File Coverage

Char/Arabic.pm

Criterion	Covered	Total	%
statement	49	2194	2.2
branch	11	2162	0.5
condition	1	199	0.5
subroutine	14	40	35.0
pod	0	25	0.0
total	75	4620	1.6

package Char::Arabic;
######################################################################
#
# Char::Arabic - Source code filter to escape Arabic script
#
# http://search.cpan.org/dist/Char-Arabic/
#
# Copyright (c) 2008, 2009, 2010, 2011, 2012, 2013, 2014 INABA Hitoshi
######################################################################

use 5.00503;    # Galapagos Consensus 1998 for primetools
# use 5.008001; # Lancaster Consensus 2013 for toolchains

# 12.3. Delaying use Until Runtime
# in Chapter 12. Packages, Libraries, and Modules
# of ISBN 0-596-00313-7 Perl Cookbook, 2nd Edition.
# (and so on)

BEGIN { eval q{ use vars qw($VERSION) } }
$VERSION = sprintf '%d.%02d', q$Revision: 0.99 $ =~ /(\d+)/oxmsg;

BEGIN {
    if ($^X =~ / jperl /oxmsi) {
        die __FILE__, ": needs perl(not jperl) 5.00503 or later. (\$^X==$^X)";
    }
    if (CORE::ord('A') == 193) {
        die __FILE__, ": is not US-ASCII script (may be EBCDIC or EBCDIK script).";
    }
    if (CORE::ord('A') != 0x41) {
        die __FILE__, ": is not US-ASCII script (must be US-ASCII script).";
    }
}

BEGIN { CORE::require Char::Earabic; }

# instead of Symbol.pm
BEGIN {
    my $genpkg = "Symbol::";
    my $genseq = 0;
    sub gensym () {
        my $name = "GEN" . $genseq++;

        # here, no strict qw(refs); if strict.pm exists

        my $ref = \*{$genpkg . $name};
        delete $$genpkg{$name};
        $ref;
    }
}

# Column: local $@
# in Chapter 9. Osaete okitai Perl no kiso
# of ISBN 10: 4798119172 | ISBN 13: 978-4798119175 MODAN Perl NYUMON
# (and so on)

# use strict; if strict.pm exists
BEGIN {
    if (eval { local $@; CORE::require strict }) {
        strict::->import;
    }
}

# P.714 29.2.39. flock
# in Chapter 29: Functions
# of ISBN 0-596-00027-8 Programming Perl Third Edition.

# P.863 flock
# in Chapter 27: Functions
# of ISBN 978-0-596-00492-7 Programming Perl 4th Edition.

# P.228 Inlining Constant Functions
# in Chapter 6: Subroutines
# of ISBN 0-596-00027-8 Programming Perl Third Edition.

# P.331 Inlining Constant Functions
# in Chapter 7: Subroutines
# of ISBN 978-0-596-00492-7 Programming Perl 4th Edition.

sub LOCK_SH() {1}
sub LOCK_EX() {2}
sub LOCK_UN() {8}
sub LOCK_NB() {4}

sub unimport {}
sub Char::Arabic::escape_script;

# 6.18. Matching Multiple-Byte Characters
# in Chapter 6. Pattern Matching
# of ISBN 978-1-56592-243-3 Perl Perl Cookbook.
# (and so on)

# regexp of character
my $your_char = q{[\x00-\xFF]};
my $qq_char   = qr/\\c[\x40-\x5F]|\\?(?:$your_char)/oxms;
my  $q_char   = qr/$your_char/oxms;

# P.1023 Appendix W.9 Multibyte Anchoring
# of ISBN 1-56592-224-7 CJKV Information Processing

my $anchor = '';

BEGIN { eval q{ use vars qw($nest) } }

# regexp of nested parens in qqXX

# P.340 Matching Nested Constructs with Embedded Code
# in Chapter 7: Perl
# of ISBN 0-596-00289-0 Mastering Regular Expressions, Second edition

my $qq_paren   = qr{(?{local $nest=0}) (?>(?:
                    \\c[\x40-\x5F] |
                    \\ [\x00-\xFF] |
                    [^()] |
                             \(  (?{$nest++}) |
                             \)  (?(?{$nest>0})(?{$nest--})|(?!)))*) (?(?{$nest!=0})(?!))
                 }xms;
my $qq_brace   = qr{(?{local $nest=0}) (?>(?:
                    \\c[\x40-\x5F] |
                    \\ [\x00-\xFF] |
                    [^{}] |
                             \{  (?{$nest++}) |
                             \}  (?(?{$nest>0})(?{$nest--})|(?!)))*) (?(?{$nest!=0})(?!))
                 }xms;
my $qq_bracket = qr{(?{local $nest=0}) (?>(?:
                    \\c[\x40-\x5F] |
                    \\ [\x00-\xFF] |
                    [^[\]] |
                             \[  (?{$nest++}) |
                             \]  (?(?{$nest>0})(?{$nest--})|(?!)))*) (?(?{$nest!=0})(?!))
                 }xms;
my $qq_angle   = qr{(?{local $nest=0}) (?>(?:
                    \\c[\x40-\x5F] |
                    \\ [\x00-\xFF] |
                    [^<>] |
                             \<  (?{$nest++}) |
                             \>  (?(?{$nest>0})(?{$nest--})|(?!)))*) (?(?{$nest!=0})(?!))
                 }xms;
my $qq_scalar  = qr{(?: \{ (?:$qq_brace)*? \} |
                       (?: ::)? (?:
                             [a-zA-Z_][a-zA-Z_0-9]*
                       (?: ::[a-zA-Z_][a-zA-Z_0-9]* )* (?:                                   \[ (?: \$\[ | \$\] | $qq_char )*? \] |           \{ (?:$qq_brace)*? \} )*
                                         (?: (?: -> )? (?: [\$\@\%\&\*]\* | \$\#\* | [\@\%]? \[ (?: \$\[ | \$\] | $qq_char )*? \] | [\@\%\*]? \{ (?:$qq_brace)*? \} ) )*
                   ))
                 }xms;
my $qq_variable = qr{(?: \{ (?:$qq_brace)*? \}                    |
                        (?: ::)? (?:
                              [0-9]+                              |
                              [^a-zA-Z_0-9\[\]] |
                              ^[A-Z]                              |
                              [a-zA-Z_][a-zA-Z_0-9]*
                        (?: ::[a-zA-Z_][a-zA-Z_0-9]* )* (?:                                   \[ (?: \$\[ | \$\] | $qq_char )*? \] |           \{ (?:$qq_brace)*? \} )*
                                          (?: (?: -> )? (?: [\$\@\%\&\*]\* | \$\#\* | [\@\%]? \[ (?: \$\[ | \$\] | $qq_char )*? \] | [\@\%\*]? \{ (?:$qq_brace)*? \} ) )*
                    ))
                  }xms;
my $qq_substr  = qr{(?: Char::Arabic::substr | CORE::substr | substr ) \( $qq_paren \)
                 }xms;

# regexp of nested parens in qXX
my $q_paren    = qr{(?{local $nest=0}) (?>(?:
                    [^()] |
                             \(  (?{$nest++}) |
                             \)  (?(?{$nest>0})(?{$nest--})|(?!)))*) (?(?{$nest!=0})(?!))
                 }xms;
my $q_brace    = qr{(?{local $nest=0}) (?>(?:
                    [^{}] |
                             \{  (?{$nest++}) |
                             \}  (?(?{$nest>0})(?{$nest--})|(?!)))*) (?(?{$nest!=0})(?!))
                 }xms;
my $q_bracket  = qr{(?{local $nest=0}) (?>(?:
                    [^[\]] |
                             \[  (?{$nest++}) |
                             \]  (?(?{$nest>0})(?{$nest--})|(?!)))*) (?(?{$nest!=0})(?!))
                 }xms;
my $q_angle    = qr{(?{local $nest=0}) (?>(?:
                    [^<>] |
                             \<  (?{$nest++}) |
                             \>  (?(?{$nest>0})(?{$nest--})|(?!)))*) (?(?{$nest!=0})(?!))
                 }xms;

my $matched     = '';
my $s_matched   = '';

my $tr_variable   = '';   # variable of tr///
my $sub_variable  = '';   # variable of s///
my $bind_operator = '';   # =~ or !~
BEGIN { eval q{ use vars qw($slash) } }
                          # when 'm//', '/' means regexp match 'm//' and '?' means regexp match '??'
                          # when 'div', '/' means division operator and '?' means conditional operator (condition ? then : else)
my @heredoc = ();         # here document
my @heredoc_delimiter = ();
my $here_script = '';     # here script
my $function_ord;         # ord()   to ord()   or Char::Arabic::ord()
my $function_ord_;        # ord     to ord     or Char::Arabic::ord_
my $function_reverse;     # reverse to reverse or Char::Arabic::reverse
my $function_getc;        # getc    to getc    or Char::Arabic::getc

my $ignore_modules = join('|', qw(
    utf8
    bytes
    charnames
    I18N::Japanese
    I18N::Collate
    I18N::JExt
    File::DosGlob
    Wild
    Wildcard
    Japanese
));

# when this script is main program
if ($0 eq __FILE__) {

    # show usage
    unless (@ARGV) {
        die <
$0: usage

perl $0 Arabic_script.pl > Escaped_script.pl.e
END
    }

    print Char::Arabic::escape_script($ARGV[0]);
    exit 0;
}

my($package,$filename,$line,$subroutine,$hasargs,$wantarray,$evaltext,$is_require,$hints,$bitmask) = caller 0;

# called any package not main
if ($package ne 'main') {
    die <
@{[__FILE__]}: escape by manually command '$^X @{[__FILE__]} "$filename" > "@{[__PACKAGE__]}::$filename"'
and rewrite "use $package;" to "use @{[__PACKAGE__]}::$package;" of script "$0".
END
}

# P.302 Module Privacy and the Exporter
# in Chapter 11: Modules
# of ISBN 0-596-00027-8 Programming Perl Third Edition.
#
# A module can do anything it jolly well pleases when it's used, since use just
# calls the ordinary import method for the module, and you can define that
# method to do anything you like.

# P.406 Module Privacy and the Exporter
# in Chapter 11: Modules
# of ISBN 978-0-596-00492-7 Programming Perl 4th Edition.
#
# A module can do anything it jolly well pleases when it's used, since use just
# calls the ordinary import method for the module, and you can define that
# method to do anything you like.

sub import {

    if (-e("$filename.e")) {
        if (exists $ENV{'SJIS_DEBUG'}) {
            unlink "$filename.e";
        }
        elsif (-z("$filename.e")) {
            unlink "$filename.e";
        }
        else {

            #----------------------------------------------------
            #  older >
            #  newer >>>>>
            #----------------------------------------------------
            # Filter >
            # Source >>>>>
            # Escape >>>   needs re-escape (Source was changed)
            #
            # Filter >>>
            # Source >>>>>
            # Escape >     needs re-escape (Source was changed)
            #
            # Filter >>>>>
            # Source >>>
            # Escape >     needs re-escape (Source was changed)
            #
            # Filter >>>>>
            # Source >
            # Escape >>>   needs re-escape (Filter was changed)
            #
            # Filter >
            # Source >>>
            # Escape >>>>> executable without re-escape
            #
            # Filter >>>
            # Source >
            # Escape >>>>> executable without re-escape
            #----------------------------------------------------

            my $mtime_filter = (stat(__FILE__     ))[9];
            my $mtime_source = (stat($filename    ))[9];
            my $mtime_escape = (stat("$filename.e"))[9];
            if (($mtime_escape < $mtime_source) or ($mtime_escape < $mtime_filter)) {
                unlink "$filename.e";
            }
        }
    }

    if (not -e("$filename.e")) {
        my $fh = gensym();

        if (eval q{ use Fcntl qw(O_WRONLY O_APPEND O_CREAT); 1 } and CORE::sysopen($fh,"$filename.e",&O_WRONLY|&O_APPEND|&O_CREAT)) {
        }
        else {
            Char::Earabic::_open_a($fh, "$filename.e") or die __FILE__, ": Can't write open file: $filename.e";
        }

        if (0) {
        }
        elsif (exists $ENV{'SJIS_NONBLOCK'}) {

            # P.419 File Locking
            # in Chapter 16: Interprocess Communication
            # of ISBN 0-596-00027-8 Programming Perl Third Edition.

            # P.524 File Locking
            # in Chapter 15: Interprocess Communication
            # of ISBN 978-0-596-00492-7 Programming Perl 4th Edition.

            # P.571 Handling Race Conditions
            # in Chapter 23: Security
            # of ISBN 0-596-00027-8 Programming Perl Third Edition.

            # P.663 Handling Race Conditions
            # in Chapter 20: Security
            # of ISBN 978-0-596-00492-7 Programming Perl 4th Edition.

            # (and so on)

            eval q{
                unless (flock($fh, LOCK_EX | LOCK_NB)) {
                    warn __FILE__, ": Can't immediately write-lock the file: $filename.e";
                    exit;
                }
            };
        }
        else {
            eval q{ flock($fh, LOCK_EX) };
        }

        eval q{ truncate($fh, 0) };
        seek($fh, 0, 0) or die __FILE__, ": Can't seek file: $filename.e";

        my $e_script = Char::Arabic::escape_script($filename);
        print {$fh} $e_script;

        my $mode = (stat($filename))[2] & 0777;
        chmod $mode, "$filename.e";

        close($fh) or die __FILE__, ": Can't close file: $filename.e";
    }

    my $fh = gensym();
    Char::Earabic::_open_r($fh, "$filename.e") or die __FILE__, ": Can't read open file: $filename.e";

    if (0) {
    }
    elsif (exists $ENV{'SJIS_NONBLOCK'}) {
        eval q{
            unless (flock($fh, LOCK_SH | LOCK_NB)) {
                warn __FILE__, ": Can't immediately read-lock the file: $filename.e";
                exit;
            }
        };
    }
    else {
        eval q{ flock($fh, LOCK_SH) };
    }

    my @switch = ();
    if ($^W) {
        push @switch, '-w';
    }

    # P.707 29.2.33. exec
    # in Chapter 29: Functions
    # of ISBN 0-596-00027-8 Programming Perl Third Edition.
    #
    # If there is more than one argument in LIST, or if LIST is an array with more
    # than one value, the system shell will never be used. This also bypasses any
    # shell processing of the command. The presence or absence of metacharacters in
    # the arguments doesn't affect this list-triggered behavior, which makes it the
    # preferred from in security-conscious programs that do not with to expose
    # themselves to potential shell escapes.
    # Environment variable PERL5SHELL(Microsoft ports only) will never be used, too.

    # P.855 exec
    # in Chapter 27: Functions
    # of ISBN 978-0-596-00492-7 Programming Perl 4th Edition.
    #
    # If there is more than one argument in LIST, or if LIST is an array with more
    # than one value, the system shell will never be used. This also bypasses any
    # shell processing of the command. The presence or absence of metacharacters in
    # the arguments doesn't affect this list-triggered behavior, which makes it the
    # preferred from in security-conscious programs that do not wish to expose
    # themselves to injection attacks via shell escapes.
    # Environment variable PERL5SHELL(Microsoft ports only) will never be used, too.

    # P.489 #! and Quoting on Non-Unix Systems
    # in Chapter 19: The Command-Line Interface
    # of ISBN 0-596-00027-8 Programming Perl Third Edition.

    # P.578 #! and Quoting on Non-Unix Systems
    # in Chapter 17: The Command-Line Interface
    # of ISBN 978-0-596-00492-7 Programming Perl 4th Edition.

    # DOS-like system
    if ($^O =~ /\A (?: MSWin32 | NetWare | symbian | dos ) \z/oxms) {
        exit Char::Earabic::_systemx(
            _escapeshellcmd_MSWin32($^X),

        # -I switch can not treat space included path
        #   (map { '-I' . _escapeshellcmd_MSWin32($_) } @INC),
            (map { '-I' .                         $_  } @INC),

            @switch,
            '--',
            map { _escapeshellcmd_MSWin32($_) } "$filename.e", @ARGV
        );
    }

    # UNIX-like system
    else {
        exit Char::Earabic::_systemx(
            _escapeshellcmd($^X),
            (map { '-I' . _escapeshellcmd($_) } @INC),
            @switch,
            '--',
            map { _escapeshellcmd($_) } "$filename.e", @ARGV
        );
    }
}

# escape shell command line on DOS-like system
sub _escapeshellcmd_MSWin32 {
    my($word) = @_;
    if ($word =~ / [ ] /oxms) {

        # both DOS-like and UNIX-like shell quote
        return qq{"$word"};
    }
    else {
        return $word;
    }
}

# escape shell command line on UNIX-like system
sub _escapeshellcmd {
    my($word) = @_;
    return $word;
}

# P.619 Source Filters
# in Chapter 24: Common Practices
# of ISBN 0-596-00027-8 Programming Perl Third Edition.

# P.718 Source Filters
# in Chapter 21: Common Practices
# of ISBN 978-0-596-00492-7 Programming Perl 4th Edition.

# escape Arabic script
sub Char::Arabic::escape_script {
    my($script) = @_;
    my $e_script = '';

    # read Arabic script
    my $fh = gensym();
    Char::Earabic::_open_r($fh, $script) or die __FILE__, ": Can't open file: $script";
    local $/ = undef; # slurp mode
    $_ = <$fh>;
    close($fh) or die __FILE__, ": Can't close file: $script";

    if (/^ use Char::Earabic(?:\s+[0-9\.]*)?\s*; $/oxms) {
        return $_;
    }
    else {

        # #! shebang line
        if (s/\A(#!.+?\n)//oms) {
            my $head = $1;
            $head =~ s/\bjperl\b/perl/gi;
            $e_script .= $head;
        }

        # DOS-like system header
        if (s/\A(\@rem\s*=\s*'.*?'\s*;\s*\n)//oms) {
            my $head = $1;
            $head =~ s/\bjperl\b/perl/gi;
            $e_script .= $head;
        }

        # P.618 Generating Perl in Other Languages
        # in Chapter 24: Common Practices
        # of ISBN 0-596-00027-8 Programming Perl Third Edition.

        # P.717 Generating Perl in Other Languages
        # in Chapter 21: Common Practices
        # of ISBN 978-0-596-00492-7 Programming Perl 4th Edition.

        if (s/(.*^#\s*line\s+\d+(?:\s+"(?:$q_char)+?")?\s*\n)//oms) {
            my $head = $1;
            $head =~ s/\bjperl\b/perl/gi;
            $e_script .= $head;
        }

        # P.210 5.10.3.3. Match-time code evaluation
        # in Chapter 5: Pattern Matching
        # of ISBN 0-596-00027-8 Programming Perl Third Edition.

        # P.255 Match-time code evaluation
        # in Chapter 5: Pattern Matching
        # of ISBN 978-0-596-00492-7 Programming Perl 4th Edition.

        $e_script .= sprintf("use Char::Earabic %s.0;\n", $Char::Arabic::VERSION); # require run-time routines version

        # use Char::Arabic version qw(ord reverse getc);
        $function_ord     = 'ord';
        $function_ord_    = 'ord';
        $function_reverse = 'reverse';
        $function_getc    = 'getc';
        if (s/^ \s* use \s+ Char::Arabic \s* ([^;]*) ; \s* \n? $//oxms) {

            # require version
            my $list = $1;
            if ($list =~ s/\A ([0-9]+\.[0-9]+) \.0 \s* //oxms) {
                my $version = $1;
                if ($version ne $Char::Arabic::VERSION) {
                    my @file = grep -e, map {qq{$_/Char/Arabic.pm}} @INC;
                    my %file = map { $_ => 1 } @file;
                    if (scalar(keys %file) >= 2) {
                        my $file = join "\n", sort keys %file;
                        warn <
****************************************************
                   C A U T I O N

              CONFLICT Char/Arabic.pm FILE

$file
****************************************************

END
                    }
                    die "Script $0 expects Char/Arabic.pm $version, but @{[__FILE__]} is version $Char::Arabic::VERSION\n";
                }
                $e_script .= qq{die "Script \$0 expects Char/Earabic.pm $version, but \\\$Char::Earabic::VERSION is \$Char::Earabic::VERSION" if \$Char::Earabic::VERSION ne '$version';\n};
            }
            elsif ($list =~ s/\A ([0-9]+(?:\.[0-9]*)) \s* //oxms) {
                my $version = $1;
                if ($version > $Char::Arabic::VERSION) {
                    die "Script $0 required Char/Arabic.pm $version, but @{[__FILE__]} is only version $Char::Arabic::VERSION\n";
                }
            }

            # demand ord, reverse, and getc
            if ($list !~ /\A \s* \z/oxms) {
                local $@;
                my @list = eval $list;
                for (@list) {
                    $function_ord     = 'Char::Arabic::ord'     if /\A ord \z/oxms;
                    $function_ord_    = 'Char::Arabic::ord_'    if /\A ord \z/oxms;
                    $function_reverse = 'Char::Arabic::reverse' if /\A reverse \z/oxms;
                    $function_getc    = 'Char::Arabic::getc'    if /\A getc \z/oxms;
                }
            }
        }
    }

    $slash = 'm//';

    # P.359 The Study Function
    # in Chapter 7: Perl
    # of ISBN 0-596-00289-0 Mastering Regular Expressions, Second edition

    study $_; # Yes, I studied study yesterday.

