            push @arg_desc, $arg_desc_without;
        }
        if ( $arg_desc_with =~ m/ [[(] ([^][()]*) [])] /xms ) {
            my $option = $1;
            for my $alternative ( split /\|/, $option ) {
                my $arg_desc = $arg_desc_with;
                $arg_desc =~ s{[[(] [^][()]* [])]}{$alternative}xms;
                push @arg_desc, $arg_desc;
            }
        }

        $arg_desc_with =~ s/[][]//gxms;
        $arg_desc_with =~ s/\b[^-\w] .* \z//xms;
        $variants{$arg_desc_with} = 1;
    }

    return keys %variants;
}


sub _longestname {
    return ( sort { length $a <=> length $b || $a cmp $b } @_ )[-1];
}


sub _export_var {
    my ( $prefix, $key, $value ) = @_;
    my $export_as = $prefix . $key;
    $export_as =~ s{\W}{_}gxms;    # mainly for '-'
    my $callpkg = caller( $export_lvl + ($Exporter::ExportLevel || 0) );
    no strict 'refs';
    *{"$callpkg\::$export_as"} = ( ref $value ) ? $value : \$value;
    return 1;
}


# Utility sub to factor out hash key aliasing...
sub _make_equivalent {
    my ( $hash_ref, %alias_hash ) = @_;

    while ( my ( $name_re, $aliases ) = each %alias_hash ) {
        for my $alias (@$aliases) {
            $hash_ref->{$alias} = $hash_ref->{$name_re};
        }
    }

    return 1;
}


# Report problems in specification and die
sub _fail {
    my (@msg) = @_;
    croak "Getopt::Euclid: @msg";
}


sub _get_pod_names {
    # Parse the POD of the caller program and its modules.
    my @caller = caller(1);

    # Sanity check
    if ($has_run) {
        carp 'Getopt::Euclid loaded a second time';
        warn "Second attempt to parse command-line was ignored\n";
        return 0;
    }

    # Handle calls from .pm files
    if ( $caller[1] =~ m/[.]pm \z/xms ) {
        my @caller = caller(1); # at import()'s level
        push @pod_names, $caller[1];
        # Install this import() sub as module's import sub...
        no strict 'refs';
        croak '.pm file cannot define an explicit import() when using Getopt::Euclid'
          if *{"$caller[0]::import"}{CODE};
        my $lambda;    # Needed so the anon sub is generated at run-time
        *{"$caller[0]::import"}
          = bless sub { $lambda = 1; goto &Getopt::Euclid::import },
          'Getopt::Euclid::Importer';

        return 0;
    }

    # Add name of caller program
    push @pod_names, $0 if (-e $0); # When calling perl -e '...', $0 is '-e', i.e. not a actual file

    return 1;
}


sub _insert_default_values {
    my ($args) = @_;
    my $pod_string = '';
    # Retrieve item names in sequential order
    for my $item_name ( sort { $args->{$a}->{'seq'} <=> $args->{$b}->{'seq'} } (keys %$args) ) {
        my $item_spec = $args->{$item_name}->{'src'};
        $item_spec =~ s/=for(.*)//ms;
        $pod_string .= "=item $item_name\n\n";
        # Get list of variable for this argument
        while ( my ($var_name, $var) = each %{$args->{$item_name}->{var}} ) {
            # Get default for this variable
            for my $default_type ( 'default', 'opt_default' ) {
                my $var_default;
                if (exists $var->{$default_type}) {
                    if (ref($var->{$default_type}) eq 'ARRAY') {
                        $var_default = join(' ', @{$var->{$default_type}});
                    } elsif (ref($var->{$default_type}) eq '') {
                        $var_default = $var->{$default_type};
                    } else {
                        carp 'Getopt::Euclid found an unexpected default value type';
                    }
                } else {
                    $var_default = 'none';
                }
                $item_spec =~ s/$var_name\.$default_type/$var_default/g;
            }
        }
        if ($item_spec =~ m/(\S+(\.(?:opt_)?default))/) {
            my ($reference, $default_type) = ($1, $2);
            _fail( "Invalid reference to field $reference in argument ".
                   "description:\n$item_spec" );
        }
        $pod_string .= $item_spec;
    }
    $pod_string = "=over\n\n".$pod_string."=back\n\n";
    return $pod_string;
}


1;                                 # Magic true value required at end of module


=head1 NAME

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

=head1 VERSION

This document describes Getopt::Euclid version 0.4.5

=head1 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)


=head1 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:

=over

=item 1.