    # while all script

    # 6.14. Matching from Where the Last Pattern Left Off
    # in Chapter 6. Pattern Matching
    # of ISBN 0-596-00313-7 Perl Cookbook, 2nd Edition.
    # (and so on)

    # one member of Tag-team
    #
    # P.128 Start of match (or end of previous match): \G
    # P.130 Advanced Use of \G with Perl
    # in Chapter 3: Overview of Regular Expression Features and Flavors
    # P.255 Use leading anchors
    # P.256 Expose ^ and \G at the front expressions
    # in Chapter 6: Crafting an Efficient Expression
    # P.315 "Tag-team" matching with /gc
    # in Chapter 7: Perl
    # of ISBN 0-596-00289-0 Mastering Regular Expressions, Second edition

    while (not /\G \z/oxgc) { # member
        $e_script .= escape();
    }

    return $e_script;
}

# escape Arabic part of script
sub escape {

# \n output here document

    # another member of Tag-team
    #
    # P.315 "Tag-team" matching with /gc
    # in Chapter 7: Perl
    # of ISBN 0-596-00289-0 Mastering Regular Expressions, Second edition

    if (/\G ( \n ) /oxgc) { # another member (and so on)
        my $heredoc = '';
        if (scalar(@heredoc_delimiter) >= 1) {
            $slash = 'm//';

            $heredoc = join '', @heredoc;
            @heredoc = ();

            # skip here document
            for my $heredoc_delimiter (@heredoc_delimiter) {
                /\G .*? \n $heredoc_delimiter \n/xmsgc;
            }
            @heredoc_delimiter = ();

            $here_script = '';
        }
        return "\n" . $heredoc;
    }

# ignore space, comment
    elsif (/\G (\s+|\#.*) /oxgc) { return $1; }

# if (, elsif (, unless (, while (, until (, given (, and when (

    # given, when

    # P.225 The given Statement
    # in Chapter 15: Smart Matching and given-when
    # of ISBN 978-0-596-52010-6 Learning Perl, Fifth Edition

    # P.133 The given Statement
    # in Chapter 4: Statements and Declarations
    # of ISBN 978-0-596-00492-7 Programming Perl 4th Edition.

    elsif (/\G ( (?: if | elsif | unless | while | until | given | when ) \s* \( ) /oxgc) {
        $slash = 'm//';
        return $1;
    }

# scalar variable ($scalar = ...) =~ tr///;
# scalar variable ($scalar = ...) =~ s///;

    # state

    # P.68 Persistent, Private Variables
    # in Chapter 4: Subroutines
    # of ISBN 978-0-596-52010-6 Learning Perl, Fifth Edition

    # P.160 Persistent Lexically Scoped Variables: state
    # in Chapter 4: Statements and Declarations
    # of ISBN 978-0-596-00492-7 Programming Perl 4th Edition.

    # (and so on)

    elsif (/\G ( \( \s* (?: local \b | my \b | our \b | state \b )? \s* \$ $qq_scalar ) /oxgc) {
        my $e_string = e_string($1);

        if (/\G ( \s* = $qq_paren \) ) ( \s* (?: =~ | !~ ) \s* ) (?= (?: tr | y ) \b ) /oxgc) {
            $tr_variable = $e_string . e_string($1);
            $bind_operator = $2;
            $slash = 'm//';
            return '';
        }
        elsif (/\G ( \s* = $qq_paren \) ) ( \s* (?: =~ | !~ ) \s* ) (?= s \b ) /oxgc) {
            $sub_variable = $e_string . e_string($1);
            $bind_operator = $2;
            $slash = 'm//';
            return '';
        }
        else {
            $slash = 'div';
            return $e_string;
        }
    }

# $`, ${`}, $PREMATCH, ${PREMATCH}, ${^PREMATCH} --> Char::Earabic::PREMATCH()
    elsif (/\G ( \$` | \$\{`\} | \$ \s* PREMATCH  \b | \$ \s* \{ \s* PREMATCH  \s* \} | \$ \s* \{\^PREMATCH\}  ) /oxmsgc) {
        $slash = 'div';
        return q{Char::Earabic::PREMATCH()};
    }

# $&, ${&}, $MATCH, ${MATCH}, ${^MATCH} --> Char::Earabic::MATCH()
    elsif (/\G ( \$& | \$\{&\} | \$ \s* MATCH     \b | \$ \s* \{ \s* MATCH     \s* \} | \$ \s* \{\^MATCH\}     ) /oxmsgc) {
        $slash = 'div';
        return q{Char::Earabic::MATCH()};
    }

# $', ${'} --> $', ${'}
    elsif (/\G ( \$' | \$\{'\}                                                                                 ) /oxmsgc) {
        $slash = 'div';
        return $1;
    }

# $POSTMATCH, ${POSTMATCH}, ${^POSTMATCH} --> Char::Earabic::POSTMATCH()
    elsif (/\G (                 \$ \s* POSTMATCH \b | \$ \s* \{ \s* POSTMATCH \s* \} | \$ \s* \{\^POSTMATCH\} ) /oxmsgc) {
        $slash = 'div';
        return q{Char::Earabic::POSTMATCH()};
    }

# scalar variable $scalar =~ tr///;
# scalar variable $scalar =~ s///;
# substr() =~ tr///;
# substr() =~ s///;
    elsif (/\G ( \$ $qq_scalar | $qq_substr ) /oxgc) {
        my $scalar = e_string($1);

        if (/\G ( \s* (?: =~ | !~ ) \s* ) (?= (?: tr | y ) \b ) /oxgc) {
            $tr_variable = $scalar;
            $bind_operator = $1;
            $slash = 'm//';
            return '';
        }
        elsif (/\G ( \s* (?: =~ | !~ ) \s* ) (?= s \b ) /oxgc) {
            $sub_variable = $scalar;
            $bind_operator = $1;
            $slash = 'm//';
            return '';
        }
        else {
            $slash = 'div';
            return $scalar;
        }
    }

    # end of statement
    elsif (/\G ( [,;] ) /oxgc) {
        $slash = 'm//';

        # clear tr/// variable
        $tr_variable  = '';

        # clear s/// variable
        $sub_variable  = '';

        $bind_operator = '';

        return $1;
    }

# bareword
    elsif (/\G ( \{ \s* (?: tr | index | rindex | reverse ) \s* \} ) /oxmsgc) {
        return $1;
    }

# $0 --> $0
    elsif (/\G ( \$ 0 ) /oxmsgc) {
        $slash = 'div';
        return $1;
    }
    elsif (/\G ( \$ \{ \s* 0 \s* \} ) /oxmsgc) {
        $slash = 'div';
        return $1;
    }

# $$ --> $$
    elsif (/\G ( \$ \$ ) (?![\w\{]) /oxmsgc) {
        $slash = 'div';
        return $1;
    }

# $1, $2, $3 --> $2, $3, $4 after s/// with multibyte anchoring
# $1, $2, $3 --> $1, $2, $3 otherwise
    elsif (/\G \$ ([1-9][0-9]*) /oxmsgc) {
        $slash = 'div';
        return e_capture($1);
    }
    elsif (/\G \$ \{ \s* ([1-9][0-9]*) \s* \} /oxmsgc) {
        $slash = 'div';
        return e_capture($1);
    }

# $$foo[ ... ] --> $ $foo->[ ... ]
    elsif (/\G \$ ( \$ [A-Za-z_][A-Za-z0-9_]*(?: ::[A-Za-z_][A-Za-z0-9_]*)* ) ( \[ .+? \] ) /oxmsgc) {
        $slash = 'div';
        return e_capture($1.'->'.$2);
    }

# $$foo{ ... } --> $ $foo->{ ... }
    elsif (/\G \$ ( \$ [A-Za-z_][A-Za-z0-9_]*(?: ::[A-Za-z_][A-Za-z0-9_]*)* ) ( \{ .+? \} ) /oxmsgc) {
        $slash = 'div';
        return e_capture($1.'->'.$2);
    }

# $$foo
    elsif (/\G \$ ( \$ [A-Za-z_][A-Za-z0-9_]*(?: ::[A-Za-z_][A-Za-z0-9_]*)* ) /oxmsgc) {
        $slash = 'div';
        return e_capture($1);
    }

# ${ foo }
    elsif (/\G \$ \s* \{ ( \s* [A-Za-z_][A-Za-z0-9_]*(?: ::[A-Za-z_][A-Za-z0-9_]*)* \s* ) \} /oxmsgc) {
        $slash = 'div';
        return '${' . $1 . '}';
    }

# ${ ... }
    elsif (/\G \$ \s* \{ \s* ( $qq_brace ) \s* \} /oxmsgc) {
        $slash = 'div';
        return e_capture($1);
    }

# variable or function
    #                  $ @ % & *     $ #
    elsif (/\G ( (?: [\$\@\%\&\*] | \$\# | -> | \b sub \b) \s* (?: split | chop | index | rindex | lc | uc | fc | chr | ord | reverse | getc | tr | y | q | qq | qx | qw | m | s | qr | glob | lstat | opendir | stat | unlink | chdir ) ) \b /oxmsgc) {
        $slash = 'div';
        return $1;
    }
    #                $ $ $ $ $ $ $ $ $ $ $ $ $ $
    #                $ @ # \ ' " / ? ( ) [ ] < >
    elsif (/\G ( \$[\$\@\#\\\'\"\/\?\(\)\[\]\<\>] ) /oxmsgc) {
        $slash = 'div';
        return $1;
    }

# while ()
    elsif (/\G \b (while \s* \( \s* <[\$]?[A-Za-z_][A-Za-z_0-9]*> \s* \)) \b /oxgc) {
        return $1;
    }

# while () --- glob

    # avoid "Error: Runtime exception" of perl version 5.005_03

    elsif (/\G \b while \s* \( \s* < ((?:[^>\0\a\e\f\n\r\t])+?) > \s* \) \b /oxgc) {
        return 'while ($_ = Char::Earabic::glob("' . $1 . '"))';
    }

# while (glob)
    elsif (/\G \b while \s* \( \s* glob \s* \) /oxgc) {
        return 'while ($_ = Char::Earabic::glob_)';
    }

# while (glob(WILDCARD))
    elsif (/\G \b while \s* \( \s* glob \b /oxgc) {
        return 'while ($_ = Char::Earabic::glob';
    }

# doit if, doit unless, doit while, doit until, doit for, doit when
    elsif (/\G \b ( if | unless | while | until | for | when ) \b /oxgc) { $slash = 'm//'; return $1; }

# subroutines of package Char::Earabic
    elsif (/\G \b (CORE:: | ->[ ]* (?: atan2 | [a-z]{2,})) \b  /oxgc) { $slash = 'm//'; return $1;                    }
    elsif (/\G \b bytes::substr \b (?! \s* => )                /oxgc) { $slash = 'm//'; return 'substr';              }
    elsif (/\G \b chop \b          (?! \s* => )                /oxgc) { $slash = 'm//'; return 'Char::Earabic::chop';         }
    elsif (/\G \b bytes::index \b  (?! \s* => )                /oxgc) { $slash = 'm//'; return 'index';               }
    elsif (/\G \b Char::Arabic::index \b   (?! \s* => )                /oxgc) { $slash = 'm//'; return 'Char::Arabic::index';         }
    elsif (/\G \b index \b         (?! \s* => )                /oxgc) { $slash = 'm//'; return 'Char::Earabic::index';        }
    elsif (/\G \b bytes::rindex \b (?! \s* => )                /oxgc) { $slash = 'm//'; return 'rindex';              }
    elsif (/\G \b Char::Arabic::rindex \b  (?! \s* => )                /oxgc) { $slash = 'm//'; return 'Char::Arabic::rindex';        }
    elsif (/\G \b rindex \b        (?! \s* => )                /oxgc) { $slash = 'm//'; return 'Char::Earabic::rindex';       }
    elsif (/\G \b lc      (?= \s+[A-Za-z_]|\s*['"`\$\@\&\*\(]) /oxgc) { $slash = 'm//'; return 'Char::Earabic::lc';           }
    elsif (/\G \b lcfirst (?= \s+[A-Za-z_]|\s*['"`\$\@\&\*\(]) /oxgc) { $slash = 'm//'; return 'Char::Earabic::lcfirst';      }
    elsif (/\G \b uc      (?= \s+[A-Za-z_]|\s*['"`\$\@\&\*\(]) /oxgc) { $slash = 'm//'; return 'Char::Earabic::uc';           }
    elsif (/\G \b ucfirst (?= \s+[A-Za-z_]|\s*['"`\$\@\&\*\(]) /oxgc) { $slash = 'm//'; return 'Char::Earabic::ucfirst';      }
    elsif (/\G \b fc      (?= \s+[A-Za-z_]|\s*['"`\$\@\&\*\(]) /oxgc) { $slash = 'm//'; return 'Char::Earabic::fc';           }

    # "-s '' ..." means file test "-s 'filename' ..." (not means "- s/// ...")
    elsif (/\G -s                               \s+    \s* (\") ((?:$qq_char)+?)               (\") /oxgc) { $slash = 'm//'; return '-s ' . e_qq('',  $1,$3,$2); }
    elsif (/\G -s                               \s+ qq \s* (\#) ((?:$qq_char)+?)               (\#) /oxgc) { $slash = 'm//'; return '-s ' . e_qq('qq',$1,$3,$2); }
    elsif (/\G -s                               \s+ qq \s* (\() ((?:$qq_paren)+?)              (\)) /oxgc) { $slash = 'm//'; return '-s ' . e_qq('qq',$1,$3,$2); }
    elsif (/\G -s                               \s+ qq \s* (\{) ((?:$qq_brace)+?)              (\}) /oxgc) { $slash = 'm//'; return '-s ' . e_qq('qq',$1,$3,$2); }
    elsif (/\G -s                               \s+ qq \s* (\[) ((?:$qq_bracket)+?)            (\]) /oxgc) { $slash = 'm//'; return '-s ' . e_qq('qq',$1,$3,$2); }
    elsif (/\G -s                               \s+ qq \s* (\<) ((?:$qq_angle)+?)              (\>) /oxgc) { $slash = 'm//'; return '-s ' . e_qq('qq',$1,$3,$2); }
    elsif (/\G -s                               \s+ qq \s* (\S) ((?:$qq_char)+?)               (\1) /oxgc) { $slash = 'm//'; return '-s ' . e_qq('qq',$1,$3,$2); }

    elsif (/\G -s                               \s+    \s* (\') ((?:\\\'|\\\\|$q_char)+?)      (\') /oxgc) { $slash = 'm//'; return '-s ' . e_q ('',  $1,$3,$2); }
    elsif (/\G -s                               \s+ q  \s* (\#) ((?:\\\#|\\\\|$q_char)+?)      (\#) /oxgc) { $slash = 'm//'; return '-s ' . e_q ('q', $1,$3,$2); }
    elsif (/\G -s                               \s+ q  \s* (\() ((?:\\\)|\\\\|$q_paren)+?)     (\)) /oxgc) { $slash = 'm//'; return '-s ' . e_q ('q', $1,$3,$2); }
    elsif (/\G -s                               \s+ q  \s* (\{) ((?:\\\}|\\\\|$q_brace)+?)     (\}) /oxgc) { $slash = 'm//'; return '-s ' . e_q ('q', $1,$3,$2); }
    elsif (/\G -s                               \s+ q  \s* (\[) ((?:\\\]|\\\\|$q_bracket)+?)   (\]) /oxgc) { $slash = 'm//'; return '-s ' . e_q ('q', $1,$3,$2); }
    elsif (/\G -s                               \s+ q  \s* (\<) ((?:\\\>|\\\\|$q_angle)+?)     (\>) /oxgc) { $slash = 'm//'; return '-s ' . e_q ('q', $1,$3,$2); }
    elsif (/\G -s                               \s+ q  \s* (\S) ((?:\\\1|\\\\|$q_char)+?)      (\1) /oxgc) { $slash = 'm//'; return '-s ' . e_q ('q', $1,$3,$2); }

    elsif (/\G -s                               \s* (\$ \w+(?: ::\w+)* (?: (?: ->)? (?: [\$\@\%\&\*]\* | \$\#\* | \( (?:$qq_paren)*? \) | [\@\%\*]? \{ (?:$qq_brace)+? \} | [\@\%]? \[ (?:$qq_bracket)+? \] ) )*) /oxgc)
                                                                                                           { $slash = 'm//'; return "-s $1";   }
    elsif (/\G -s                               \s* \( ((?:$qq_paren)*?) \)                         /oxgc) { $slash = 'm//'; return "-s ($1)"; }
    elsif (/\G -s                               (?= \s+ [a-z]+)                                     /oxgc) { $slash = 'm//'; return '-s';      }
    elsif (/\G -s                               \s+ (\w+)                                           /oxgc) { $slash = 'm//'; return "-s $1";   }

    elsif (/\G \b bytes::length (?= \s+[A-Za-z_]|\s*['"`\$\@\&\*\(]) /oxgc) { $slash = 'm//'; return 'length';                   }
    elsif (/\G \b bytes::chr    (?= \s+[A-Za-z_]|\s*['"`\$\@\&\*\(]) /oxgc) { $slash = 'm//'; return 'chr';                      }
    elsif (/\G \b chr           (?= \s+[A-Za-z_]|\s*['"`\$\@\&\*\(]) /oxgc) { $slash = 'm//'; return 'Char::Earabic::chr';               }
    elsif (/\G \b bytes::ord    (?= \s+[A-Za-z_]|\s*['"`\$\@\&\*\(]) /oxgc) { $slash = 'div'; return 'ord';                      }
    elsif (/\G \b ord           (?= \s+[A-Za-z_]|\s*['"`\$\@\&\*\(]) /oxgc) { $slash = 'div'; return $function_ord;              }
    elsif (/\G \b glob          (?= \s+[A-Za-z_]|\s*['"`\$\@\&\*\(]) /oxgc) { $slash = 'm//'; return 'Char::Earabic::glob';              }
    elsif (/\G \b lc \b            (?! \s* => )                      /oxgc) { $slash = 'm//'; return 'Char::Earabic::lc_';               }
    elsif (/\G \b lcfirst \b       (?! \s* => )                      /oxgc) { $slash = 'm//'; return 'Char::Earabic::lcfirst_';          }
    elsif (/\G \b uc \b            (?! \s* => )                      /oxgc) { $slash = 'm//'; return 'Char::Earabic::uc_';               }
    elsif (/\G \b ucfirst \b       (?! \s* => )                      /oxgc) { $slash = 'm//'; return 'Char::Earabic::ucfirst_';          }
    elsif (/\G \b fc \b            (?! \s* => )                      /oxgc) { $slash = 'm//'; return 'Char::Earabic::fc_';               }
    elsif (/\G    -s \b            (?! \s* => )                      /oxgc) { $slash = 'm//'; return '-s ';                      }

    elsif (/\G \b bytes::length \b (?! \s* => )                      /oxgc) { $slash = 'm//'; return 'length';                   }
    elsif (/\G \b bytes::chr \b    (?! \s* => )                      /oxgc) { $slash = 'm//'; return 'chr';                      }
    elsif (/\G \b chr \b           (?! \s* => )                      /oxgc) { $slash = 'm//'; return 'Char::Earabic::chr_';              }
    elsif (/\G \b bytes::ord \b    (?! \s* => )                      /oxgc) { $slash = 'div'; return 'ord';                      }
    elsif (/\G \b ord \b           (?! \s* => )                      /oxgc) { $slash = 'div'; return $function_ord_;             }
    elsif (/\G \b glob \b          (?! \s* => )                      /oxgc) { $slash = 'm//'; return 'Char::Earabic::glob_';             }
    elsif (/\G \b reverse \b       (?! \s* => )                      /oxgc) { $slash = 'm//'; return $function_reverse;          }
    elsif (/\G \b getc \b          (?! \s* => )                      /oxgc) { $slash = 'm//'; return $function_getc;             }

# split
    elsif (/\G \b (split) \b (?! \s* => ) /oxgc) {
        $slash = 'm//';

        my $e = '';
        while (/\G ( \s+ | \( | \#.* ) /oxgc) {
            $e .= $1;
        }

# end of split
        if    (/\G (?= [,;\)\}\]] )          /oxgc) { return 'Char::Earabic::split' . $e;                 }

# split scalar value
        elsif (/\G ( [\$\@\&\*] $qq_scalar ) /oxgc) { return 'Char::Earabic::split' . $e . e_string($1);  }

# split literal space
        elsif (/\G \b qq       (\#) [ ] (\#) /oxgc) { return 'Char::Earabic::split' . $e . qq  {qq$1 $2}; }
        elsif (/\G \b qq (\s*) (\() [ ] (\)) /oxgc) { return 'Char::Earabic::split' . $e . qq{$1qq$2 $3}; }
        elsif (/\G \b qq (\s*) (\{) [ ] (\}) /oxgc) { return 'Char::Earabic::split' . $e . qq{$1qq$2 $3}; }
        elsif (/\G \b qq (\s*) (\[) [ ] (\]) /oxgc) { return 'Char::Earabic::split' . $e . qq{$1qq$2 $3}; }
        elsif (/\G \b qq (\s*) (\<) [ ] (\>) /oxgc) { return 'Char::Earabic::split' . $e . qq{$1qq$2 $3}; }
        elsif (/\G \b qq (\s*) (\S) [ ] (\2) /oxgc) { return 'Char::Earabic::split' . $e . qq{$1qq$2 $3}; }
        elsif (/\G \b q        (\#) [ ] (\#) /oxgc) { return 'Char::Earabic::split' . $e . qq   {q$1 $2}; }
        elsif (/\G \b q  (\s*) (\() [ ] (\)) /oxgc) { return 'Char::Earabic::split' . $e . qq {$1q$2 $3}; }
        elsif (/\G \b q  (\s*) (\{) [ ] (\}) /oxgc) { return 'Char::Earabic::split' . $e . qq {$1q$2 $3}; }
        elsif (/\G \b q  (\s*) (\[) [ ] (\]) /oxgc) { return 'Char::Earabic::split' . $e . qq {$1q$2 $3}; }
        elsif (/\G \b q  (\s*) (\<) [ ] (\>) /oxgc) { return 'Char::Earabic::split' . $e . qq {$1q$2 $3}; }
        elsif (/\G \b q  (\s*) (\S) [ ] (\2) /oxgc) { return 'Char::Earabic::split' . $e . qq {$1q$2 $3}; }
        elsif (/\G                ' [ ] '    /oxgc) { return 'Char::Earabic::split' . $e . qq     {' '};  }
        elsif (/\G                " [ ] "    /oxgc) { return 'Char::Earabic::split' . $e . qq     {" "};  }

# split qq//
        elsif (/\G \b (qq) \b /oxgc) {
            if (/\G (\#) ((?:$qq_char)*?) (\#) /oxgc)                        { return e_split($e.'qr',$1,$3,$2,'');   } # qq# #  --> qr # #
            else {
                while (not /\G \z/oxgc) {
                    if    (/\G (\s+|\#.*)                             /oxgc) { $e .= $1; }
                    elsif (/\G (\()          ((?:$qq_paren)*?)   (\)) /oxgc) { return e_split($e.'qr',$1,$3,$2,'');   } # qq ( ) --> qr ( )
                    elsif (/\G (\{)          ((?:$qq_brace)*?)   (\}) /oxgc) { return e_split($e.'qr',$1,$3,$2,'');   } # qq { } --> qr { }
                    elsif (/\G (\[)          ((?:$qq_bracket)*?) (\]) /oxgc) { return e_split($e.'qr',$1,$3,$2,'');   } # qq [ ] --> qr [ ]
                    elsif (/\G (\<)          ((?:$qq_angle)*?)   (\>) /oxgc) { return e_split($e.'qr',$1,$3,$2,'');   } # qq < > --> qr < >
                    elsif (/\G ([*\-:?\\^|]) ((?:$qq_char)*?)    (\1) /oxgc) { return e_split($e.'qr','{','}',$2,''); } # qq | | --> qr { }
                    elsif (/\G (\S)          ((?:$qq_char)*?)    (\1) /oxgc) { return e_split($e.'qr',$1,$3,$2,'');   } # qq * * --> qr * *
                }
                die __FILE__, ": Can't find string terminator anywhere before EOF";
            }
        }

# split qr//
        elsif (/\G \b (qr) \b /oxgc) {
            if (/\G (\#) ((?:$qq_char)*?) (\#) ([imosxpadlubB]*) /oxgc)                        { return e_split  ($e.'qr',$1,$3,$2,$4);   } # qr# #
            else {
                while (not /\G \z/oxgc) {
                    if    (/\G (\s+|\#.*)                                               /oxgc) { $e .= $1; }
                    elsif (/\G (\()          ((?:$qq_paren)*?)   (\)) ([imosxpadlubB]*) /oxgc) { return e_split  ($e.'qr',$1, $3, $2,$4); } # qr ( )
                    elsif (/\G (\{)          ((?:$qq_brace)*?)   (\}) ([imosxpadlubB]*) /oxgc) { return e_split  ($e.'qr',$1, $3, $2,$4); } # qr { }
                    elsif (/\G (\[)          ((?:$qq_bracket)*?) (\]) ([imosxpadlubB]*) /oxgc) { return e_split  ($e.'qr',$1, $3, $2,$4); } # qr [ ]
                    elsif (/\G (\<)          ((?:$qq_angle)*?)   (\>) ([imosxpadlubB]*) /oxgc) { return e_split  ($e.'qr',$1, $3, $2,$4); } # qr < >
                    elsif (/\G (\')          ((?:$qq_char)*?)    (\') ([imosxpadlubB]*) /oxgc) { return e_split_q($e.'qr',$1, $3, $2,$4); } # qr ' '
                    elsif (/\G ([*\-:?\\^|]) ((?:$qq_char)*?)    (\1) ([imosxpadlubB]*) /oxgc) { return e_split  ($e.'qr','{','}',$2,$4); } # qr | | --> qr { }
                    elsif (/\G (\S)          ((?:$qq_char)*?)    (\1) ([imosxpadlubB]*) /oxgc) { return e_split  ($e.'qr',$1, $3, $2,$4); } # qr * *
                }
                die __FILE__, ": Can't find string terminator anywhere before EOF";
            }
        }

# split q//
        elsif (/\G \b (q) \b /oxgc) {
            if (/\G (\#) ((?:\\\#|\\\\|$q_char)*?) (\#) /oxgc)                    { return e_split_q($e.'qr',$1,$3,$2,'');   } # q# #  --> qr # #
            else {
                while (not /\G \z/oxgc) {
                    if    (/\G (\s+|\#.*)                                  /oxgc) { $e .= $1; }
                    elsif (/\G (\() ((?:\\\\|\\\)|\\\(|$q_paren)*?)   (\)) /oxgc) { return e_split_q($e.'qr',$1,$3,$2,'');   } # q ( ) --> qr ( )
                    elsif (/\G (\{) ((?:\\\\|\\\}|\\\{|$q_brace)*?)   (\}) /oxgc) { return e_split_q($e.'qr',$1,$3,$2,'');   } # q { } --> qr { }
                    elsif (/\G (\[) ((?:\\\\|\\\]|\\\[|$q_bracket)*?) (\]) /oxgc) { return e_split_q($e.'qr',$1,$3,$2,'');   } # q [ ] --> qr [ ]
                    elsif (/\G (\<) ((?:\\\\|\\\>|\\\<|$q_angle)*?)   (\>) /oxgc) { return e_split_q($e.'qr',$1,$3,$2,'');   } # q < > --> qr < >
                    elsif (/\G ([*\-:?\\^|])       ((?:$q_char)*?)    (\1) /oxgc) { return e_split_q($e.'qr','{','}',$2,''); } # q | | --> qr { }
                    elsif (/\G (\S) ((?:\\\\|\\\1|     $q_char)*?)    (\1) /oxgc) { return e_split_q($e.'qr',$1,$3,$2,'');   } # q * * --> qr * *
                }
                die __FILE__, ": Can't find string terminator anywhere before EOF";
            }
        }

# split m//
        elsif (/\G \b (m) \b /oxgc) {
            if (/\G (\#) ((?:$qq_char)*?) (\#) ([cgimosxpadlubB]*) /oxgc)                        { return e_split  ($e.'qr',$1,$3,$2,$4);   } # m# #  --> qr # #
            else {
                while (not /\G \z/oxgc) {
                    if    (/\G (\s+|\#.*)                                                 /oxgc) { $e .= $1; }
                    elsif (/\G (\()          ((?:$qq_paren)*?)   (\)) ([cgimosxpadlubB]*) /oxgc) { return e_split  ($e.'qr',$1, $3, $2,$4); } # m ( ) --> qr ( )
                    elsif (/\G (\{)          ((?:$qq_brace)*?)   (\}) ([cgimosxpadlubB]*) /oxgc) { return e_split  ($e.'qr',$1, $3, $2,$4); } # m { } --> qr { }
                    elsif (/\G (\[)          ((?:$qq_bracket)*?) (\]) ([cgimosxpadlubB]*) /oxgc) { return e_split  ($e.'qr',$1, $3, $2,$4); } # m [ ] --> qr [ ]
                    elsif (/\G (\<)          ((?:$qq_angle)*?)   (\>) ([cgimosxpadlubB]*) /oxgc) { return e_split  ($e.'qr',$1, $3, $2,$4); } # m < > --> qr < >
                    elsif (/\G (\')          ((?:$qq_char)*?)    (\') ([cgimosxpadlubB]*) /oxgc) { return e_split_q($e.'qr',$1, $3, $2,$4); } # m ' ' --> qr ' '
                    elsif (/\G ([*\-:?\\^|]) ((?:$qq_char)*?)    (\1) ([cgimosxpadlubB]*) /oxgc) { return e_split  ($e.'qr','{','}',$2,$4); } # m | | --> qr { }
                    elsif (/\G (\S)          ((?:$qq_char)*?)    (\1) ([cgimosxpadlubB]*) /oxgc) { return e_split  ($e.'qr',$1, $3, $2,$4); } # m * * --> qr * *
                }
                die __FILE__, ": Search pattern not terminated";
            }
        }

# split ''
        elsif (/\G (\') /oxgc) {
            my $q_string = '';
            while (not /\G \z/oxgc) {
                if    (/\G (\\\\)    /oxgc) { $q_string .= $1; }
                elsif (/\G (\\\')    /oxgc) { $q_string .= $1; }                               # splitqr'' --> split qr''
                elsif (/\G \'        /oxgc)                                                    { return e_split_q($e.q{ qr},"'","'",$q_string,''); } # ' ' --> qr ' '
                elsif (/\G ($q_char) /oxgc) { $q_string .= $1; }
            }
            die __FILE__, ": Can't find string terminator anywhere before EOF";
        }

# split ""
        elsif (/\G (\") /oxgc) {
            my $qq_string = '';
            while (not /\G \z/oxgc) {
                if    (/\G (\\\\)    /oxgc) { $qq_string .= $1; }
                elsif (/\G (\\\")    /oxgc) { $qq_string .= $1; }                              # splitqr"" --> split qr""
                elsif (/\G \"        /oxgc)                                                    { return e_split($e.q{ qr},'"','"',$qq_string,''); } # " " --> qr " "
                elsif (/\G ($q_char) /oxgc) { $qq_string .= $1; }
            }
            die __FILE__, ": Can't find string terminator anywhere before EOF";
        }

# split //
        elsif (/\G (\/) /oxgc) {
            my $regexp = '';
            while (not /\G \z/oxgc) {
                if    (/\G (\\\\)                 /oxgc) { $regexp .= $1; }
                elsif (/\G (\\\/)                 /oxgc) { $regexp .= $1; }                    # splitqr// --> split qr//
                elsif (/\G \/ ([cgimosxpadlubB]*) /oxgc)                                       { return e_split($e.q{ qr}, '/','/',$regexp,$1); } # / / --> qr / /
                elsif (/\G ($q_char)              /oxgc) { $regexp .= $1; }
            }
            die __FILE__, ": Search pattern not terminated";
        }
    }

# tr/// or y///

    # about [cdsrbB]* (/B modifier)
    #
    # P.559 appendix C
    # of ISBN 4-89052-384-7 Programming perl
    # (Japanese title is: Perl puroguramingu)

    elsif (/\G \b ( tr | y ) \b /oxgc) {
        my $ope = $1;

        #        $1   $2               $3   $4               $5   $6
        if (/\G (\#) ((?:$qq_char)*?) (\#) ((?:$qq_char)*?) (\#) ([cdsrbB]*) /oxgc) { # tr# # #
            my @tr = ($tr_variable,$2);
            return e_tr(@tr,'',$4,$6);
        }
        else {
            my $e = '';
            while (not /\G \z/oxgc) {
                if    (/\G (\s+|\#.*)                  /oxgc) { $e .= $1; }
                elsif (/\G (\() ((?:$qq_paren)*?) (\)) /oxgc) {
                    my @tr = ($tr_variable,$2);
                    while (not /\G \z/oxgc) {
                        if    (/\G (\s+|\#.*)                                /oxgc) { $e .= $1; }
                        elsif (/\G (\() ((?:$qq_paren)*?)   (\)) ([cdsrbB]*) /oxgc) { return e_tr(@tr,$e,$2,$4); } # tr ( ) ( )
                        elsif (/\G (\{) ((?:$qq_brace)*?)   (\}) ([cdsrbB]*) /oxgc) { return e_tr(@tr,$e,$2,$4); } # tr ( ) { }
                        elsif (/\G (\[) ((?:$qq_bracket)*?) (\]) ([cdsrbB]*) /oxgc) { return e_tr(@tr,$e,$2,$4); } # tr ( ) [ ]
                        elsif (/\G (\<) ((?:$qq_angle)*?)   (\>) ([cdsrbB]*) /oxgc) { return e_tr(@tr,$e,$2,$4); } # tr ( ) < >
                        elsif (/\G (\S) ((?:$qq_char)*?)    (\1) ([cdsrbB]*) /oxgc) { return e_tr(@tr,$e,$2,$4); } # tr ( ) * *
                    }
                    die __FILE__, ": Transliteration replacement not terminated";
                }
                elsif (/\G (\{) ((?:$qq_brace)*?) (\}) /oxgc) {
                    my @tr = ($tr_variable,$2);
                    while (not /\G \z/oxgc) {
                        if    (/\G (\s+|\#.*)                                /oxgc) { $e .= $1; }
                        elsif (/\G (\() ((?:$qq_paren)*?)   (\)) ([cdsrbB]*) /oxgc) { return e_tr(@tr,$e,$2,$4); } # tr { } ( )
                        elsif (/\G (\{) ((?:$qq_brace)*?)   (\}) ([cdsrbB]*) /oxgc) { return e_tr(@tr,$e,$2,$4); } # tr { } { }
                        elsif (/\G (\[) ((?:$qq_bracket)*?) (\]) ([cdsrbB]*) /oxgc) { return e_tr(@tr,$e,$2,$4); } # tr { } [ ]
                        elsif (/\G (\<) ((?:$qq_angle)*?)   (\>) ([cdsrbB]*) /oxgc) { return e_tr(@tr,$e,$2,$4); } # tr { } < >
                        elsif (/\G (\S) ((?:$qq_char)*?)    (\1) ([cdsrbB]*) /oxgc) { return e_tr(@tr,$e,$2,$4); } # tr { } * *
                    }
                    die __FILE__, ": Transliteration replacement not terminated";
                }
                elsif (/\G (\[) ((?:$qq_bracket)*?) (\]) /oxgc) {
                    my @tr = ($tr_variable,$2);
                    while (not /\G \z/oxgc) {
                        if    (/\G (\s+|\#.*)                                /oxgc) { $e .= $1; }
                        elsif (/\G (\() ((?:$qq_paren)*?)   (\)) ([cdsrbB]*) /oxgc) { return e_tr(@tr,$e,$2,$4); } # tr [ ] ( )
                        elsif (/\G (\{) ((?:$qq_brace)*?)   (\}) ([cdsrbB]*) /oxgc) { return e_tr(@tr,$e,$2,$4); } # tr [ ] { }
                        elsif (/\G (\[) ((?:$qq_bracket)*?) (\]) ([cdsrbB]*) /oxgc) { return e_tr(@tr,$e,$2,$4); } # tr [ ] [ ]
                        elsif (/\G (\<) ((?:$qq_angle)*?)   (\>) ([cdsrbB]*) /oxgc) { return e_tr(@tr,$e,$2,$4); } # tr [ ] < >
                        elsif (/\G (\S) ((?:$qq_char)*?)    (\1) ([cdsrbB]*) /oxgc) { return e_tr(@tr,$e,$2,$4); } # tr [ ] * *
                    }
                    die __FILE__, ": Transliteration replacement not terminated";
                }
                elsif (/\G (\<) ((?:$qq_angle)*?) (\>) /oxgc) {
                    my @tr = ($tr_variable,$2);
                    while (not /\G \z/oxgc) {
                        if    (/\G (\s+|\#.*)                                /oxgc) { $e .= $1; }
                        elsif (/\G (\() ((?:$qq_paren)*?)   (\)) ([cdsrbB]*) /oxgc) { return e_tr(@tr,$e,$2,$4); } # tr < > ( )
                        elsif (/\G (\{) ((?:$qq_brace)*?)   (\}) ([cdsrbB]*) /oxgc) { return e_tr(@tr,$e,$2,$4); } # tr < > { }
                        elsif (/\G (\[) ((?:$qq_bracket)*?) (\]) ([cdsrbB]*) /oxgc) { return e_tr(@tr,$e,$2,$4); } # tr < > [ ]
                        elsif (/\G (\<) ((?:$qq_angle)*?)   (\>) ([cdsrbB]*) /oxgc) { return e_tr(@tr,$e,$2,$4); } # tr < > < >
                        elsif (/\G (\S) ((?:$qq_char)*?)    (\1) ([cdsrbB]*) /oxgc) { return e_tr(@tr,$e,$2,$4); } # tr < > * *
                    }
                    die __FILE__, ": Transliteration replacement not terminated";
                }
                #           $1   $2               $3   $4               $5   $6
                elsif (/\G (\S) ((?:$qq_char)*?) (\1) ((?:$qq_char)*?) (\1) ([cdsrbB]*) /oxgc) { # tr * * *
                    my @tr = ($tr_variable,$2);
                    return e_tr(@tr,'',$4,$6);
                }
            }
            die __FILE__, ": Transliteration pattern not terminated";
        }
    }

# qq//
    elsif (/\G \b (qq) \b /oxgc) {
        my $ope = $1;

#       if (/\G (\#) ((?:$qq_char)*?) (\#) /oxgc) { return e_qq($ope,$1,$3,$2); } # qq# #
        if (/\G (\#) /oxgc) {                                                     # qq# #
            my $qq_string = '';
            while (not /\G \z/oxgc) {
                if    (/\G (\\\\)     /oxgc) { $qq_string .= $1;                     }
                elsif (/\G (\\\#)     /oxgc) { $qq_string .= $1;                     }
                elsif (/\G (\#)       /oxgc) { return e_qq($ope,'#','#',$qq_string); }
                elsif (/\G ($qq_char) /oxgc) { $qq_string .= $1;                     }
            }
            die __FILE__, ": Can't find string terminator anywhere before EOF";
        }

        else {
            my $e = '';
            while (not /\G \z/oxgc) {
                if    (/\G (\s+|\#.*)                  /oxgc) { $e .= $1; }

#               elsif (/\G (\() ((?:$qq_paren)*?) (\)) /oxgc) { return $e . e_qq($ope,$1,$3,$2); } # qq ( )
                elsif (/\G (\() /oxgc) {                                                           # qq ( )
                    my $qq_string = '';
                    local $nest = 1;
                    while (not /\G \z/oxgc) {
                        if    (/\G (\\\\)     /oxgc) { $qq_string .= $1;                          }
                        elsif (/\G (\\\))     /oxgc) { $qq_string .= $1;                          }
                        elsif (/\G (\()       /oxgc) { $qq_string .= $1; $nest++;                 }
                        elsif (/\G (\))       /oxgc) {
                            if (--$nest == 0)        { return $e . e_qq($ope,'(',')',$qq_string); }
                            else                     { $qq_string .= $1;                          }
                        }
                        elsif (/\G ($qq_char) /oxgc) { $qq_string .= $1;                          }
                    }
                    die __FILE__, ": Can't find string terminator anywhere before EOF";
                }

#               elsif (/\G (\{) ((?:$qq_brace)*?) (\}) /oxgc) { return $e . e_qq($ope,$1,$3,$2); } # qq { }
                elsif (/\G (\{) /oxgc) {                                                           # qq { }
                    my $qq_string = '';
                    local $nest = 1;
                    while (not /\G \z/oxgc) {
                        if    (/\G (\\\\)     /oxgc) { $qq_string .= $1;                          }
                        elsif (/\G (\\\})     /oxgc) { $qq_string .= $1;                          }
                        elsif (/\G (\{)       /oxgc) { $qq_string .= $1; $nest++;                 }
                        elsif (/\G (\})       /oxgc) {
                            if (--$nest == 0)        { return $e . e_qq($ope,'{','}',$qq_string); }
                            else                     { $qq_string .= $1;                          }
                        }
                        elsif (/\G ($qq_char) /oxgc) { $qq_string .= $1;                          }
                    }
                    die __FILE__, ": Can't find string terminator anywhere before EOF";
                }

#               elsif (/\G (\[) ((?:$qq_bracket)*?) (\]) /oxgc) { return $e . e_qq($ope,$1,$3,$2); } # qq [ ]
                elsif (/\G (\[) /oxgc) {                                                             # qq [ ]
                    my $qq_string = '';
                    local $nest = 1;
                    while (not /\G \z/oxgc) {
                        if    (/\G (\\\\)     /oxgc) { $qq_string .= $1;                          }
                        elsif (/\G (\\\])     /oxgc) { $qq_string .= $1;                          }
                        elsif (/\G (\[)       /oxgc) { $qq_string .= $1; $nest++;                 }
                        elsif (/\G (\])       /oxgc) {
                            if (--$nest == 0)        { return $e . e_qq($ope,'[',']',$qq_string); }
                            else                     { $qq_string .= $1;                          }
                        }
                        elsif (/\G ($qq_char) /oxgc) { $qq_string .= $1;                          }
                    }
                    die __FILE__, ": Can't find string terminator anywhere before EOF";
                }

#               elsif (/\G (\<) ((?:$qq_angle)*?) (\>) /oxgc) { return $e . e_qq($ope,$1,$3,$2); } # qq < >
                elsif (/\G (\<) /oxgc) {                                                           # qq < >
                    my $qq_string = '';
                    local $nest = 1;
                    while (not /\G \z/oxgc) {
                        if    (/\G (\\\\)     /oxgc) { $qq_string .= $1;                          }
                        elsif (/\G (\\\>)     /oxgc) { $qq_string .= $1;                          }
                        elsif (/\G (\<)       /oxgc) { $qq_string .= $1; $nest++;                 }
                        elsif (/\G (\>)       /oxgc) {
                            if (--$nest == 0)        { return $e . e_qq($ope,'<','>',$qq_string); }
                            else                     { $qq_string .= $1;                          }
                        }
                        elsif (/\G ($qq_char) /oxgc) { $qq_string .= $1;                          }
                    }
                    die __FILE__, ": Can't find string terminator anywhere before EOF";
                }

#               elsif (/\G (\S) ((?:$qq_char)*?) (\1) /oxgc) { return $e . e_qq($ope,$1,$3,$2); } # qq * *
                elsif (/\G (\S) /oxgc) {                                                          # qq * *
                    my $delimiter = $1;
                    my $qq_string = '';
                    while (not /\G \z/oxgc) {
                        if    (/\G (\\\\)             /oxgc) { $qq_string .= $1;                                        }
                        elsif (/\G (\\\Q$delimiter\E) /oxgc) { $qq_string .= $1;                                        }
                        elsif (/\G (\Q$delimiter\E)   /oxgc) { return $e . e_qq($ope,$delimiter,$delimiter,$qq_string); }
                        elsif (/\G ($qq_char)         /oxgc) { $qq_string .= $1;                                        }
                    }
                    die __FILE__, ": Can't find string terminator anywhere before EOF";
                }
            }
            die __FILE__, ": Can't find string terminator anywhere before EOF";
        }
    }

# qr//
    elsif (/\G \b (qr) \b /oxgc) {
        my $ope = $1;
        if (/\G (\#) ((?:$qq_char)*?) (\#) ([imosxpadlubB]*) /oxgc) { # qr# # #
            return e_qr($ope,$1,$3,$2,$4);
        }
        else {
            my $e = '';
            while (not /\G \z/oxgc) {
                if    (/\G (\s+|\#.*)                                               /oxgc) { $e .= $1; }
                elsif (/\G (\()          ((?:$qq_paren)*?)   (\)) ([imosxpadlubB]*) /oxgc) { return $e . e_qr  ($ope,$1, $3, $2,$4); } # qr ( )
                elsif (/\G (\{)          ((?:$qq_brace)*?)   (\}) ([imosxpadlubB]*) /oxgc) { return $e . e_qr  ($ope,$1, $3, $2,$4); } # qr { }
                elsif (/\G (\[)          ((?:$qq_bracket)*?) (\]) ([imosxpadlubB]*) /oxgc) { return $e . e_qr  ($ope,$1, $3, $2,$4); } # qr [ ]
                elsif (/\G (\<)          ((?:$qq_angle)*?)   (\>) ([imosxpadlubB]*) /oxgc) { return $e . e_qr  ($ope,$1, $3, $2,$4); } # qr < >
                elsif (/\G (\')          ((?:$qq_char)*?)    (\') ([imosxpadlubB]*) /oxgc) { return $e . e_qr_q($ope,$1, $3, $2,$4); } # qr ' '
                elsif (/\G ([*\-:?\\^|]) ((?:$qq_char)*?)    (\1) ([imosxpadlubB]*) /oxgc) { return $e . e_qr  ($ope,'{','}',$2,$4); } # qr | | --> qr { }
                elsif (/\G (\S)          ((?:$qq_char)*?)    (\1) ([imosxpadlubB]*) /oxgc) { return $e . e_qr  ($ope,$1, $3, $2,$4); } # qr * *
            }
            die __FILE__, ": Can't find string terminator anywhere before EOF";
        }
    }