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

=item 2.

extract information from that POD, most especially from 
the C<=head1 REQUIRED ARGUMENTS> and C<=head1 OPTIONS> sections,

=item 3.

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

=item 4.

remove the command-line arguments from C<@ARGV> and parse them, and

=item 5.

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

=back

As a special case, if the module is loaded within some other module
(i.e. from within a C<.pm> file), it still locates and extracts POD
information, but instead of parsing C<@ARGV> immediately, it caches that
information and installs an C<import()> subroutine in the caller module.
This new C<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 L<Module interface> for more details.

=head1 INTERFACE 

=head2 Program interface

You write:

    use Getopt::Euclid;

and your command-line is parsed automagically.

=head2 Module interface

=over

=item import()

You write:

    use Getopt::Euclid;

and your module will then act just like Getopt::Euclid (i.e. you can use
your module I<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 C<import()>
subroutine in your module. If your module already has an C<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

=item process_args()

Alternatively, to parse arguments from a source different from C<@ARGV>, use the
C<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 C<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.

=item process_pods()

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

=back

=head2 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 L<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-C<.pm> file, it searches that file for
the following POD documentation:

=over

=item =head1 NAME

Getopt::Euclid ignores the name specified here. In fact, if you use the
standard C<--help>, C<--usage>, C<--man>, C<--podfile>, or C<--version>
arguments (see L<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 C<$0>).

=item =head1 USAGE

Getopt::Euclid ignores the usage line specified here. If you use the
standard C<--help>, C<--usage>, C<--man> or C<--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.

=item =head1 VERSION

Getopt::Euclid extracts the current version number from this POD section.
To do that it simply takes the first substring that matches
I<< <digit> >>.I<< <digit> >> or I<< <digit> >>_I<< <digit> >>. It also
accepts one or more additional trailing .I<< <digit> >> or _I<< <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 C<BEGIN> block:

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

Euclid stores the version as C<$Getopt::Euclid::SCRIPT_VERSION>.

=item =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 L<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]

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

=item =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 L<Specifying arguments> for details of the specification syntax.

Typically a program will specify both C<REQUIRED ARGUMENTS> and C<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]

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

=item =head1 COPYRIGHT

Getopt::Euclid prints this section whenever the standard C<--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".

=back

=head2 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

=head3 Argument structure

=over 

=item *

Each argument is specified as an C<=item>.

=item *

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

=item *

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

=item *

Any placeholder that is immediately followed by C<...> may be repeated as many
times as desired.

=item *

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

=item *

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

=back

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 C<'data.txt'>.

You could allow the optional C<=> 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 C<-i>, C<-in>, and C<-input> as synonyms for this
argument and would set all three of C<$ARGV{'-i'}>, C<$ARGV{'-in'}>, and
C<$ARGV{'-input'}> to the supplied file name.

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

=head2 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 C<repeatable> option:

    =item file=<filename>

    =for Euclid:
        repeatable

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

=head2 Boolean arguments

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

For a boolean argument, you can also specify variations that are I<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 C<$ARGV{'--print'}>
and C<$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 C<--print> appears in C<@ARGV>, then
C<$ARGV{'--print'}> will be true and C<$ARGV{'--noprint'}> will be false.
On the other hand, if C<--noprint> appears in C<@ARGV>, then
C<$ARGV{'--print'}> will be false and C<$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.

=head2 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
C<%ARGV> becomes a hash reference, with each of the placeholder names as one
key. That is, the above command-line would set both C<$ARGV{'-size'}{'h'}> and
C<$ARGV{'-size'}{'w'}>.

=head2 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 C<$ARGV{'-size'}{'w'}> entry is set to C<undef>. See also
L<Placeholder defaults>.

=head2 Unflagged placeholders

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

    =item <filename>

then the corresponding entry in C<%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 C<%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 I<do not> retain their angle-brackets.