# qw//
    elsif (/\G \b (qw) \b /oxgc) {
        my $ope = $1;
        if (/\G (\#) (.*?) (\#) /oxmsgc) { # qw# #
            return e_qw($ope,$1,$3,$2);
        }
        else {
            my $e = '';
            while (not /\G \z/oxgc) {
                if    (/\G (\s+|\#.*)                            /oxgc)   { $e .= $1; }

                elsif (/\G (\()          ([^(]*?)           (\)) /oxmsgc) { return $e . e_qw($ope,$1,$3,$2); } # qw ( )
                elsif (/\G (\()          ((?:$q_paren)*?)   (\)) /oxmsgc) { return $e . e_qw($ope,$1,$3,$2); } # qw ( )

                elsif (/\G (\{)          ([^{]*?)           (\}) /oxmsgc) { return $e . e_qw($ope,$1,$3,$2); } # qw { }
                elsif (/\G (\{)          ((?:$q_brace)*?)   (\}) /oxmsgc) { return $e . e_qw($ope,$1,$3,$2); } # qw { }

                elsif (/\G (\[)          ([^[]*?)           (\]) /oxmsgc) { return $e . e_qw($ope,$1,$3,$2); } # qw [ ]
                elsif (/\G (\[)          ((?:$q_bracket)*?) (\]) /oxmsgc) { return $e . e_qw($ope,$1,$3,$2); } # qw [ ]

                elsif (/\G (\<)          ([^<]*?)           (\>) /oxmsgc) { return $e . e_qw($ope,$1,$3,$2); } # qw < >
                elsif (/\G (\<)          ((?:$q_angle)*?)   (\>) /oxmsgc) { return $e . e_qw($ope,$1,$3,$2); } # qw < >

                elsif (/\G ([\x21-\x3F]) (.*?)              (\1) /oxmsgc) { return $e . e_qw($ope,$1,$3,$2); } # qw * *
                elsif (/\G (\S)          ((?:$q_char)*?)    (\1) /oxmsgc) { return $e . e_qw($ope,$1,$3,$2); } # qw * *
            }
            die __FILE__, ": Can't find string terminator anywhere before EOF";
        }
    }

# qx//
    elsif (/\G \b (qx) \b /oxgc) {
        my $ope = $1;
        if (/\G (\#) ((?:$qq_char)*?) (\#) /oxgc) { # qx# #
            return e_qq($ope,$1,$3,$2);
        }
        else {
            my $e = '';
            while (not /\G \z/oxgc) {
                if    (/\G (\s+|\#.*)                    /oxgc) { $e .= $1; }
                elsif (/\G (\() ((?:$qq_paren)*?)   (\)) /oxgc) { return $e . e_qq($ope,$1,$3,$2); } # qx ( )
                elsif (/\G (\{) ((?:$qq_brace)*?)   (\}) /oxgc) { return $e . e_qq($ope,$1,$3,$2); } # qx { }
                elsif (/\G (\[) ((?:$qq_bracket)*?) (\]) /oxgc) { return $e . e_qq($ope,$1,$3,$2); } # qx [ ]
                elsif (/\G (\<) ((?:$qq_angle)*?)   (\>) /oxgc) { return $e . e_qq($ope,$1,$3,$2); } # qx < >
                elsif (/\G (\') ((?:$qq_char)*?)    (\') /oxgc) { return $e . e_q ($ope,$1,$3,$2); } # qx ' '
                elsif (/\G (\S) ((?:$qq_char)*?)    (\1) /oxgc) { return $e . e_qq($ope,$1,$3,$2); } # qx * *
            }
            die __FILE__, ": Can't find string terminator anywhere before EOF";
        }
    }

# q//
    elsif (/\G \b (q) \b /oxgc) {
        my $ope = $1;

#       if (/\G (\#) ((?:\\\#|\\\\|$q_char)*?) (\#) /oxgc) { return e_q($ope,$1,$3,$2); } # q# #

        # avoid "Error: Runtime exception" of perl version 5.005_03
        # (and so on)

        if (/\G (\#) /oxgc) {                                                             # q# #
            my $q_string = '';
            while (not /\G \z/oxgc) {
                if    (/\G (\\\\)    /oxgc) { $q_string .= $1;                    }
                elsif (/\G (\\\#)    /oxgc) { $q_string .= $1;                    }
                elsif (/\G (\#)      /oxgc) { return e_q($ope,'#','#',$q_string); }
                elsif (/\G ($q_char) /oxgc) { $q_string .= $1;                    }
            }
            die __FILE__, ": Can't find string terminator anywhere before EOF";
        }

        else {
            my $e = '';
            while (not /\G \z/oxgc) {
                if    (/\G (\s+|\#.*)                           /oxgc) { $e .= $1; }

#               elsif (/\G (\() ((?:\\\)|\\\\|$q_paren)*?) (\)) /oxgc) { return $e . e_q($ope,$1,$3,$2); } # q ( )
                elsif (/\G (\() /oxgc) {                                                                   # q ( )
                    my $q_string = '';
                    local $nest = 1;
                    while (not /\G \z/oxgc) {
                        if    (/\G (\\\\)    /oxgc) { $q_string .= $1;                         }
                        elsif (/\G (\\\))    /oxgc) { $q_string .= $1;                         }
                        elsif (/\G (\\\()    /oxgc) { $q_string .= $1;                         }
                        elsif (/\G (\()      /oxgc) { $q_string .= $1; $nest++;                }
                        elsif (/\G (\))      /oxgc) {
                            if (--$nest == 0)       { return $e . e_q($ope,'(',')',$q_string); }
                            else                    { $q_string .= $1;                         }
                        }
                        elsif (/\G ($q_char) /oxgc) { $q_string .= $1;                         }
                    }
                    die __FILE__, ": Can't find string terminator anywhere before EOF";
                }

#               elsif (/\G (\{) ((?:\\\}|\\\\|$q_brace)*?) (\}) /oxgc) { return $e . e_q($ope,$1,$3,$2); } # q { }
                elsif (/\G (\{) /oxgc) {                                                                   # q { }
                    my $q_string = '';
                    local $nest = 1;
                    while (not /\G \z/oxgc) {
                        if    (/\G (\\\\)    /oxgc) { $q_string .= $1;                         }
                        elsif (/\G (\\\})    /oxgc) { $q_string .= $1;                         }
                        elsif (/\G (\\\{)    /oxgc) { $q_string .= $1;                         }
                        elsif (/\G (\{)      /oxgc) { $q_string .= $1; $nest++;                }
                        elsif (/\G (\})      /oxgc) {
                            if (--$nest == 0)       { return $e . e_q($ope,'{','}',$q_string); }
                            else                    { $q_string .= $1;                         }
                        }
                        elsif (/\G ($q_char) /oxgc) { $q_string .= $1;                         }
                    }
                    die __FILE__, ": Can't find string terminator anywhere before EOF";
                }

#               elsif (/\G (\[) ((?:\\\]|\\\\|$q_bracket)*?) (\]) /oxgc) { return $e . e_q($ope,$1,$3,$2); } # q [ ]
                elsif (/\G (\[) /oxgc) {                                                                     # q [ ]
                    my $q_string = '';
                    local $nest = 1;
                    while (not /\G \z/oxgc) {
                        if    (/\G (\\\\)    /oxgc) { $q_string .= $1;                         }
                        elsif (/\G (\\\])    /oxgc) { $q_string .= $1;                         }
                        elsif (/\G (\\\[)    /oxgc) { $q_string .= $1;                         }
                        elsif (/\G (\[)      /oxgc) { $q_string .= $1; $nest++;                }
                        elsif (/\G (\])      /oxgc) {
                            if (--$nest == 0)       { return $e . e_q($ope,'[',']',$q_string); }
                            else                    { $q_string .= $1;                         }
                        }
                        elsif (/\G ($q_char) /oxgc) { $q_string .= $1;                         }
                    }
                    die __FILE__, ": Can't find string terminator anywhere before EOF";
                }

#               elsif (/\G (\<) ((?:\\\>|\\\\|$q_angle)*?) (\>) /oxgc) { return $e . e_q($ope,$1,$3,$2); } # q < >
                elsif (/\G (\<) /oxgc) {                                                                   # q < >
                    my $q_string = '';
                    local $nest = 1;
                    while (not /\G \z/oxgc) {
                        if    (/\G (\\\\)    /oxgc) { $q_string .= $1;                         }
                        elsif (/\G (\\\>)    /oxgc) { $q_string .= $1;                         }
                        elsif (/\G (\\\<)    /oxgc) { $q_string .= $1;                         }
                        elsif (/\G (\<)      /oxgc) { $q_string .= $1; $nest++;                }
                        elsif (/\G (\>)      /oxgc) {
                            if (--$nest == 0)       { return $e . e_q($ope,'<','>',$q_string); }
                            else                    { $q_string .= $1;                         }
                        }
                        elsif (/\G ($q_char) /oxgc) { $q_string .= $1;                         }
                    }
                    die __FILE__, ": Can't find string terminator anywhere before EOF";
                }

#               elsif (/\G (\S) ((?:\\\1|\\\\|$q_char)*?) (\1) /oxgc) { return $e . e_q($ope,$1,$3,$2); } # q * *
                elsif (/\G (\S) /oxgc) {                                                                  # q * *
                    my $delimiter = $1;
                    my $q_string = '';
                    while (not /\G \z/oxgc) {
                        if    (/\G (\\\\)             /oxgc) { $q_string .= $1;                                       }
                        elsif (/\G (\\\Q$delimiter\E) /oxgc) { $q_string .= $1;                                       }
                        elsif (/\G (\Q$delimiter\E)   /oxgc) { return $e . e_q($ope,$delimiter,$delimiter,$q_string); }
                        elsif (/\G ($q_char)          /oxgc) { $q_string .= $1;                                       }
                    }
                    die __FILE__, ": Can't find string terminator anywhere before EOF";
                }
            }
            die __FILE__, ": Can't find string terminator anywhere before EOF";
        }
    }

# m//
    elsif (/\G \b (m) \b /oxgc) {
        my $ope = $1;
        if (/\G (\#) ((?:$qq_char)*?) (\#) ([cgimosxpadlubB]*) /oxgc) { # m# #
            return e_qr($ope,$1,$3,$2,$4);
        }
        else {
            my $e = '';
            while (not /\G \z/oxgc) {
                if    (/\G (\s+|\#.*)                                                /oxgc) { $e .= $1; }
                elsif (/\G (\()         ((?:$qq_paren)*?)   (\)) ([cgimosxpadlubB]*) /oxgc) { return $e . e_qr  ($ope,$1, $3, $2,$4); } # m ( )
                elsif (/\G (\{)         ((?:$qq_brace)*?)   (\}) ([cgimosxpadlubB]*) /oxgc) { return $e . e_qr  ($ope,$1, $3, $2,$4); } # m { }
                elsif (/\G (\[)         ((?:$qq_bracket)*?) (\]) ([cgimosxpadlubB]*) /oxgc) { return $e . e_qr  ($ope,$1, $3, $2,$4); } # m [ ]
                elsif (/\G (\<)         ((?:$qq_angle)*?)   (\>) ([cgimosxpadlubB]*) /oxgc) { return $e . e_qr  ($ope,$1, $3, $2,$4); } # m < >
                elsif (/\G (\?)         ((?:$qq_char)*?)    (\?) ([cgimosxpadlubB]*) /oxgc) { return $e . e_qr  ($ope,$1, $3, $2,$4); } # m ? ?
                elsif (/\G (\')         ((?:$qq_char)*?)    (\') ([cgimosxpadlubB]*) /oxgc) { return $e . e_qr_q($ope,$1, $3, $2,$4); } # m ' '
                elsif (/\G ([*\-:\\^|]) ((?:$qq_char)*?)    (\1) ([cgimosxpadlubB]*) /oxgc) { return $e . e_qr  ($ope,'{','}',$2,$4); } # m | | --> m { }
                elsif (/\G (\S)         ((?:$qq_char)*?)    (\1) ([cgimosxpadlubB]*) /oxgc) { return $e . e_qr  ($ope,$1, $3, $2,$4); } # m * *
            }
            die __FILE__, ": Search pattern not terminated";
        }
    }

# s///

    # about [cegimosxpradlubB]* (/cg modifier)
    #
    # P.67 Pattern-Matching Operators
    # of ISBN 0-596-00241-6 Perl in a Nutshell, Second Edition.

    elsif (/\G \b (s) \b /oxgc) {
        my $ope = $1;

        #        $1   $2               $3   $4               $5   $6
        if (/\G (\#) ((?:$qq_char)*?) (\#) ((?:$qq_char)*?) (\#) ([cegimosxpradlubB]*) /oxgc) { # s# # #
            return e_sub($sub_variable,$1,$2,$3,$3,$4,$5,$6);
        }
        else {
            my $e = '';
            while (not /\G \z/oxgc) {
                if (/\G (\s+|\#.*) /oxgc) { $e .= $1; }
                elsif (/\G (\() ((?:$qq_paren)*?) (\)) /oxgc) {
                    my @s = ($1,$2,$3);
                    while (not /\G \z/oxgc) {
                        if    (/\G (\s+|\#.*)                                          /oxgc) { $e .= $1; }
                        #           $1   $2                  $3   $4
                        elsif (/\G (\() ((?:$qq_paren)*?)   (\)) ([cegimosxpradlubB]*) /oxgc) { return e_sub($sub_variable,@s,$1,$2,$3,$4); }
                        elsif (/\G (\{) ((?:$qq_brace)*?)   (\}) ([cegimosxpradlubB]*) /oxgc) { return e_sub($sub_variable,@s,$1,$2,$3,$4); }
                        elsif (/\G (\[) ((?:$qq_bracket)*?) (\]) ([cegimosxpradlubB]*) /oxgc) { return e_sub($sub_variable,@s,$1,$2,$3,$4); }
                        elsif (/\G (\<) ((?:$qq_angle)*?)   (\>) ([cegimosxpradlubB]*) /oxgc) { return e_sub($sub_variable,@s,$1,$2,$3,$4); }
                        elsif (/\G (\') ((?:$qq_char)*?)    (\') ([cegimosxpradlubB]*) /oxgc) { return e_sub($sub_variable,@s,$1,$2,$3,$4); }
                        elsif (/\G (\$) ((?:$qq_char)*?)    (\$) ([cegimosxpradlubB]*) /oxgc) { return e_sub($sub_variable,@s,$1,$2,$3,$4); }
                        elsif (/\G (\:) ((?:$qq_char)*?)    (\:) ([cegimosxpradlubB]*) /oxgc) { return e_sub($sub_variable,@s,$1,$2,$3,$4); }
                        elsif (/\G (\@) ((?:$qq_char)*?)    (\@) ([cegimosxpradlubB]*) /oxgc) { return e_sub($sub_variable,@s,$1,$2,$3,$4); }
                        elsif (/\G (\S) ((?:$qq_char)*?)    (\1) ([cegimosxpradlubB]*) /oxgc) { return e_sub($sub_variable,@s,$1,$2,$3,$4); }
                    }
                    die __FILE__, ": Substitution replacement not terminated";
                }
                elsif (/\G (\{) ((?:$qq_brace)*?) (\}) /oxgc) {
                    my @s = ($1,$2,$3);
                    while (not /\G \z/oxgc) {
                        if    (/\G (\s+|\#.*)                                          /oxgc) { $e .= $1; }
                        #           $1   $2                  $3   $4
                        elsif (/\G (\() ((?:$qq_paren)*?)   (\)) ([cegimosxpradlubB]*) /oxgc) { return e_sub($sub_variable,@s,$1,$2,$3,$4); }
                        elsif (/\G (\{) ((?:$qq_brace)*?)   (\}) ([cegimosxpradlubB]*) /oxgc) { return e_sub($sub_variable,@s,$1,$2,$3,$4); }
                        elsif (/\G (\[) ((?:$qq_bracket)*?) (\]) ([cegimosxpradlubB]*) /oxgc) { return e_sub($sub_variable,@s,$1,$2,$3,$4); }
                        elsif (/\G (\<) ((?:$qq_angle)*?)   (\>) ([cegimosxpradlubB]*) /oxgc) { return e_sub($sub_variable,@s,$1,$2,$3,$4); }
                        elsif (/\G (\') ((?:$qq_char)*?)    (\') ([cegimosxpradlubB]*) /oxgc) { return e_sub($sub_variable,@s,$1,$2,$3,$4); }
                        elsif (/\G (\$) ((?:$qq_char)*?)    (\$) ([cegimosxpradlubB]*) /oxgc) { return e_sub($sub_variable,@s,$1,$2,$3,$4); }
                        elsif (/\G (\:) ((?:$qq_char)*?)    (\:) ([cegimosxpradlubB]*) /oxgc) { return e_sub($sub_variable,@s,$1,$2,$3,$4); }
                        elsif (/\G (\@) ((?:$qq_char)*?)    (\@) ([cegimosxpradlubB]*) /oxgc) { return e_sub($sub_variable,@s,$1,$2,$3,$4); }
                        elsif (/\G (\S) ((?:$qq_char)*?)    (\1) ([cegimosxpradlubB]*) /oxgc) { return e_sub($sub_variable,@s,$1,$2,$3,$4); }
                    }
                    die __FILE__, ": Substitution replacement not terminated";
                }
                elsif (/\G (\[) ((?:$qq_bracket)*?) (\]) /oxgc) {
                    my @s = ($1,$2,$3);
                    while (not /\G \z/oxgc) {
                        if    (/\G (\s+|\#.*)                                          /oxgc) { $e .= $1; }
                        #           $1   $2                  $3   $4
                        elsif (/\G (\() ((?:$qq_paren)*?)   (\)) ([cegimosxpradlubB]*) /oxgc) { return e_sub($sub_variable,@s,$1,$2,$3,$4); }
                        elsif (/\G (\{) ((?:$qq_brace)*?)   (\}) ([cegimosxpradlubB]*) /oxgc) { return e_sub($sub_variable,@s,$1,$2,$3,$4); }
                        elsif (/\G (\[) ((?:$qq_bracket)*?) (\]) ([cegimosxpradlubB]*) /oxgc) { return e_sub($sub_variable,@s,$1,$2,$3,$4); }
                        elsif (/\G (\<) ((?:$qq_angle)*?)   (\>) ([cegimosxpradlubB]*) /oxgc) { return e_sub($sub_variable,@s,$1,$2,$3,$4); }
                        elsif (/\G (\') ((?:$qq_char)*?)    (\') ([cegimosxpradlubB]*) /oxgc) { return e_sub($sub_variable,@s,$1,$2,$3,$4); }
                        elsif (/\G (\$) ((?:$qq_char)*?)    (\$) ([cegimosxpradlubB]*) /oxgc) { return e_sub($sub_variable,@s,$1,$2,$3,$4); }
                        elsif (/\G (\S) ((?:$qq_char)*?)    (\1) ([cegimosxpradlubB]*) /oxgc) { return e_sub($sub_variable,@s,$1,$2,$3,$4); }
                    }
                    die __FILE__, ": Substitution replacement not terminated";
                }
                elsif (/\G (\<) ((?:$qq_angle)*?) (\>) /oxgc) {
                    my @s = ($1,$2,$3);
                    while (not /\G \z/oxgc) {
                        if    (/\G (\s+|\#.*)                                          /oxgc) { $e .= $1; }
                        #           $1   $2                  $3   $4
                        elsif (/\G (\() ((?:$qq_paren)*?)   (\)) ([cegimosxpradlubB]*) /oxgc) { return e_sub($sub_variable,@s,$1,$2,$3,$4); }
                        elsif (/\G (\{) ((?:$qq_brace)*?)   (\}) ([cegimosxpradlubB]*) /oxgc) { return e_sub($sub_variable,@s,$1,$2,$3,$4); }
                        elsif (/\G (\[) ((?:$qq_bracket)*?) (\]) ([cegimosxpradlubB]*) /oxgc) { return e_sub($sub_variable,@s,$1,$2,$3,$4); }
                        elsif (/\G (\<) ((?:$qq_angle)*?)   (\>) ([cegimosxpradlubB]*) /oxgc) { return e_sub($sub_variable,@s,$1,$2,$3,$4); }
                        elsif (/\G (\') ((?:$qq_char)*?)    (\') ([cegimosxpradlubB]*) /oxgc) { return e_sub($sub_variable,@s,$1,$2,$3,$4); }
                        elsif (/\G (\$) ((?:$qq_char)*?)    (\$) ([cegimosxpradlubB]*) /oxgc) { return e_sub($sub_variable,@s,$1,$2,$3,$4); }
                        elsif (/\G (\:) ((?:$qq_char)*?)    (\:) ([cegimosxpradlubB]*) /oxgc) { return e_sub($sub_variable,@s,$1,$2,$3,$4); }
                        elsif (/\G (\@) ((?:$qq_char)*?)    (\@) ([cegimosxpradlubB]*) /oxgc) { return e_sub($sub_variable,@s,$1,$2,$3,$4); }
                        elsif (/\G (\S) ((?:$qq_char)*?)    (\1) ([cegimosxpradlubB]*) /oxgc) { return e_sub($sub_variable,@s,$1,$2,$3,$4); }
                    }
                    die __FILE__, ": Substitution replacement not terminated";
                }
                #           $1   $2               $3   $4               $5   $6
                elsif (/\G (\') ((?:$qq_char)*?) (\') ((?:$qq_char)*?) (\') ([cegimosxpradlubB]*) /oxgc) {
                    return e_sub($sub_variable,$1,$2,$3,$3,$4,$5,$6);
                }
                #           $1            $2               $3   $4               $5   $6
                elsif (/\G ([*\-:?\\^|]) ((?:$qq_char)*?) (\1) ((?:$qq_char)*?) (\1) ([cegimosxpradlubB]*) /oxgc) {
                    return e_sub($sub_variable,'{',$2,'}','{',$4,'}',$6); # s | | | --> s { } { }
                }
                #           $1   $2               $3   $4               $5   $6
                elsif (/\G (\$) ((?:$qq_char)*?) (\1) ((?:$qq_char)*?) (\1) ([cegimosxpradlubB]*) /oxgc) {
                    return e_sub($sub_variable,$1,$2,$3,$3,$4,$5,$6);
                }
                #           $1   $2               $3   $4               $5   $6
                elsif (/\G (\S) ((?:$qq_char)*?) (\1) ((?:$qq_char)*?) (\1) ([cegimosxpradlubB]*) /oxgc) {
                    return e_sub($sub_variable,$1,$2,$3,$3,$4,$5,$6);
                }
            }
            die __FILE__, ": Substitution pattern not terminated";
        }
    }

# require ignore module
    elsif (/\G \b require (\s+ (?:$ignore_modules) .*? ;) ([ \t]* [#\n])                  /oxmsgc) { return "# require$1$2";   }
    elsif (/\G \b require (\s+ (?:$ignore_modules) .*? ;) ([ \t]* [^#]) /oxmsgc) { return "# require$1\n$2"; }
    elsif (/\G \b require (\s+ (?:$ignore_modules)) \b                                    /oxmsgc) { return "# require$1";     }

# use strict; --> use strict; no strict qw(refs);
    elsif (/\G \b use (\s+ strict .*? ;) ([ \t]* [#\n])                               /oxmsgc) { return "use$1 no strict qw(refs);$2";   }
    elsif (/\G \b use (\s+ strict .*? ;) ([ \t]* [^#])              /oxmsgc) { return "use$1 no strict qw(refs);\n$2"; }
    elsif (/\G \b use (\s+ strict) \b                                                 /oxmsgc) { return "use$1; no strict qw(refs)";     }

# use 5.12.0; --> use 5.12.0; no strict qw(refs);
    elsif (/\G \b use \s+ (([1-9][0-9_]*)(?:\.([0-9_]+))*)  \s* ;                     /oxmsgc) {
        if (($2 >= 6) or (($2 == 5) and ($3 ge '012'))) {
            return "use $1; no strict qw(refs);";
        }
        else {
            return "use $1;";
        }
    }
    elsif (/\G \b use \s+ (v([0-9][0-9_]*)(?:\.([0-9_]+))*) \s* ;                    /oxmsgc)  {
        if (($2 >= 6) or (($2 == 5) and ($3 >= 12))) {
            return "use $1; no strict qw(refs);";
        }
        else {
            return "use $1;";
        }
    }

# ignore use module
    elsif (/\G \b use (\s+ (?:$ignore_modules) .*? ;) ([ \t]* [#\n])                  /oxmsgc) { return "# use$1$2";         }
    elsif (/\G \b use (\s+ (?:$ignore_modules) .*? ;) ([ \t]* [^#]) /oxmsgc) { return "# use$1\n$2";       }
    elsif (/\G \b use (\s+ (?:$ignore_modules)) \b                                    /oxmsgc) { return "# use$1";           }

# ignore no module
    elsif (/\G \b no  (\s+ (?:$ignore_modules) .*? ;) ([ \t]* [#\n])                  /oxmsgc) { return "# no$1$2";          }
    elsif (/\G \b no  (\s+ (?:$ignore_modules) .*? ;) ([ \t]* [^#]) /oxmsgc) { return "# no$1\n$2";        }
    elsif (/\G \b no  (\s+ (?:$ignore_modules)) \b                                    /oxmsgc) { return "# no$1";            }

# use else
    elsif (/\G \b use \b                                                              /oxmsgc) { return "use";               }

# use else
    elsif (/\G \b no  \b                                                              /oxmsgc) { return "no";                }

# ''
    elsif (/\G (?
        my $q_string = '';
        while (not /\G \z/oxgc) {
            if    (/\G (\\\\)                 /oxgc) { $q_string .= $1;                   }
            elsif (/\G (\\\')                 /oxgc) { $q_string .= $1;                   }
            elsif (/\G \'                     /oxgc) { return e_q('', "'","'",$q_string); }
            elsif (/\G ($q_char)              /oxgc) { $q_string .= $1;                   }
        }
        die __FILE__, ": Can't find string terminator anywhere before EOF";
    }

# ""
    elsif (/\G (\") /oxgc) {
        my $qq_string = '';
        while (not /\G \z/oxgc) {
            if    (/\G (\\\\)                 /oxgc) { $qq_string .= $1;                    }
            elsif (/\G (\\\")                 /oxgc) { $qq_string .= $1;                    }
            elsif (/\G \"                     /oxgc) { return e_qq('', '"','"',$qq_string); }
            elsif (/\G ($q_char)              /oxgc) { $qq_string .= $1;                    }
        }
        die __FILE__, ": Can't find string terminator anywhere before EOF";
    }

# ``
    elsif (/\G (\`) /oxgc) {
        my $qx_string = '';
        while (not /\G \z/oxgc) {
            if    (/\G (\\\\)                 /oxgc) { $qx_string .= $1;                    }
            elsif (/\G (\\\`)                 /oxgc) { $qx_string .= $1;                    }
            elsif (/\G \`                     /oxgc) { return e_qq('', '`','`',$qx_string); }
            elsif (/\G ($q_char)              /oxgc) { $qx_string .= $1;                    }
        }
        die __FILE__, ": Can't find string terminator anywhere before EOF";
    }

# //   --- not divide operator (num / num), not defined-or
    elsif (($slash eq 'm//') and /\G (\/) /oxgc) {
        my $regexp = '';
        while (not /\G \z/oxgc) {
            if    (/\G (\\\\)                 /oxgc) { $regexp .= $1;                       }
            elsif (/\G (\\\/)                 /oxgc) { $regexp .= $1;                       }
            elsif (/\G \/ ([cgimosxpadlubB]*) /oxgc) { return e_qr('', '/','/',$regexp,$1); }
            elsif (/\G ($q_char)              /oxgc) { $regexp .= $1;                       }
        }
        die __FILE__, ": Search pattern not terminated";
    }

# ??   --- not conditional operator (condition ? then : else)
    elsif (($slash eq 'm//') and /\G (\?) /oxgc) {
        my $regexp = '';
        while (not /\G \z/oxgc) {
            if    (/\G (\\\\)                 /oxgc) { $regexp .= $1;                       }
            elsif (/\G (\\\?)                 /oxgc) { $regexp .= $1;                       }
            elsif (/\G \? ([cgimosxpadlubB]*) /oxgc) { return e_qr('m','?','?',$regexp,$1); }
            elsif (/\G ($q_char)              /oxgc) { $regexp .= $1;                       }
        }
        die __FILE__, ": Search pattern not terminated";
    }

# << (bit shift)   --- not here document
    elsif (/\G ( << \s* ) (?= [0-9\$\@\&] ) /oxgc)   { $slash = 'm//'; return $1;           }

# <<'HEREDOC'
    elsif (/\G ( << '([a-zA-Z_0-9]*)' ) /oxgc) {
        $slash = 'm//';
        my $here_quote = $1;
        my $delimiter  = $2;

        # get here document
        if ($here_script eq '') {
            $here_script = CORE::substr $_, pos $_;
            $here_script =~ s/.*?\n//oxm;
        }
        if ($here_script =~ s/\A (.*?) \n $delimiter \n //xms) {
            push @heredoc, $1 . qq{\n$delimiter\n};
            push @heredoc_delimiter, $delimiter;
        }
        else {
            die __FILE__, ": Can't find string terminator $delimiter anywhere before EOF";
        }
        return $here_quote;
    }

# <<\HEREDOC

    # P.66 2.6.6. "Here" Documents
    # in Chapter 2: Bits and Pieces
    # of ISBN 0-596-00027-8 Programming Perl Third Edition.

    # P.73 "Here" Documents
    # in Chapter 2: Bits and Pieces
    # of ISBN 978-0-596-00492-7 Programming Perl 4th Edition.

    elsif (/\G ( << \\([a-zA-Z_0-9]+) ) /oxgc) {
        $slash = 'm//';
        my $here_quote = $1;
        my $delimiter  = $2;

        # get here document
        if ($here_script eq '') {
            $here_script = CORE::substr $_, pos $_;
            $here_script =~ s/.*?\n//oxm;
        }
        if ($here_script =~ s/\A (.*?) \n $delimiter \n //xms) {
            push @heredoc, $1 . qq{\n$delimiter\n};
            push @heredoc_delimiter, $delimiter;
        }
        else {
            die __FILE__, ": Can't find string terminator $delimiter anywhere before EOF";
        }
        return $here_quote;
    }

# <<"HEREDOC"
    elsif (/\G ( << "([a-zA-Z_0-9]*)" ) /oxgc) {
        $slash = 'm//';
        my $here_quote = $1;
        my $delimiter  = $2;

        # get here document
        if ($here_script eq '') {
            $here_script = CORE::substr $_, pos $_;
            $here_script =~ s/.*?\n//oxm;
        }
        if ($here_script =~ s/\A (.*?) \n $delimiter \n //xms) {
            push @heredoc, e_heredoc($1) . qq{\n$delimiter\n};
            push @heredoc_delimiter, $delimiter;
        }
        else {
            die __FILE__, ": Can't find string terminator $delimiter anywhere before EOF";
        }
        return $here_quote;
    }

# <
    elsif (/\G ( << ([a-zA-Z_0-9]+) ) /oxgc) {
        $slash = 'm//';
        my $here_quote = $1;
        my $delimiter  = $2;

        # get here document
        if ($here_script eq '') {
            $here_script = CORE::substr $_, pos $_;
            $here_script =~ s/.*?\n//oxm;
        }
        if ($here_script =~ s/\A (.*?) \n $delimiter \n //xms) {
            push @heredoc, e_heredoc($1) . qq{\n$delimiter\n};
            push @heredoc_delimiter, $delimiter;
        }
        else {
            die __FILE__, ": Can't find string terminator $delimiter anywhere before EOF";
        }
        return $here_quote;
    }

# <<`HEREDOC`
    elsif (/\G ( << `([a-zA-Z_0-9]*)` ) /oxgc) {
        $slash = 'm//';
        my $here_quote = $1;
        my $delimiter  = $2;

        # get here document
        if ($here_script eq '') {
            $here_script = CORE::substr $_, pos $_;
            $here_script =~ s/.*?\n//oxm;
        }
        if ($here_script =~ s/\A (.*?) \n $delimiter \n //xms) {
            push @heredoc, e_heredoc($1) . qq{\n$delimiter\n};
            push @heredoc_delimiter, $delimiter;
        }
        else {
            die __FILE__, ": Can't find string terminator $delimiter anywhere before EOF";
        }
        return $here_quote;
    }

# <<= <=> <= < operator
    elsif (/\G ( <<= | <=> | <= | < ) (?= \s* [A-Za-z_0-9'"`\$\@\&\*\(\+\-] )/oxgc) {
        return $1;
    }

#
    elsif (/\G (<[\$]?[A-Za-z_][A-Za-z_0-9]*>) /oxgc) {
        return $1;
    }

#  --- glob

    # avoid "Error: Runtime exception" of perl version 5.005_03

    elsif (/\G < ((?:[^>\0\a\e\f\n\r\t])+?) > /oxgc) {
        return 'Char::Earabic::glob("' . $1 . '")';
    }

# __DATA__
    elsif (/\G ^ ( __DATA__ \n .*) \z /oxmsgc) { return $1; }

# __END__
    elsif (/\G ^ ( __END__  \n .*) \z /oxmsgc) { return $1; }

# \cD Control-D

    # P.68 2.6.8. Other Literal Tokens
    # in Chapter 2: Bits and Pieces
    # of ISBN 0-596-00027-8 Programming Perl Third Edition.

    # P.76 Other Literal Tokens
    # in Chapter 2: Bits and Pieces
    # of ISBN 978-0-596-00492-7 Programming Perl 4th Edition.

    elsif (/\G   ( \cD         .*) \z /oxmsgc) { return $1; }

# \cZ Control-Z
    elsif (/\G   ( \cZ         .*) \z /oxmsgc) { return $1; }

    # any operator before div
    elsif (/\G (
            -- | \+\+ |
            [\)\}\]]

            ) /oxgc) { $slash = 'div'; return $1; }

    # any operator before m//

    # //, //= (defined-or)

    # P.164 Logical Operators
    # in Chapter 10: More Control Structures
    # of ISBN 978-0-596-52010-6 Learning Perl, Fifth Edition

    # P.119 C-Style Logical (Short-Circuit) Operators
    # in Chapter 3: Unary and Binary Operators
    # of ISBN 978-0-596-00492-7 Programming Perl 4th Edition.

    # (and so on)

    # ~~

    # P.221 The Smart Match Operator
    # in Chapter 15: Smart Matching and given-when
    # of ISBN 978-0-596-52010-6 Learning Perl, Fifth Edition

    # P.112 Smartmatch Operator
    # in Chapter 3: Unary and Binary Operators
    # of ISBN 978-0-596-00492-7 Programming Perl 4th Edition.

    # (and so on)

    elsif (/\G (

            !~~ | !~ | != | ! |
            %= | % |
            &&= | && | &= | & |
            -= | -> | - |
            :\s*= |
            : |
            <<= | <=> | <= | < |
            == | => | =~ | = |
            >>= | >> | >= | > |
            \*\*= | \*\* | \*= | \* |
            \+= | \+ |
            \.\.\. | \.\. | \.= | \. |
            \/\/= | \/\/ |
            \/= | \/ |
            \? |
            \\ |
            \^= | \^ |
            \b x= |
            \|\|= | \|\| | \|= | \| |
            ~~ | ~ |
            \b(?: and | cmp | eq | ge | gt | le | lt | ne | not | or | xor | x )\b |
            \b(?: print )\b |

            [,;\(\{\[]

            ) /oxgc) { $slash = 'm//'; return $1; }

    # other any character
    elsif (/\G ($q_char) /oxgc) { $slash = 'div'; return $1; }

    # system error
    else {
        die __FILE__, ": Oops, this shouldn't happen!";
    }
}

# escape Arabic string
sub e_string {
    my($string) = @_;
    my $e_string = '';

    local $slash = 'm//';

    # P.1024 Appendix W.10 Multibyte Processing
    # of ISBN 1-56592-224-7 CJKV Information Processing
    # (and so on)

    my @char = $string =~ / \G (\\?(?:$q_char)) /oxmsg;

    # without { ... }
    if (not (grep(/\A \{ \z/xms, @char) and grep(/\A \} \z/xms, @char))) {
        if ($string !~ /<
            return $string;
        }
    }

E_STRING_LOOP:
    while ($string !~ /\G \z/oxgc) {
        if (0) {
        }

# $`, ${`}, $PREMATCH, ${PREMATCH}, ${^PREMATCH} --> @{[Char::Earabic::PREMATCH()]}
        elsif ($string =~ /\G ( \$` | \$\{`\} | \$ \s* PREMATCH  \b | \$ \s* \{ \s* PREMATCH  \s* \} | \$ \s* \{\^PREMATCH\}  ) /oxmsgc) {
            $e_string .= q{Char::Earabic::PREMATCH()};
            $slash = 'div';
        }

# $&, ${&}, $MATCH, ${MATCH}, ${^MATCH} --> @{[Char::Earabic::MATCH()]}
        elsif ($string =~ /\G ( \$& | \$\{&\} | \$ \s* MATCH     \b | \$ \s* \{ \s* MATCH     \s* \} | \$ \s* \{\^MATCH\}     ) /oxmsgc) {
            $e_string .= q{Char::Earabic::MATCH()};
            $slash = 'div';
        }

# $', ${'} --> $', ${'}
        elsif ($string =~ /\G ( \$' | \$\{'\}                                                                                 ) /oxmsgc) {
            $e_string .= $1;
            $slash = 'div';
        }

# $POSTMATCH, ${POSTMATCH}, ${^POSTMATCH} --> @{[Char::Earabic::POSTMATCH()]}
        elsif ($string =~ /\G (                 \$ \s* POSTMATCH \b | \$ \s* \{ \s* POSTMATCH \s* \} | \$ \s* \{\^POSTMATCH\} ) /oxmsgc) {
            $e_string .= q{Char::Earabic::POSTMATCH()};
            $slash = 'div';
        }

# bareword
        elsif ($string =~ /\G ( \{ \s* (?: tr | index | rindex | reverse ) \s* \} ) /oxmsgc) {
            $e_string .= $1;
            $slash = 'div';
        }

# $0 --> $0
        elsif ($string =~ /\G ( \$ 0 ) /oxmsgc) {
            $e_string .= $1;
            $slash = 'div';
        }
        elsif ($string =~ /\G ( \$ \{ \s* 0 \s* \} ) /oxmsgc) {
            $e_string .= $1;
            $slash = 'div';
        }

# $$ --> $$
        elsif ($string =~ /\G ( \$ \$ ) (?![\w\{]) /oxmsgc) {
            $e_string .= $1;
            $slash = 'div';
        }

# $1, $2, $3 --> $2, $3, $4 after s/// with multibyte anchoring
# $1, $2, $3 --> $1, $2, $3 otherwise
        elsif ($string =~ /\G \$ ([1-9][0-9]*) /oxmsgc) {
            $e_string .= e_capture($1);
            $slash = 'div';
        }
        elsif ($string =~ /\G \$ \{ \s* ([1-9][0-9]*) \s* \} /oxmsgc) {
            $e_string .= e_capture($1);
            $slash = 'div';
        }

# $$foo[ ... ] --> $ $foo->[ ... ]
        elsif ($string =~ /\G \$ ( \$ [A-Za-z_][A-Za-z0-9_]*(?: ::[A-Za-z_][A-Za-z0-9_]*)* ) ( \[ .+? \] ) /oxmsgc) {
            $e_string .= e_capture($1.'->'.$2);
            $slash = 'div';
        }

# $$foo{ ... } --> $ $foo->{ ... }
        elsif ($string =~ /\G \$ ( \$ [A-Za-z_][A-Za-z0-9_]*(?: ::[A-Za-z_][A-Za-z0-9_]*)* ) ( \{ .+? \} ) /oxmsgc) {
            $e_string .= e_capture($1.'->'.$2);
            $slash = 'div';
        }

# $$foo
        elsif ($string =~ /\G \$ ( \$ [A-Za-z_][A-Za-z0-9_]*(?: ::[A-Za-z_][A-Za-z0-9_]*)* ) /oxmsgc) {
            $e_string .= e_capture($1);
            $slash = 'div';
        }

# ${ foo }
        elsif ($string =~ /\G \$ \s* \{ ( \s* [A-Za-z_][A-Za-z0-9_]*(?: ::[A-Za-z_][A-Za-z0-9_]*)* \s* ) \} /oxmsgc) {
            $e_string .= '${' . $1 . '}';
            $slash = 'div';
        }

# ${ ... }
        elsif ($string =~ /\G \$ \s* \{ \s* ( $qq_brace ) \s* \} /oxmsgc) {
            $e_string .= e_capture($1);
            $slash = 'div';
        }

# variable or function
        #                             $ @ % & *     $ #
        elsif ($string =~ /\G ( (?: [\$\@\%\&\*] | \$\# | -> | \b sub \b) \s* (?: split | chop | index | rindex | lc | uc | fc | chr | ord | reverse | getc | tr | y | q | qq | qx | qw | m | s | qr | glob | lstat | opendir | stat | unlink | chdir ) ) \b /oxmsgc) {
            $e_string .= $1;
            $slash = 'div';
        }
        #                           $ $ $ $ $ $ $ $ $ $ $ $ $ $
        #                           $ @ # \ ' " / ? ( ) [ ] < >
        elsif ($string =~ /\G ( \$[\$\@\#\\\'\"\/\?\(\)\[\]\<\>] ) /oxmsgc) {
            $e_string .= $1;
            $slash = 'div';
        }

# subroutines of package Char::Earabic
        elsif ($string =~ /\G \b (CORE:: | ->[ ]* (?: atan2 | [a-z]{2,})) \b  /oxgc) { $e_string .= $1;               $slash = 'm//'; }
        elsif ($string =~ /\G \b bytes::substr \b                             /oxgc) { $e_string .= 'substr';         $slash = 'm//'; }
        elsif ($string =~ /\G \b chop \b                                      /oxgc) { $e_string .= 'Char::Earabic::chop';    $slash = 'm//'; }
        elsif ($string =~ /\G \b bytes::index \b                              /oxgc) { $e_string .= 'index';          $slash = 'm//'; }
        elsif ($string =~ /\G \b Char::Arabic::index \b                               /oxgc) { $e_string .= 'Char::Arabic::index';    $slash = 'm//'; }
        elsif ($string =~ /\G \b index \b                                     /oxgc) { $e_string .= 'Char::Earabic::index';   $slash = 'm//'; }
        elsif ($string =~ /\G \b bytes::rindex \b                             /oxgc) { $e_string .= 'rindex';         $slash = 'm//'; }
        elsif ($string =~ /\G \b Char::Arabic::rindex \b                              /oxgc) { $e_string .= 'Char::Arabic::rindex';   $slash = 'm//'; }
        elsif ($string =~ /\G \b rindex \b                                    /oxgc) { $e_string .= 'Char::Earabic::rindex';  $slash = 'm//'; }
        elsif ($string =~ /\G \b lc      (?= \s+[A-Za-z_]|\s*['"`\$\@\&\*\(]) /oxgc) { $e_string .= 'Char::Earabic::lc';      $slash = 'm//'; }
        elsif ($string =~ /\G \b lcfirst (?= \s+[A-Za-z_]|\s*['"`\$\@\&\*\(]) /oxgc) { $e_string .= 'Char::Earabic::lcfirst'; $slash = 'm//'; }
        elsif ($string =~ /\G \b uc      (?= \s+[A-Za-z_]|\s*['"`\$\@\&\*\(]) /oxgc) { $e_string .= 'Char::Earabic::uc';      $slash = 'm//'; }
        elsif ($string =~ /\G \b ucfirst (?= \s+[A-Za-z_]|\s*['"`\$\@\&\*\(]) /oxgc) { $e_string .= 'Char::Earabic::ucfirst'; $slash = 'm//'; }
        elsif ($string =~ /\G \b fc      (?= \s+[A-Za-z_]|\s*['"`\$\@\&\*\(]) /oxgc) { $e_string .= 'Char::Earabic::fc';      $slash = 'm//'; }