=head2 Repeated placeholders

Any placeholder that is immediately followed by C<...>, 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 L<Placeholder constraints>
and L<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 C<%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>'} };

=head2 Placeholder constraints

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

    =item -size <h>x<w>

    =for Euclid:
        h.type: integer
        w.type: integer

specifies that both the C<< <h> >> and C<< <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 C<< <h> >> has to be given an integer that is greater than zero,
and that C<< <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 C<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)
I<before> the C<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


=head2 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
L<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

=head2 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 C<.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).

=head2 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 C<< <w> >> value is supplied:

    -size 20

then C<$ARGV{'-size'}{'w'}> is set to 80. Likewise, of the C<-size> argument is
omitted entirely, both C<$ARGV{'-size'}{'h'}> and C<$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 C<< --debug >> is not specified, then
C<$ARGV{'--debug'}> is set to 0, the regular default. But if no C<< <log_level> >>
value is supplied:

    --debug
    
then C<$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 L<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

=head2 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.

=head2 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 C<z> argument has also been specified.

On the other hand, if the argument:

    =item -e <cmd>

had been specified, the module I<would> accept:

    -vxe'print time'

as a cuddled version of:

    -v -x -e'print time'

=head2 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 C<':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 C<%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
C<use strict>.

=head2 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:

=over

=item --usage  usage()

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

=item --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
C<help()>, convert it to plaintext, display it and exit. The message is paged
using IO::Pager::Page (or IO::Page) if possible.

=item --man  man()

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

=item --podfile  podfile()

The --podfile argument is provided for authors. It causes the program to take
the POD manual from C<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 C<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 L<Module::Install> to bundle your script, you might be interested in
using L<Module::Install::PodFromEuclid> to include the --podfile step into the
installation process.

=item --version  version()

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

=back

=head2 Minimalist keys

By default, the keys of C<%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 C<%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
C<%ARGV> be simply:

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

You can arrange this by loading the module with the special C<':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 C<--step> and
a C<< <step> >> option). The module throws an exception if this happens.

=head2 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 C<@ARGV> before it is processed (and emptied) by
Getopt::Euclid. Or you may intend to pass your own arguments manually only
using C<process_args()>.

To defer the parsing of arguments, use the specifier C<':defer'>:

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

=head1 DIAGNOSTICS

=head2 Compile-time diagnostics

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

=over

=item 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 I<perl> was able to run the program in the
first place, if it is not readable.

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

You tried to define an C<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 C<import()> for you, supplying your own
C<import()> as well defeats the purpose.

=item Unknown specification: %s

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

=item Unknown type (%s) in specification: %s

=item 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 L<Standard placeholder types> for a list of types that Getopt::Euclid
I<does> recognize.

=item 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

=item 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: '*$@!&'

=item Invalid .opt_default value: %s

Same as previous diagnostic, but for optional defaults.

=item 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

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

Same as previous diagnostic, but for optional defaults.

=item 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


=item 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

=item 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.

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

You attempted to define a C<.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 C<=for Euclid:> that has drifted away from its argument:

    =item -foo <bar>

    =item -verbose

    =for Euclid:
        bar.type: integer

=item 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.

=item Unknown mode ('%s')

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

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

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

=back

=head2 Run-time diagnostics

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

=over 

=item Missing required argument(s): %s

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

=item 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.

=item 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.

=back

=head1 CONFIGURATION AND ENVIRONMENT

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

=head1 DEPENDENCIES

=over 

=item *

version

=item *

Pod::Select

=item *

Pod::PlainText

=item *

File::Basename

=item *

File::Spec::Functions

=item *

List::Util

=item *

Text::Balanced

=item *

IO::Pager::Page (recommended)

=back

=head1 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.

=head1 BUGS AND LIMITATIONS

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

Pull requests to fix bugs / add features welcome!

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

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

The L<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.


=head1 AUTHOR

Damian Conway  C<< <DCONWAY@cpan.org> >>

Florent Angly C<< <florent.angly@gmail.com> >>

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

=head1 LICENCE AND COPYRIGHT

Copyright (c) 2005, Damian Conway C<< <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.


=head1 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.