        # "-s '' ..." means file test "-s 'filename' ..." (not means "- s/// ...")
        elsif ($string =~ /\G -s                               \s+    \s* (\") ((?:$qq_char)+?)                (\") /oxgc) { $e_string .= '-s ' . e_qq('',  $1,$3,$2); $slash = 'm//'; }
        elsif ($string =~ /\G -s                               \s+ qq \s* (\#) ((?:$qq_char)+?)                (\#) /oxgc) { $e_string .= '-s ' . e_qq('qq',$1,$3,$2); $slash = 'm//'; }
        elsif ($string =~ /\G -s                               \s+ qq \s* (\() ((?:$qq_paren)+?)               (\)) /oxgc) { $e_string .= '-s ' . e_qq('qq',$1,$3,$2); $slash = 'm//'; }
        elsif ($string =~ /\G -s                               \s+ qq \s* (\{) ((?:$qq_brace)+?)               (\}) /oxgc) { $e_string .= '-s ' . e_qq('qq',$1,$3,$2); $slash = 'm//'; }
        elsif ($string =~ /\G -s                               \s+ qq \s* (\[) ((?:$qq_bracket)+?)             (\]) /oxgc) { $e_string .= '-s ' . e_qq('qq',$1,$3,$2); $slash = 'm//'; }
        elsif ($string =~ /\G -s                               \s+ qq \s* (\<) ((?:$qq_angle)+?)               (\>) /oxgc) { $e_string .= '-s ' . e_qq('qq',$1,$3,$2); $slash = 'm//'; }
        elsif ($string =~ /\G -s                               \s+ qq \s* (\S) ((?:$qq_char)+?)                (\1) /oxgc) { $e_string .= '-s ' . e_qq('qq',$1,$3,$2); $slash = 'm//'; }

        elsif ($string =~ /\G -s                               \s+    \s* (\') ((?:\\\'|\\\\|$q_char)+?)       (\') /oxgc) { $e_string .= '-s ' . e_q ('',  $1,$3,$2); $slash = 'm//'; }
        elsif ($string =~ /\G -s                               \s+ q  \s* (\#) ((?:\\\#|\\\\|$q_char)+?)       (\#) /oxgc) { $e_string .= '-s ' . e_q ('q', $1,$3,$2); $slash = 'm//'; }
        elsif ($string =~ /\G -s                               \s+ q  \s* (\() ((?:\\\)|\\\\|$q_paren)+?)      (\)) /oxgc) { $e_string .= '-s ' . e_q ('q', $1,$3,$2); $slash = 'm//'; }
        elsif ($string =~ /\G -s                               \s+ q  \s* (\{) ((?:\\\}|\\\\|$q_brace)+?)      (\}) /oxgc) { $e_string .= '-s ' . e_q ('q', $1,$3,$2); $slash = 'm//'; }
        elsif ($string =~ /\G -s                               \s+ q  \s* (\[) ((?:\\\]|\\\\|$q_bracket)+?)    (\]) /oxgc) { $e_string .= '-s ' . e_q ('q', $1,$3,$2); $slash = 'm//'; }
        elsif ($string =~ /\G -s                               \s+ q  \s* (\<) ((?:\\\>|\\\\|$q_angle)+?)      (\>) /oxgc) { $e_string .= '-s ' . e_q ('q', $1,$3,$2); $slash = 'm//'; }
        elsif ($string =~ /\G -s                               \s+ q  \s* (\S) ((?:\\\1|\\\\|$q_char)+?)       (\1) /oxgc) { $e_string .= '-s ' . e_q ('q', $1,$3,$2); $slash = 'm//'; }

        elsif ($string =~ /\G -s                               \s* (\$ \w+(?: ::\w+)* (?: (?: ->)? (?: [\$\@\%\&\*]\* | \$\#\* | \( (?:$qq_paren)*? \) | [\@\%\*]? \{ (?:$qq_brace)+? \} | [\@\%]? \[ (?:$qq_bracket)+? \] ) )*) /oxgc)
                                                                                                                           { $e_string .= "-s $1";   $slash = 'm//'; }
        elsif ($string =~ /\G -s                               \s* \( ((?:$qq_paren)*?) \)                          /oxgc) { $e_string .= "-s ($1)"; $slash = 'm//'; }
        elsif ($string =~ /\G -s                               (?= \s+ [a-z]+)                                      /oxgc) { $e_string .= '-s';      $slash = 'm//'; }
        elsif ($string =~ /\G -s                               \s+ (\w+)                                            /oxgc) { $e_string .= "-s $1";   $slash = 'm//'; }

        elsif ($string =~ /\G \b bytes::length (?= \s+[A-Za-z_]|\s*['"`\$\@\&\*\(]) /oxgc) { $e_string .= 'length';                   $slash = 'm//'; }
        elsif ($string =~ /\G \b bytes::chr    (?= \s+[A-Za-z_]|\s*['"`\$\@\&\*\(]) /oxgc) { $e_string .= 'chr';                      $slash = 'm//'; }
        elsif ($string =~ /\G \b chr           (?= \s+[A-Za-z_]|\s*['"`\$\@\&\*\(]) /oxgc) { $e_string .= 'Char::Earabic::chr';               $slash = 'm//'; }
        elsif ($string =~ /\G \b bytes::ord    (?= \s+[A-Za-z_]|\s*['"`\$\@\&\*\(]) /oxgc) { $e_string .= 'ord';                      $slash = 'div'; }
        elsif ($string =~ /\G \b ord           (?= \s+[A-Za-z_]|\s*['"`\$\@\&\*\(]) /oxgc) { $e_string .= $function_ord;              $slash = 'div'; }
        elsif ($string =~ /\G \b glob          (?= \s+[A-Za-z_]|\s*['"`\$\@\&\*\(]) /oxgc) { $e_string .= 'Char::Earabic::glob';              $slash = 'm//'; }
        elsif ($string =~ /\G \b lc \b                                              /oxgc) { $e_string .= 'Char::Earabic::lc_';               $slash = 'm//'; }
        elsif ($string =~ /\G \b lcfirst \b                                         /oxgc) { $e_string .= 'Char::Earabic::lcfirst_';          $slash = 'm//'; }
        elsif ($string =~ /\G \b uc \b                                              /oxgc) { $e_string .= 'Char::Earabic::uc_';               $slash = 'm//'; }
        elsif ($string =~ /\G \b ucfirst \b                                         /oxgc) { $e_string .= 'Char::Earabic::ucfirst_';          $slash = 'm//'; }
        elsif ($string =~ /\G \b fc \b                                              /oxgc) { $e_string .= 'Char::Earabic::fc_';               $slash = 'm//'; }
        elsif ($string =~ /\G    -s                              \b                 /oxgc) { $e_string .= '-s ';                      $slash = 'm//'; }

        elsif ($string =~ /\G \b bytes::length \b                                   /oxgc) { $e_string .= 'length';                   $slash = 'm//'; }
        elsif ($string =~ /\G \b bytes::chr \b                                      /oxgc) { $e_string .= 'chr';                      $slash = 'm//'; }
        elsif ($string =~ /\G \b chr \b                                             /oxgc) { $e_string .= 'Char::Earabic::chr_';              $slash = 'm//'; }
        elsif ($string =~ /\G \b bytes::ord \b                                      /oxgc) { $e_string .= 'ord';                      $slash = 'div'; }
        elsif ($string =~ /\G \b ord \b                                             /oxgc) { $e_string .= $function_ord_;             $slash = 'div'; }
        elsif ($string =~ /\G \b glob \b                                            /oxgc) { $e_string .= 'Char::Earabic::glob_';             $slash = 'm//'; }
        elsif ($string =~ /\G \b reverse \b                                         /oxgc) { $e_string .= $function_reverse;          $slash = 'm//'; }
        elsif ($string =~ /\G \b getc \b                                            /oxgc) { $e_string .= $function_getc;             $slash = 'm//'; }

# split
        elsif ($string =~ /\G \b (split) \b (?! \s* => ) /oxgc) {
            $slash = 'm//';

            my $e = '';
            while ($string =~ /\G ( \s+ | \( | \#.* ) /oxgc) {
                $e .= $1;
            }

# end of split
            if    ($string =~ /\G (?= [,;\)\}\]] )          /oxgc) { return 'Char::Earabic::split' . $e;                                           }

# split scalar value
            elsif ($string =~ /\G ( [\$\@\&\*] $qq_scalar ) /oxgc) { $e_string .= 'Char::Earabic::split' . $e . e_string($1);  next E_STRING_LOOP; }

# split literal space
            elsif ($string =~ /\G \b qq       (\#) [ ] (\#) /oxgc) { $e_string .= 'Char::Earabic::split' . $e . qq  {qq$1 $2}; next E_STRING_LOOP; }
            elsif ($string =~ /\G \b qq (\s*) (\() [ ] (\)) /oxgc) { $e_string .= 'Char::Earabic::split' . $e . qq{$1qq$2 $3}; next E_STRING_LOOP; }
            elsif ($string =~ /\G \b qq (\s*) (\{) [ ] (\}) /oxgc) { $e_string .= 'Char::Earabic::split' . $e . qq{$1qq$2 $3}; next E_STRING_LOOP; }
            elsif ($string =~ /\G \b qq (\s*) (\[) [ ] (\]) /oxgc) { $e_string .= 'Char::Earabic::split' . $e . qq{$1qq$2 $3}; next E_STRING_LOOP; }
            elsif ($string =~ /\G \b qq (\s*) (\<) [ ] (\>) /oxgc) { $e_string .= 'Char::Earabic::split' . $e . qq{$1qq$2 $3}; next E_STRING_LOOP; }
            elsif ($string =~ /\G \b qq (\s*) (\S) [ ] (\2) /oxgc) { $e_string .= 'Char::Earabic::split' . $e . qq{$1qq$2 $3}; next E_STRING_LOOP; }
            elsif ($string =~ /\G \b q        (\#) [ ] (\#) /oxgc) { $e_string .= 'Char::Earabic::split' . $e . qq   {q$1 $2}; next E_STRING_LOOP; }
            elsif ($string =~ /\G \b q  (\s*) (\() [ ] (\)) /oxgc) { $e_string .= 'Char::Earabic::split' . $e . qq {$1q$2 $3}; next E_STRING_LOOP; }
            elsif ($string =~ /\G \b q  (\s*) (\{) [ ] (\}) /oxgc) { $e_string .= 'Char::Earabic::split' . $e . qq {$1q$2 $3}; next E_STRING_LOOP; }
            elsif ($string =~ /\G \b q  (\s*) (\[) [ ] (\]) /oxgc) { $e_string .= 'Char::Earabic::split' . $e . qq {$1q$2 $3}; next E_STRING_LOOP; }
            elsif ($string =~ /\G \b q  (\s*) (\<) [ ] (\>) /oxgc) { $e_string .= 'Char::Earabic::split' . $e . qq {$1q$2 $3}; next E_STRING_LOOP; }
            elsif ($string =~ /\G \b q  (\s*) (\S) [ ] (\2) /oxgc) { $e_string .= 'Char::Earabic::split' . $e . qq {$1q$2 $3}; next E_STRING_LOOP; }
            elsif ($string =~ /\G                ' [ ] '    /oxgc) { $e_string .= 'Char::Earabic::split' . $e . qq     {' '};  next E_STRING_LOOP; }
            elsif ($string =~ /\G                " [ ] "    /oxgc) { $e_string .= 'Char::Earabic::split' . $e . qq     {" "};  next E_STRING_LOOP; }

# split qq//
            elsif ($string =~ /\G \b (qq) \b /oxgc) {
                if ($string =~ /\G (\#) ((?:$qq_char)*?) (\#) /oxgc)                        { $e_string .= e_split($e.'qr',$1,$3,$2,'');   next E_STRING_LOOP; } # qq# #  --> qr # #
                else {
                    while ($string !~ /\G \z/oxgc) {
                        if    ($string =~ /\G (\s+|\#.*)                             /oxgc) { $e_string .= $e . $1; }
                        elsif ($string =~ /\G (\()          ((?:$qq_paren)*?)   (\)) /oxgc) { $e_string .= e_split($e.'qr',$1,$3,$2,'');   next E_STRING_LOOP; } # qq ( ) --> qr ( )
                        elsif ($string =~ /\G (\{)          ((?:$qq_brace)*?)   (\}) /oxgc) { $e_string .= e_split($e.'qr',$1,$3,$2,'');   next E_STRING_LOOP; } # qq { } --> qr { }
                        elsif ($string =~ /\G (\[)          ((?:$qq_bracket)*?) (\]) /oxgc) { $e_string .= e_split($e.'qr',$1,$3,$2,'');   next E_STRING_LOOP; } # qq [ ] --> qr [ ]
                        elsif ($string =~ /\G (\<)          ((?:$qq_angle)*?)   (\>) /oxgc) { $e_string .= e_split($e.'qr',$1,$3,$2,'');   next E_STRING_LOOP; } # qq < > --> qr < >
                        elsif ($string =~ /\G ([*\-:?\\^|]) ((?:$qq_char)*?)    (\1) /oxgc) { $e_string .= e_split($e.'qr','{','}',$2,''); next E_STRING_LOOP; } # qq | | --> qr { }
                        elsif ($string =~ /\G (\S)          ((?:$qq_char)*?)    (\1) /oxgc) { $e_string .= e_split($e.'qr',$1,$3,$2,'');   next E_STRING_LOOP; } # qq * * --> qr * *
                    }
                    die __FILE__, ": Can't find string terminator anywhere before EOF";
                }
            }

# split qr//
            elsif ($string =~ /\G \b (qr) \b /oxgc) {
                if ($string =~ /\G (\#) ((?:$qq_char)*?) (\#) ([imosxpadlubB]*) /oxgc)                        { $e_string .= e_split  ($e.'qr',$1,$3,$2,$4);   next E_STRING_LOOP; } # qr# #
                else {
                    while ($string !~ /\G \z/oxgc) {
                        if    ($string =~ /\G (\s+|\#.*)                                               /oxgc) { $e_string .= $e . $1; }
                        elsif ($string =~ /\G (\()          ((?:$qq_paren)*?)   (\)) ([imosxpadlubB]*) /oxgc) { $e_string .= e_split  ($e.'qr',$1, $3, $2,$4); next E_STRING_LOOP; } # qr ( )
                        elsif ($string =~ /\G (\{)          ((?:$qq_brace)*?)   (\}) ([imosxpadlubB]*) /oxgc) { $e_string .= e_split  ($e.'qr',$1, $3, $2,$4); next E_STRING_LOOP; } # qr { }
                        elsif ($string =~ /\G (\[)          ((?:$qq_bracket)*?) (\]) ([imosxpadlubB]*) /oxgc) { $e_string .= e_split  ($e.'qr',$1, $3, $2,$4); next E_STRING_LOOP; } # qr [ ]
                        elsif ($string =~ /\G (\<)          ((?:$qq_angle)*?)   (\>) ([imosxpadlubB]*) /oxgc) { $e_string .= e_split  ($e.'qr',$1, $3, $2,$4); next E_STRING_LOOP; } # qr < >
                        elsif ($string =~ /\G (\')          ((?:$qq_char)*?)    (\') ([imosxpadlubB]*) /oxgc) { $e_string .= e_split_q($e.'qr',$1, $3, $2,$4); next E_STRING_LOOP; } # qr ' '
                        elsif ($string =~ /\G ([*\-:?\\^|]) ((?:$qq_char)*?)    (\1) ([imosxpadlubB]*) /oxgc) { $e_string .= e_split  ($e.'qr','{','}',$2,$4); next E_STRING_LOOP; } # qr | | --> qr { }
                        elsif ($string =~ /\G (\S)          ((?:$qq_char)*?)    (\1) ([imosxpadlubB]*) /oxgc) { $e_string .= e_split  ($e.'qr',$1, $3, $2,$4); next E_STRING_LOOP; } # qr * *
                    }
                    die __FILE__, ": Can't find string terminator anywhere before EOF";
                }
            }

# split q//
            elsif ($string =~ /\G \b (q) \b /oxgc) {
                if ($string =~ /\G (\#) ((?:\\\#|\\\\|$q_char)*?) (\#) /oxgc)                    { $e_string .= e_split_q($e.'qr',$1,$3,$2,'');   next E_STRING_LOOP; } # q# #  --> qr # #
                else {
                    while ($string !~ /\G \z/oxgc) {
                        if    ($string =~ /\G (\s+|\#.*)                                  /oxgc) { $e_string .= $e . $1; }
                        elsif ($string =~ /\G (\() ((?:\\\\|\\\)|\\\(|$q_paren)*?)   (\)) /oxgc) { $e_string .= e_split_q($e.'qr',$1,$3,$2,'');   next E_STRING_LOOP; } # q ( ) --> qr ( )
                        elsif ($string =~ /\G (\{) ((?:\\\\|\\\}|\\\{|$q_brace)*?)   (\}) /oxgc) { $e_string .= e_split_q($e.'qr',$1,$3,$2,'');   next E_STRING_LOOP; } # q { } --> qr { }
                        elsif ($string =~ /\G (\[) ((?:\\\\|\\\]|\\\[|$q_bracket)*?) (\]) /oxgc) { $e_string .= e_split_q($e.'qr',$1,$3,$2,'');   next E_STRING_LOOP; } # q [ ] --> qr [ ]
                        elsif ($string =~ /\G (\<) ((?:\\\\|\\\>|\\\<|$q_angle)*?)   (\>) /oxgc) { $e_string .= e_split_q($e.'qr',$1,$3,$2,'');   next E_STRING_LOOP; } # q < > --> qr < >
                        elsif ($string =~ /\G ([*\-:?\\^|])       ((?:$q_char)*?)    (\1) /oxgc) { $e_string .= e_split_q($e.'qr','{','}',$2,''); next E_STRING_LOOP; } # q | | --> qr { }
                        elsif ($string =~ /\G (\S) ((?:\\\\|\\\1|     $q_char)*?)    (\1) /oxgc) { $e_string .= e_split_q($e.'qr',$1,$3,$2,'');   next E_STRING_LOOP; } # q * * --> qr * *
                    }
                    die __FILE__, ": Can't find string terminator anywhere before EOF";
                }
            }

# split m//
            elsif ($string =~ /\G \b (m) \b /oxgc) {
                if ($string =~ /\G (\#) ((?:$qq_char)*?) (\#) ([cgimosxpadlubB]*) /oxgc)                        { $e_string .= e_split  ($e.'qr',$1,$3,$2,$4);   next E_STRING_LOOP; } # m# #  --> qr # #
                else {
                    while ($string !~ /\G \z/oxgc) {
                        if    ($string =~ /\G (\s+|\#.*)                                                 /oxgc) { $e_string .= $e . $1; }
                        elsif ($string =~ /\G (\()          ((?:$qq_paren)*?)   (\)) ([cgimosxpadlubB]*) /oxgc) { $e_string .= e_split  ($e.'qr',$1, $3, $2,$4); next E_STRING_LOOP; } # m ( ) --> qr ( )
                        elsif ($string =~ /\G (\{)          ((?:$qq_brace)*?)   (\}) ([cgimosxpadlubB]*) /oxgc) { $e_string .= e_split  ($e.'qr',$1, $3, $2,$4); next E_STRING_LOOP; } # m { } --> qr { }
                        elsif ($string =~ /\G (\[)          ((?:$qq_bracket)*?) (\]) ([cgimosxpadlubB]*) /oxgc) { $e_string .= e_split  ($e.'qr',$1, $3, $2,$4); next E_STRING_LOOP; } # m [ ] --> qr [ ]
                        elsif ($string =~ /\G (\<)          ((?:$qq_angle)*?)   (\>) ([cgimosxpadlubB]*) /oxgc) { $e_string .= e_split  ($e.'qr',$1, $3, $2,$4); next E_STRING_LOOP; } # m < > --> qr < >
                        elsif ($string =~ /\G (\')          ((?:$qq_char)*?)    (\') ([cgimosxpadlubB]*) /oxgc) { $e_string .= e_split_q($e.'qr',$1, $3, $2,$4); next E_STRING_LOOP; } # m ' ' --> qr ' '
                        elsif ($string =~ /\G ([*\-:?\\^|]) ((?:$qq_char)*?)    (\1) ([cgimosxpadlubB]*) /oxgc) { $e_string .= e_split  ($e.'qr','{','}',$2,$4); next E_STRING_LOOP; } # m | | --> qr { }
                        elsif ($string =~ /\G (\S)          ((?:$qq_char)*?)    (\1) ([cgimosxpadlubB]*) /oxgc) { $e_string .= e_split  ($e.'qr',$1, $3, $2,$4); next E_STRING_LOOP; } # m * * --> qr * *
                    }
                    die __FILE__, ": Search pattern not terminated";
                }
            }

# split ''
            elsif ($string =~ /\G (\') /oxgc) {
                my $q_string = '';
                while ($string !~ /\G \z/oxgc) {
                    if    ($string =~ /\G (\\\\)    /oxgc) { $q_string .= $1; }
                    elsif ($string =~ /\G (\\\')    /oxgc) { $q_string .= $1; } # splitqr'' --> split qr''
                    elsif ($string =~ /\G \'        /oxgc)                      { $e_string .= e_split_q($e.q{ qr},"'","'",$q_string,''); next E_STRING_LOOP; } # ' ' --> qr ' '
                    elsif ($string =~ /\G ($q_char) /oxgc) { $q_string .= $1; }
                }
                die __FILE__, ": Can't find string terminator anywhere before EOF";
            }

# split ""
            elsif ($string =~ /\G (\") /oxgc) {
                my $qq_string = '';
                while ($string !~ /\G \z/oxgc) {
                    if    ($string =~ /\G (\\\\)    /oxgc) { $qq_string .= $1; }
                    elsif ($string =~ /\G (\\\")    /oxgc) { $qq_string .= $1; } # splitqr"" --> split qr""
                    elsif ($string =~ /\G \"        /oxgc)                       { $e_string .= e_split($e.q{ qr},'"','"',$qq_string,''); next E_STRING_LOOP; } # " " --> qr " "
                    elsif ($string =~ /\G ($q_char) /oxgc) { $qq_string .= $1; }
                }
                die __FILE__, ": Can't find string terminator anywhere before EOF";
            }

# split //
            elsif ($string =~ /\G (\/) /oxgc) {
                my $regexp = '';
                while ($string !~ /\G \z/oxgc) {
                    if    ($string =~ /\G (\\\\)                 /oxgc) { $regexp .= $1; }
                    elsif ($string =~ /\G (\\\/)                 /oxgc) { $regexp .= $1; } # splitqr// --> split qr//
                    elsif ($string =~ /\G \/ ([cgimosxpadlubB]*) /oxgc)                    { $e_string .= e_split($e.q{ qr}, '/','/',$regexp,$1); next E_STRING_LOOP; } # / / --> qr / /
                    elsif ($string =~ /\G ($q_char)              /oxgc) { $regexp .= $1; }
                }
                die __FILE__, ": Search pattern not terminated";
            }
        }

# qq//
        elsif ($string =~ /\G \b (qq) \b /oxgc) {
            my $ope = $1;
            if ($string =~ /\G (\#) ((?:$qq_char)*?) (\#) /oxgc) { # qq# #
                $e_string .= e_qq($ope,$1,$3,$2);
            }
            else {
                my $e = '';
                while ($string !~ /\G \z/oxgc) {
                    if    ($string =~ /\G (\s+|\#.*)                    /oxgc) { $e .= $1; }
                    elsif ($string =~ /\G (\() ((?:$qq_paren)*?)   (\)) /oxgc) { $e_string .= $e . e_qq($ope,$1,$3,$2); next E_STRING_LOOP; } # qq ( )
                    elsif ($string =~ /\G (\{) ((?:$qq_brace)*?)   (\}) /oxgc) { $e_string .= $e . e_qq($ope,$1,$3,$2); next E_STRING_LOOP; } # qq { }
                    elsif ($string =~ /\G (\[) ((?:$qq_bracket)*?) (\]) /oxgc) { $e_string .= $e . e_qq($ope,$1,$3,$2); next E_STRING_LOOP; } # qq [ ]
                    elsif ($string =~ /\G (\<) ((?:$qq_angle)*?)   (\>) /oxgc) { $e_string .= $e . e_qq($ope,$1,$3,$2); next E_STRING_LOOP; } # qq < >
                    elsif ($string =~ /\G (\S) ((?:$qq_char)*?)    (\1) /oxgc) { $e_string .= $e . e_qq($ope,$1,$3,$2); next E_STRING_LOOP; } # qq * *
                }
                die __FILE__, ": Can't find string terminator anywhere before EOF";
            }
        }

# qx//
        elsif ($string =~ /\G \b (qx) \b /oxgc) {
            my $ope = $1;
            if ($string =~ /\G (\#) ((?:$qq_char)*?) (\#) /oxgc) { # qx# #
                $e_string .= e_qq($ope,$1,$3,$2);
            }
            else {
                my $e = '';
                while ($string !~ /\G \z/oxgc) {
                    if    ($string =~ /\G (\s+|\#.*)                    /oxgc) { $e .= $1; }
                    elsif ($string =~ /\G (\() ((?:$qq_paren)*?)   (\)) /oxgc) { $e_string .= $e . e_qq($ope,$1,$3,$2); next E_STRING_LOOP; } # qx ( )
                    elsif ($string =~ /\G (\{) ((?:$qq_brace)*?)   (\}) /oxgc) { $e_string .= $e . e_qq($ope,$1,$3,$2); next E_STRING_LOOP; } # qx { }
                    elsif ($string =~ /\G (\[) ((?:$qq_bracket)*?) (\]) /oxgc) { $e_string .= $e . e_qq($ope,$1,$3,$2); next E_STRING_LOOP; } # qx [ ]
                    elsif ($string =~ /\G (\<) ((?:$qq_angle)*?)   (\>) /oxgc) { $e_string .= $e . e_qq($ope,$1,$3,$2); next E_STRING_LOOP; } # qx < >
                    elsif ($string =~ /\G (\') ((?:$qq_char)*?)    (\') /oxgc) { $e_string .= $e . e_q ($ope,$1,$3,$2); next E_STRING_LOOP; } # qx ' '
                    elsif ($string =~ /\G (\S) ((?:$qq_char)*?)    (\1) /oxgc) { $e_string .= $e . e_qq($ope,$1,$3,$2); next E_STRING_LOOP; } # qx * *
                }
                die __FILE__, ": Can't find string terminator anywhere before EOF";
            }
        }

# q//
        elsif ($string =~ /\G \b (q) \b /oxgc) {
            my $ope = $1;
            if ($string =~ /\G (\#) ((?:\\\#|\\\\|$q_char)*?) (\#) /oxgc) { # q# #
                $e_string .= e_q($ope,$1,$3,$2);
            }
            else {
                my $e = '';
                while ($string !~ /\G \z/oxgc) {
                    if    ($string =~ /\G (\s+|\#.*)                                  /oxgc) { $e .= $1; }
                    elsif ($string =~ /\G (\() ((?:\\\\|\\\)|\\\(|$q_paren)*?)   (\)) /oxgc) { $e_string .= $e . e_q($ope,$1,$3,$2); next E_STRING_LOOP; } # q ( )
                    elsif ($string =~ /\G (\{) ((?:\\\\|\\\}|\\\{|$q_brace)*?)   (\}) /oxgc) { $e_string .= $e . e_q($ope,$1,$3,$2); next E_STRING_LOOP; } # q { }
                    elsif ($string =~ /\G (\[) ((?:\\\\|\\\]|\\\[|$q_bracket)*?) (\]) /oxgc) { $e_string .= $e . e_q($ope,$1,$3,$2); next E_STRING_LOOP; } # q [ ]
                    elsif ($string =~ /\G (\<) ((?:\\\\|\\\>|\\\<|$q_angle)*?)   (\>) /oxgc) { $e_string .= $e . e_q($ope,$1,$3,$2); next E_STRING_LOOP; } # q < >
                    elsif ($string =~ /\G (\S) ((?:\\\\|\\\1|     $q_char)*?)    (\1) /oxgc) { $e_string .= $e . e_q($ope,$1,$3,$2); next E_STRING_LOOP; } # q * *
                }
                die __FILE__, ": Can't find string terminator anywhere before EOF";
            }
        }

# ''
        elsif ($string =~ /\G (?

# ""
        elsif ($string =~ /\G (\") ((?:$qq_char)*?) (\")                              /oxgc) { $e_string .= e_qq('',$1,$3,$2); }

# ``
        elsif ($string =~ /\G (\`) ((?:$qq_char)*?) (\`)                              /oxgc) { $e_string .= e_qq('',$1,$3,$2); }

# <<= <=> <= < operator
        elsif ($string =~ /\G ( <<= | <=> | <= | < ) (?= \s* [A-Za-z_0-9'"`\$\@\&\*\(\+\-] )/oxgc)
                                                                                             { $e_string .= $1;                }

#
        elsif ($string =~ /\G (<[\$]?[A-Za-z_][A-Za-z_0-9]*>)                         /oxgc) { $e_string .= $1;                }

#    --- glob
        elsif ($string =~ /\G < ((?:$q_char)+?) > /oxgc) {
            $e_string .= 'Char::Earabic::glob("' . $1 . '")';
        }

# << (bit shift)   --- not here document
        elsif ($string =~ /\G ( << \s* ) (?= [0-9\$\@\&] ) /oxgc) { $slash = 'm//'; $e_string .= $1; }

# <<'HEREDOC'
        elsif ($string =~ /\G ( << '([a-zA-Z_0-9]*)' ) /oxgc) {
            $slash = 'm//';
            my $here_quote = $1;
            my $delimiter  = $2;

            # get here document
            if ($here_script eq '') {
                $here_script = CORE::substr $_, pos $_;
                $here_script =~ s/.*?\n//oxm;
            }
            if ($here_script =~ s/\A (.*?) \n $delimiter \n //xms) {
                push @heredoc, $1 . qq{\n$delimiter\n};
                push @heredoc_delimiter, $delimiter;
            }
            else {
                die __FILE__, ": Can't find string terminator $delimiter anywhere before EOF";
            }
            $e_string .= $here_quote;
        }

# <<\HEREDOC
        elsif ($string =~ /\G ( << \\([a-zA-Z_0-9]+) ) /oxgc) {
            $slash = 'm//';
            my $here_quote = $1;
            my $delimiter  = $2;

            # get here document
            if ($here_script eq '') {
                $here_script = CORE::substr $_, pos $_;
                $here_script =~ s/.*?\n//oxm;
            }
            if ($here_script =~ s/\A (.*?) \n $delimiter \n //xms) {
                push @heredoc, $1 . qq{\n$delimiter\n};
                push @heredoc_delimiter, $delimiter;
            }
            else {
                die __FILE__, ": Can't find string terminator $delimiter anywhere before EOF";
            }
            $e_string .= $here_quote;
        }

# <<"HEREDOC"
        elsif ($string =~ /\G ( << "([a-zA-Z_0-9]*)" ) /oxgc) {
            $slash = 'm//';
            my $here_quote = $1;
            my $delimiter  = $2;

            # get here document
            if ($here_script eq '') {
                $here_script = CORE::substr $_, pos $_;
                $here_script =~ s/.*?\n//oxm;
            }
            if ($here_script =~ s/\A (.*?) \n $delimiter \n //xms) {
                push @heredoc, e_heredoc($1) . qq{\n$delimiter\n};
                push @heredoc_delimiter, $delimiter;
            }
            else {
                die __FILE__, ": Can't find string terminator $delimiter anywhere before EOF";
            }
            $e_string .= $here_quote;
        }

# <
        elsif ($string =~ /\G ( << ([a-zA-Z_0-9]+) ) /oxgc) {
            $slash = 'm//';
            my $here_quote = $1;
            my $delimiter  = $2;

            # get here document
            if ($here_script eq '') {
                $here_script = CORE::substr $_, pos $_;
                $here_script =~ s/.*?\n//oxm;
            }
            if ($here_script =~ s/\A (.*?) \n $delimiter \n //xms) {
                push @heredoc, e_heredoc($1) . qq{\n$delimiter\n};
                push @heredoc_delimiter, $delimiter;
            }
            else {
                die __FILE__, ": Can't find string terminator $delimiter anywhere before EOF";
            }
            $e_string .= $here_quote;
        }

# <<`HEREDOC`
        elsif ($string =~ /\G ( << `([a-zA-Z_0-9]*)` ) /oxgc) {
            $slash = 'm//';
            my $here_quote = $1;
            my $delimiter  = $2;

            # get here document
            if ($here_script eq '') {
                $here_script = CORE::substr $_, pos $_;
                $here_script =~ s/.*?\n//oxm;
            }
            if ($here_script =~ s/\A (.*?) \n $delimiter \n //xms) {
                push @heredoc, e_heredoc($1) . qq{\n$delimiter\n};
                push @heredoc_delimiter, $delimiter;
            }
            else {
                die __FILE__, ": Can't find string terminator $delimiter anywhere before EOF";
            }
            $e_string .= $here_quote;
        }

        # any operator before div
        elsif ($string =~ /\G (
            -- | \+\+ |
            [\)\}\]]

            ) /oxgc) { $slash = 'div'; $e_string .= $1; }

        # any operator before m//
        elsif ($string =~ /\G (

            !~~ | !~ | != | ! |
            %= | % |
            &&= | && | &= | & |
            -= | -> | - |
            :\s*= |
            : |
            <<= | <=> | <= | < |
            == | => | =~ | = |
            >>= | >> | >= | > |
            \*\*= | \*\* | \*= | \* |
            \+= | \+ |
            \.\.\. | \.\. | \.= | \. |
            \/\/= | \/\/ |
            \/= | \/ |
            \? |
            \\ |
            \^= | \^ |
            \b x= |
            \|\|= | \|\| | \|= | \| |
            ~~ | ~ |
            \b(?: and | cmp | eq | ge | gt | le | lt | ne | not | or | xor | x )\b |
            \b(?: print )\b |

            [,;\(\{\[]

            ) /oxgc) { $slash = 'm//'; $e_string .= $1; }

        # other any character
        elsif ($string =~ /\G ($q_char) /oxgc) { $e_string .= $1; }

        # system error
        else {
            die __FILE__, ": Oops, this shouldn't happen!";
        }
    }

    return $e_string;
}

#
# character class
#
sub character_class {
    my($char,$modifier) = @_;

    if ($char eq '.') {
        if ($modifier =~ /s/) {
            return '${Char::Earabic::dot_s}';
        }
        else {
            return '${Char::Earabic::dot}';
        }
    }
    else {
        return Char::Earabic::classic_character_class($char);
    }
}

#
# escape capture ($1, $2, $3, ...)
#
sub e_capture {

    return join '', '${',                $_[0],  '}';
}

#
# escape transliteration (tr/// or y///)
#
sub e_tr {
    my($variable,$charclass,$e,$charclass2,$modifier) = @_;
    my $e_tr = '';
    $modifier ||= '';

    $slash = 'div';

    # quote character class 1
    $charclass  = q_tr($charclass);

    # quote character class 2
    $charclass2 = q_tr($charclass2);

    # /b /B modifier
    if ($modifier =~ tr/bB//d) {
        if ($variable eq '') {
            $e_tr = qq{tr$charclass$e$charclass2$modifier};
        }
        else {
            $e_tr = qq{$variable${bind_operator}tr$charclass$e$charclass2$modifier};
        }
    }
    else {
        if ($variable eq '') {
            $e_tr = qq{Char::Earabic::tr(\$_,' =~ ',$charclass,$e$charclass2,'$modifier')};
        }
        else {
            $e_tr = qq{Char::Earabic::tr($variable,'$bind_operator',$charclass,$e$charclass2,'$modifier')};
        }
    }

    # clear tr/// variable
    $tr_variable = '';
    $bind_operator = '';

    return $e_tr;
}

#
# quote for escape transliteration (tr/// or y///)
#
sub q_tr {
    my($charclass) = @_;

    # quote character class
    if ($charclass !~ /'/oxms) {
        return e_q('',  "'", "'", $charclass); # --> q' '
    }
    elsif ($charclass !~ /\//oxms) {
        return e_q('q',  '/', '/', $charclass); # --> q/ /
    }
    elsif ($charclass !~ /\#/oxms) {
        return e_q('q',  '#', '#', $charclass); # --> q# #
    }
    elsif ($charclass !~ /[\<\>]/oxms) {
        return e_q('q', '<', '>', $charclass); # --> q< >
    }
    elsif ($charclass !~ /[\(\)]/oxms) {
        return e_q('q', '(', ')', $charclass); # --> q( )
    }
    elsif ($charclass !~ /[\{\}]/oxms) {
        return e_q('q', '{', '}', $charclass); # --> q{ }
    }
    else {
        for my $char (qw( ! " $ % & * + . : = ? @ ^ ` | ~ ), "\x00".."\x1F", "\x7F", "\xFF") {
            if ($charclass !~ /\Q$char\E/xms) {
                return e_q('q', $char, $char, $charclass);
            }
        }
    }

    return e_q('q', '{', '}', $charclass);
}

#
# escape q string (q//, '')
#
sub e_q {
    my($ope,$delimiter,$end_delimiter,$string) = @_;

    $slash = 'div';

    return join '', $ope, $delimiter, $string, $end_delimiter;
}

#
# escape qq string (qq//, "", qx//, ``)
#
sub e_qq {
    my($ope,$delimiter,$end_delimiter,$string) = @_;

    $slash = 'div';

    my $left_e  = 0;
    my $right_e = 0;
    my @char = $string =~ /\G(
        \\o\{ [0-7]+          \}   |
        \\x\{ [0-9A-Fa-f]+    \}   |
        \\N\{ [^0-9\}][^\}]* \} |
        \$` | \$\{`\} | \$ \s* PREMATCH  | \$ \s* \{ \s* PREMATCH  \s* \} | \$ \s* \{\^PREMATCH\}  |
        \$& | \$\{&\} | \$ \s* MATCH     | \$ \s* \{ \s* MATCH     \s* \} | \$ \s* \{\^MATCH\}     |
                        \$ \s* POSTMATCH | \$ \s* \{ \s* POSTMATCH \s* \} | \$ \s* \{\^POSTMATCH\} |
        \$ \s* \d+                 |
        \$ \s* \{ \s* \d+ \s* \}   |
        \$ \$ (?![\w\{])           |
        \$ \s* \$ \s* $qq_variable |
        \\?(?:$q_char)
    )/oxmsg;

    for (my $i=0; $i <= $#char; $i++) {

        # "\L\u" --> "\u\L"
        if (($char[$i] eq '\L') and ($char[$i+1] eq '\u')) {
            @char[$i,$i+1] = @char[$i+1,$i];
        }

        # "\U\l" --> "\l\U"
        elsif (($char[$i] eq '\U') and ($char[$i+1] eq '\l')) {
            @char[$i,$i+1] = @char[$i+1,$i];
        }

        # octal escape sequence
        elsif ($char[$i] =~ /\A \\o \{ ([0-7]+) \} \z/oxms) {
            $char[$i] = Char::Earabic::octchr($1);
        }

        # hexadecimal escape sequence
        elsif ($char[$i] =~ /\A \\x \{ ([0-9A-Fa-f]+) \} \z/oxms) {
            $char[$i] = Char::Earabic::hexchr($1);
        }

        # \N{CHARNAME} --> N{CHARNAME}
        elsif ($char[$i] =~ /\A \\ ( N\{ ([^0-9\}][^\}]*) \} ) \z/oxms) {
            $char[$i] = $1;
        }

        if (0) {
        }

        # \F
        #
        # P.69 Table 2-6. Translation escapes
        # in Chapter 2: Bits and Pieces
        # of ISBN 978-0-596-00492-7 Programming Perl 4th Edition.
        # (and so on)

        # \u \l \U \L \F \Q \E
        elsif ($char[$i] =~ /\A ([<>]) \z/oxms) {
            if ($right_e < $left_e) {
                $char[$i] = '\\' . $char[$i];
            }
        }
        elsif ($char[$i] eq '\u') {

            # "STRING @{[ LIST EXPR ]} MORE STRING"

            # P.257 Other Tricks You Can Do with Hard References
            # in Chapter 8: References
            # of ISBN 0-596-00027-8 Programming Perl Third Edition.

            # P.353 Other Tricks You Can Do with Hard References
            # in Chapter 8: References
            # of ISBN 978-0-596-00492-7 Programming Perl 4th Edition.

            # (and so on)

            $char[$i] = '@{[Char::Earabic::ucfirst qq<';
            $left_e++;
        }
        elsif ($char[$i] eq '\l') {
            $char[$i] = '@{[Char::Earabic::lcfirst qq<';
            $left_e++;
        }
        elsif ($char[$i] eq '\U') {
            $char[$i] = '@{[Char::Earabic::uc qq<';
            $left_e++;
        }
        elsif ($char[$i] eq '\L') {
            $char[$i] = '@{[Char::Earabic::lc qq<';
            $left_e++;
        }
        elsif ($char[$i] eq '\F') {
            $char[$i] = '@{[Char::Earabic::fc qq<';
            $left_e++;
        }
        elsif ($char[$i] eq '\Q') {
            $char[$i] = '@{[CORE::quotemeta qq<';
            $left_e++;
        }
        elsif ($char[$i] eq '\E') {
            if ($right_e < $left_e) {
                $char[$i] = '>]}';
                $right_e++;
            }
            else {
                $char[$i] = '';
            }
        }
        elsif ($char[$i] eq '\Q') {
            while (1) {
                if (++$i > $#char) {
                    last;
                }
                if ($char[$i] eq '\E') {
                    last;
                }
            }
        }
        elsif ($char[$i] eq '\E') {
        }

        # $0 --> $0
        elsif ($char[$i] =~ /\A \$ 0 \z/oxms) {
        }
        elsif ($char[$i] =~ /\A \$ \{ \s* 0 \s* \} \z/oxms) {
        }

        # $$ --> $$
        elsif ($char[$i] =~ /\A \$\$ \z/oxms) {
        }

        # $1, $2, $3 --> $2, $3, $4 after s/// with multibyte anchoring
        # $1, $2, $3 --> $1, $2, $3 otherwise
        elsif ($char[$i] =~ /\A \$ ([1-9][0-9]*) \z/oxms) {
            $char[$i] = e_capture($1);
        }
        elsif ($char[$i] =~ /\A \$ \{ \s* ([1-9][0-9]*) \s* \} \z/oxms) {
            $char[$i] = e_capture($1);
        }

        # $$foo[ ... ] --> $ $foo->[ ... ]
        elsif ($char[$i] =~ /\A \$ ( \$ [A-Za-z_][A-Za-z0-9_]*(?: ::[A-Za-z_][A-Za-z0-9_]*)* ) ( \[ (?:$qq_bracket)*? \] ) \z/oxms) {
            $char[$i] = e_capture($1.'->'.$2);
        }

        # $$foo{ ... } --> $ $foo->{ ... }
        elsif ($char[$i] =~ /\A \$ ( \$ [A-Za-z_][A-Za-z0-9_]*(?: ::[A-Za-z_][A-Za-z0-9_]*)* ) ( \{ (?:$qq_brace)*? \} ) \z/oxms) {
            $char[$i] = e_capture($1.'->'.$2);
        }

        # $$foo
        elsif ($char[$i] =~ /\A \$ ( \$ [A-Za-z_][A-Za-z0-9_]*(?: ::[A-Za-z_][A-Za-z0-9_]*)* ) \z/oxms) {
            $char[$i] = e_capture($1);
        }

        # $`, ${`}, $PREMATCH, ${PREMATCH}, ${^PREMATCH} --> Char::Earabic::PREMATCH()
        elsif ($char[$i] =~ /\A ( \$` | \$\{`\} | \$ \s* PREMATCH  | \$ \s* \{ \s* PREMATCH \s* \}  | \$ \s* \{\^PREMATCH\}  ) \z/oxmsgc) {
            $char[$i] = '@{[Char::Earabic::PREMATCH()]}';
        }

        # $&, ${&}, $MATCH, ${MATCH}, ${^MATCH} --> Char::Earabic::MATCH()
        elsif ($char[$i] =~ /\A ( \$& | \$\{&\} | \$ \s* MATCH     | \$ \s* \{ \s* MATCH     \s* \} | \$ \s* \{\^MATCH\}     ) \z/oxmsgc) {
            $char[$i] = '@{[Char::Earabic::MATCH()]}';
        }

        # $POSTMATCH, ${POSTMATCH}, ${^POSTMATCH} --> Char::Earabic::POSTMATCH()
        elsif ($char[$i] =~ /\A (                 \$ \s* POSTMATCH | \$ \s* \{ \s* POSTMATCH \s* \} | \$ \s* \{\^POSTMATCH\} ) \z/oxmsgc) {
            $char[$i] = '@{[Char::Earabic::POSTMATCH()]}';
        }

        # ${ foo } --> ${ foo }
        elsif ($char[$i] =~ /\A \$ \s* \{ \s* [A-Za-z_][A-Za-z0-9_]*(?: ::[A-Za-z_][A-Za-z0-9_]*)* \s* \} \z/oxms) {
        }

        # ${ ... }
        elsif ($char[$i] =~ /\A \$ \s* \{ \s* ( .+ ) \s* \} \z/oxms) {
            $char[$i] = e_capture($1);
        }
    }

    # return string
    if ($left_e > $right_e) {
        return join '', $ope, $delimiter, @char, '>]}' x ($left_e - $right_e), $end_delimiter;
    }
    return     join '', $ope, $delimiter, @char,                               $end_delimiter;
}

#
# escape qw string (qw//)
#
sub e_qw {
    my($ope,$delimiter,$end_delimiter,$string) = @_;

    $slash = 'div';

    # choice again delimiter
    my %octet = map {$_ => 1} ($string =~ /\G ([\x00-\xFF]) /oxmsg);
    if (not $octet{$end_delimiter}) {
        return join '', $ope, $delimiter, $string, $end_delimiter;
    }
    elsif (not $octet{')'}) {
        return join '', $ope, '(',        $string, ')';
    }
    elsif (not $octet{'}'}) {
        return join '', $ope, '{',        $string, '}';
    }
    elsif (not $octet{']'}) {
        return join '', $ope, '[',        $string, ']';
    }
    elsif (not $octet{'>'}) {
        return join '', $ope, '<',        $string, '>';
    }
    else {
        for my $char (qw( ! " $ % & * + - . / : = ? @ ^ ` | ~ ), "\x00".."\x1F", "\x7F", "\xFF") {
            if (not $octet{$char}) {
                return join '', $ope,      $char, $string, $char;
            }
        }
    }

    # qw/AAA BBB C'CC/ --> ('AAA', 'BBB', 'C\'CC')
    my @string = CORE::split(/\s+/, $string);
    for my $string (@string) {
        my @octet = $string =~ /\G ([\x00-\xFF]) /oxmsg;
        for my $octet (@octet) {
            if ($octet =~ /\A (['\\]) \z/oxms) {
                $octet = '\\' . $1;
            }
        }
        $string = join '', @octet;
    }
    return join '', '(', (join ', ', map { "'$_'" } @string), ')';
}

#
# escape here document (<<"HEREDOC", <
#
sub e_heredoc {
    my($string) = @_;

    $slash = 'm//';

    my $metachar = qr/[\@\\|]/oxms; # '|' is for <<`HEREDOC`

    my $left_e  = 0;
    my $right_e = 0;
    my @char = $string =~ /\G(
        \\o\{ [0-7]+          \}   |
        \\x\{ [0-9A-Fa-f]+    \}   |
        \\N\{ [^0-9\}][^\}]* \} |
        \$` | \$\{`\} | \$ \s* PREMATCH  | \$ \s* \{ \s* PREMATCH  \s* \} | \$ \s* \{\^PREMATCH\}  |
        \$& | \$\{&\} | \$ \s* MATCH     | \$ \s* \{ \s* MATCH     \s* \} | \$ \s* \{\^MATCH\}     |
                        \$ \s* POSTMATCH | \$ \s* \{ \s* POSTMATCH \s* \} | \$ \s* \{\^POSTMATCH\} |
        \$ \s* \d+                 |
        \$ \s* \{ \s* \d+ \s* \}   |
        \$ \$ (?![\w\{])           |
        \$ \s* \$ \s* $qq_variable |
        \\?(?:$q_char)
    )/oxmsg;

    for (my $i=0; $i <= $#char; $i++) {

        # "\L\u" --> "\u\L"
        if (($char[$i] eq '\L') and ($char[$i+1] eq '\u')) {
            @char[$i,$i+1] = @char[$i+1,$i];
        }

        # "\U\l" --> "\l\U"
        elsif (($char[$i] eq '\U') and ($char[$i+1] eq '\l')) {
            @char[$i,$i+1] = @char[$i+1,$i];
        }

        # octal escape sequence
        elsif ($char[$i] =~ /\A \\o \{ ([0-7]+) \} \z/oxms) {
            $char[$i] = Char::Earabic::octchr($1);
        }

        # hexadecimal escape sequence
        elsif ($char[$i] =~ /\A \\x \{ ([0-9A-Fa-f]+) \} \z/oxms) {
            $char[$i] = Char::Earabic::hexchr($1);
        }

        # \N{CHARNAME} --> N{CHARNAME}
        elsif ($char[$i] =~ /\A \\ ( N\{ ([^0-9\}][^\}]*) \} ) \z/oxms) {
            $char[$i] = $1;
        }

        if (0) {
        }

        # \u \l \U \L \F \Q \E
        elsif ($char[$i] =~ /\A ([<>]) \z/oxms) {
            if ($right_e < $left_e) {
                $char[$i] = '\\' . $char[$i];
            }
        }
        elsif ($char[$i] eq '\u') {
            $char[$i] = '@{[Char::Earabic::ucfirst qq<';
            $left_e++;
        }
        elsif ($char[$i] eq '\l') {
            $char[$i] = '@{[Char::Earabic::lcfirst qq<';
            $left_e++;
        }
        elsif ($char[$i] eq '\U') {
            $char[$i] = '@{[Char::Earabic::uc qq<';
            $left_e++;
        }
        elsif ($char[$i] eq '\L') {
            $char[$i] = '@{[Char::Earabic::lc qq<';
            $left_e++;
        }
        elsif ($char[$i] eq '\F') {
            $char[$i] = '@{[Char::Earabic::fc qq<';
            $left_e++;
        }
        elsif ($char[$i] eq '\Q') {
            $char[$i] = '@{[CORE::quotemeta qq<';
            $left_e++;
        }
        elsif ($char[$i] eq '\E') {
            if ($right_e < $left_e) {
                $char[$i] = '>]}';
                $right_e++;
            }
            else {
                $char[$i] = '';
            }
        }
        elsif ($char[$i] eq '\Q') {
            while (1) {
                if (++$i > $#char) {
                    last;
                }
                if ($char[$i] eq '\E') {
                    last;
                }
            }
        }
        elsif ($char[$i] eq '\E') {
        }

        # $0 --> $0
        elsif ($char[$i] =~ /\A \$ 0 \z/oxms) {
        }
        elsif ($char[$i] =~ /\A \$ \{ \s* 0 \s* \} \z/oxms) {
        }

        # $$ --> $$
        elsif ($char[$i] =~ /\A \$\$ \z/oxms) {
        }

        # $1, $2, $3 --> $2, $3, $4 after s/// with multibyte anchoring
        # $1, $2, $3 --> $1, $2, $3 otherwise
        elsif ($char[$i] =~ /\A \$ ([1-9][0-9]*) \z/oxms) {
            $char[$i] = e_capture($1);
        }
        elsif ($char[$i] =~ /\A \$ \{ \s* ([1-9][0-9]*) \s* \} \z/oxms) {
            $char[$i] = e_capture($1);
        }

        # $$foo[ ... ] --> $ $foo->[ ... ]
        elsif ($char[$i] =~ /\A \$ ( \$ [A-Za-z_][A-Za-z0-9_]*(?: ::[A-Za-z_][A-Za-z0-9_]*)* ) ( \[ (?:$qq_bracket)*? \] ) \z/oxms) {
            $char[$i] = e_capture($1.'->'.$2);
        }

        # $$foo{ ... } --> $ $foo->{ ... }
        elsif ($char[$i] =~ /\A \$ ( \$ [A-Za-z_][A-Za-z0-9_]*(?: ::[A-Za-z_][A-Za-z0-9_]*)* ) ( \{ (?:$qq_brace)*? \} ) \z/oxms) {
            $char[$i] = e_capture($1.'->'.$2);
        }

        # $$foo
        elsif ($char[$i] =~ /\A \$ ( \$ [A-Za-z_][A-Za-z0-9_]*(?: ::[A-Za-z_][A-Za-z0-9_]*)* ) \z/oxms) {
            $char[$i] = e_capture($1);
        }

        # $`, ${`}, $PREMATCH, ${PREMATCH}, ${^PREMATCH} --> Char::Earabic::PREMATCH()
        elsif ($char[$i] =~ /\A ( \$` | \$\{`\} | \$ \s* PREMATCH  | \$ \s* \{ \s* PREMATCH  \s* \} | \$ \s* \{\^PREMATCH\}  ) \z/oxmsgc) {
            $char[$i] = '@{[Char::Earabic::PREMATCH()]}';
        }

        # $&, ${&}, $MATCH, ${MATCH}, ${^MATCH} --> Char::Earabic::MATCH()
        elsif ($char[$i] =~ /\A ( \$& | \$\{&\} | \$ \s* MATCH     | \$ \s* \{ \s* MATCH     \s* \} | \$ \s* \{\^MATCH\}     ) \z/oxmsgc) {
            $char[$i] = '@{[Char::Earabic::MATCH()]}';
        }

        # $POSTMATCH, ${POSTMATCH}, ${^POSTMATCH} --> Char::Earabic::POSTMATCH()
        elsif ($char[$i] =~ /\A (                 \$ \s* POSTMATCH | \$ \s* \{ \s* POSTMATCH \s* \} | \$ \s* \{\^POSTMATCH\} ) \z/oxmsgc) {
            $char[$i] = '@{[Char::Earabic::POSTMATCH()]}';
        }

        # ${ foo } --> ${ foo }
        elsif ($char[$i] =~ /\A \$ \s* \{ \s* [A-Za-z_][A-Za-z0-9_]*(?: ::[A-Za-z_][A-Za-z0-9_]*)* \s* \} \z/oxms) {
        }

        # ${ ... }
        elsif ($char[$i] =~ /\A \$ \s* \{ \s* ( .+ ) \s* \} \z/oxms) {
            $char[$i] = e_capture($1);
        }
    }

    # return string
    if ($left_e > $right_e) {
        return join '', @char, '>]}' x ($left_e - $right_e);
    }
    return     join '', @char;
}

#
# escape regexp (m//, qr//)
#
sub e_qr {
    my($ope,$delimiter,$end_delimiter,$string,$modifier) = @_;
    $modifier ||= '';

    $modifier =~ tr/p//d;
    if ($modifier =~ /([adlu])/oxms) {
        my $line = 0;
        for (my $i=0; my($package,$filename,$use_line,$subroutine) = caller($i); $i++) {
            if ($filename ne __FILE__) {
                $line = $use_line + (CORE::substr($_,0,pos($_)) =~ tr/\n//) + 1;
                last;
            }
        }
        die qq{Unsupported modifier "$1" used at line $line.\n};
    }

    $slash = 'div';

    # literal null string pattern
    if ($string eq '') {
        $modifier =~ tr/bB//d;
        $modifier =~ tr/i//d;
        return join '', $ope, $delimiter, $end_delimiter, $modifier;
    }

    # /b /B modifier
    elsif ($modifier =~ tr/bB//d) {

        # choice again delimiter
        if ($delimiter =~ / [\@:] /oxms) {
            my @char = $string =~ /\G([\x00-\xFF])/oxmsg;
            my %octet = map {$_ => 1} @char;
            if (not $octet{')'}) {
                $delimiter     = '(';
                $end_delimiter = ')';
            }
            elsif (not $octet{'}'}) {
                $delimiter     = '{';
                $end_delimiter = '}';
            }
            elsif (not $octet{']'}) {
                $delimiter     = '[';
                $end_delimiter = ']';
            }
            elsif (not $octet{'>'}) {
                $delimiter     = '<';
                $end_delimiter = '>';
            }
            else {
                for my $char (qw( ! " $ % & * + - . / = ? ^ ` | ~ ), "\x00".."\x1F", "\x7F", "\xFF") {
                    if (not $octet{$char}) {
                        $delimiter     = $char;
                        $end_delimiter = $char;
                        last;
                    }
                }
            }
        }

        if (($ope =~ /\A m? \z/oxms) and ($delimiter eq '?')) {
            return join '', $ope, $delimiter,        $string,      $matched, $end_delimiter, $modifier;
        }
        else {
            return join '', $ope, $delimiter, '(?:', $string, ')', $matched, $end_delimiter, $modifier;
        }
    }

    my $ignorecase = ($modifier =~ /i/oxms) ? 1 : 0;
    my $metachar = qr/[\@\\|[\]{^]/oxms;

    # split regexp
    my @char = $string =~ /\G(
        \\o\{ [0-7]+           \}  |
        \\    [0-7]{2,3}           |
        \\x\{ [0-9A-Fa-f]+     \}  |
        \\x   [0-9A-Fa-f]{1,2}     |
        \\c   [\x40-\x5F]          |
        \\N\{ [^0-9\}][^\}]* \} |
        \\p\{ [^0-9\}][^\}]* \} |
        \\P\{ [^0-9\}][^\}]* \} |
        \\  (?:$q_char)            |
        \$` | \$\{`\} | \$ \s* PREMATCH  | \$ \s* \{ \s* PREMATCH  \s* \} | \$ \s* \{\^PREMATCH\}  |
        \$& | \$\{&\} | \$ \s* MATCH     | \$ \s* \{ \s* MATCH     \s* \} | \$ \s* \{\^MATCH\}     |
                        \$ \s* POSTMATCH | \$ \s* \{ \s* POSTMATCH \s* \} | \$ \s* \{\^POSTMATCH\} |
        [\$\@] $qq_variable        |
        \$ \s* \d+                 |
        \$ \s* \{ \s* \d+ \s* \}   |
        \$ \$ (?![\w\{])           |
        \$ \s* \$ \s* $qq_variable |
        \[\:\^ (?:alnum|alpha|ascii|blank|cntrl|digit|graph|lower|print|punct|space|upper|word|xdigit) :\] |
        \[\:   (?:alnum|alpha|ascii|blank|cntrl|digit|graph|lower|print|punct|space|upper|word|xdigit) :\] |
        \[\^                       |
        \(\?                       |
            (?:$q_char)
    )/oxmsg;

    # choice again delimiter
    if ($delimiter =~ / [\@:] /oxms) {
        my %octet = map {$_ => 1} @char;
        if (not $octet{')'}) {
            $delimiter     = '(';
            $end_delimiter = ')';
        }
        elsif (not $octet{'}'}) {
            $delimiter     = '{';
            $end_delimiter = '}';
        }
        elsif (not $octet{']'}) {
            $delimiter     = '[';
            $end_delimiter = ']';
        }
        elsif (not $octet{'>'}) {
            $delimiter     = '<';
            $end_delimiter = '>';
        }
        else {
            for my $char (qw( ! " $ % & * + - . / = ? ^ ` | ~ ), "\x00".."\x1F", "\x7F", "\xFF") {
                if (not $octet{$char}) {
                    $delimiter     = $char;
                    $end_delimiter = $char;
                    last;
                }
            }
        }
    }

    my $left_e  = 0;
    my $right_e = 0;
    for (my $i=0; $i <= $#char; $i++) {

        # "\L\u" --> "\u\L"
        if (($char[$i] eq '\L') and ($char[$i+1] eq '\u')) {
            @char[$i,$i+1] = @char[$i+1,$i];
        }

        # "\U\l" --> "\l\U"
        elsif (($char[$i] eq '\U') and ($char[$i+1] eq '\l')) {
            @char[$i,$i+1] = @char[$i+1,$i];
        }

        # octal escape sequence
        elsif ($char[$i] =~ /\A \\o \{ ([0-7]+) \} \z/oxms) {
            $char[$i] = Char::Earabic::octchr($1);
        }

        # hexadecimal escape sequence
        elsif ($char[$i] =~ /\A \\x \{ ([0-9A-Fa-f]+) \} \z/oxms) {
            $char[$i] = Char::Earabic::hexchr($1);
        }

        # \N{CHARNAME} --> N\{CHARNAME}
        elsif ($char[$i] =~ /\A \\ (N) ( \{ ([^0-9\}][^\}]*) \} ) \z/oxms) {
            $char[$i] = $1 . '\\' . $2;
        }

        # \p{PROPERTY} --> p\{PROPERTY}
        elsif ($char[$i] =~ /\A \\ (p) ( \{ ([^0-9\}][^\}]*) \} ) \z/oxms) {
            $char[$i] = $1 . '\\' . $2;
        }

        # \P{PROPERTY} --> P\{PROPERTY}
        elsif ($char[$i] =~ /\A \\ (P) ( \{ ([^0-9\}][^\}]*) \} ) \z/oxms) {
            $char[$i] = $1 . '\\' . $2;
        }

        # \p, \P, \X --> p, P, X
        elsif ($char[$i] =~ /\A \\ ( [pPX] ) \z/oxms) {
            $char[$i] = $1;
        }

        if (0) {
        }

        # join separated multiple-octet
        elsif ($char[$i] =~ /\A (?: \\ [0-7]{2,3} | \\x [0-9A-Fa-f]{1,2}) \z/oxms) {
            if (   ($i+3 <= $#char) and (grep(/\A (?: \\ [0-7]{2,3} | \\x [0-9A-Fa-f]{1,2}) \z/oxms, @char[$i+1..$i+3]) == 3) and (eval(sprintf '"%s%s%s%s"', @char[$i..$i+3]) =~ /\A $q_char \z/oxms)) {
                $char[$i] .= join '', splice @char, $i+1, 3;
            }
            elsif (($i+2 <= $#char) and (grep(/\A (?: \\ [0-7]{2,3} | \\x [0-9A-Fa-f]{1,2}) \z/oxms, @char[$i+1..$i+2]) == 2) and (eval(sprintf '"%s%s%s"',   @char[$i..$i+2]) =~ /\A $q_char \z/oxms)) {
                $char[$i] .= join '', splice @char, $i+1, 2;
            }
            elsif (($i+1 <= $#char) and (grep(/\A (?: \\ [0-7]{2,3} | \\x [0-9A-Fa-f]{1,2}) \z/oxms, $char[$i+1      ]) == 1) and (eval(sprintf '"%s%s"',     @char[$i..$i+1]) =~ /\A $q_char \z/oxms)) {
                $char[$i] .= join '', splice @char, $i+1, 1;
            }
        }

        # open character class [...]
        elsif ($char[$i] eq '[') {
            my $left = $i;

            # [] make die "Unmatched [] in regexp ..."
            # (and so on)

            if ($char[$i+1] eq ']') {
                $i++;
            }

            while (1) {
                if (++$i > $#char) {
                    die __FILE__, ": Unmatched [] in regexp";
                }
                if ($char[$i] eq ']') {
                    my $right = $i;

                    # [...]
                    if (grep(/\A [\$\@]/oxms,@char[$left+1..$right-1]) >= 1) {
                        splice @char, $left, $right-$left+1, sprintf(q{@{[Char::Earabic::charlist_qr(%s,'%s')]}}, join(',', map {qq_stuff($delimiter,$end_delimiter,$_)} @char[$left+1..$right-1]), $modifier);
                    }
                    else {
                        splice @char, $left, $right-$left+1, Char::Earabic::charlist_qr(@char[$left+1..$right-1], $modifier);
                    }

                    $i = $left;
                    last;
                }
            }
        }

        # open character class [^...]
        elsif ($char[$i] eq '[^') {
            my $left = $i;

            # [^] make die "Unmatched [] in regexp ..."
            # (and so on)

            if ($char[$i+1] eq ']') {
                $i++;
            }

            while (1) {
                if (++$i > $#char) {
                    die __FILE__, ": Unmatched [] in regexp";
                }
                if ($char[$i] eq ']') {
                    my $right = $i;

                    # [^...]
                    if (grep(/\A [\$\@]/oxms,@char[$left+1..$right-1]) >= 1) {
                        splice @char, $left, $right-$left+1, sprintf(q{@{[Char::Earabic::charlist_not_qr(%s,'%s')]}}, join(',', map {qq_stuff($delimiter,$end_delimiter,$_)} @char[$left+1..$right-1]), $modifier);
                    }
                    else {
                        splice @char, $left, $right-$left+1, Char::Earabic::charlist_not_qr(@char[$left+1..$right-1], $modifier);
                    }

                    $i = $left;
                    last;
                }
            }
        }

        # rewrite character class or escape character
        elsif (my $char = character_class($char[$i],$modifier)) {
            $char[$i] = $char;
        }

        # /i modifier
        elsif ($ignorecase and ($char[$i] =~ /\A [\x00-\xFF] \z/oxms) and (Char::Earabic::uc($char[$i]) ne Char::Earabic::fc($char[$i]))) {
            if (CORE::length(Char::Earabic::fc($char[$i])) == 1) {
                $char[$i] = '['   . Char::Earabic::uc($char[$i])       . Char::Earabic::fc($char[$i]) . ']';
            }
            else {
                $char[$i] = '(?:' . Char::Earabic::uc($char[$i]) . '|' . Char::Earabic::fc($char[$i]) . ')';
            }
        }

        # \u \l \U \L \F \Q \E
        elsif ($char[$i] =~ /\A [<>] \z/oxms) {
            if ($right_e < $left_e) {
                $char[$i] = '\\' . $char[$i];
            }
        }
        elsif ($char[$i] eq '\u') {
            $char[$i] = '@{[Char::Earabic::ucfirst qq<';
            $left_e++;
        }
        elsif ($char[$i] eq '\l') {
            $char[$i] = '@{[Char::Earabic::lcfirst qq<';
            $left_e++;
        }
        elsif ($char[$i] eq '\U') {
            $char[$i] = '@{[Char::Earabic::uc qq<';
            $left_e++;
        }
        elsif ($char[$i] eq '\L') {
            $char[$i] = '@{[Char::Earabic::lc qq<';
            $left_e++;
        }
        elsif ($char[$i] eq '\F') {
            $char[$i] = '@{[Char::Earabic::fc qq<';
            $left_e++;
        }
        elsif ($char[$i] eq '\Q') {
            $char[$i] = '@{[CORE::quotemeta qq<';
            $left_e++;
        }
        elsif ($char[$i] eq '\E') {
            if ($right_e < $left_e) {
                $char[$i] = '>]}';
                $right_e++;
            }
            else {
                $char[$i] = '';
            }
        }
        elsif ($char[$i] eq '\Q') {
            while (1) {
                if (++$i > $#char) {
                    last;
                }
                if ($char[$i] eq '\E') {
                    last;
                }
            }
        }
        elsif ($char[$i] eq '\E') {
        }

        # $0 --> $0
        elsif ($char[$i] =~ /\A \$ 0 \z/oxms) {
            if ($ignorecase) {
                $char[$i] = '@{[Char::Earabic::ignorecase(' . $char[$i] . ')]}';
            }
        }
        elsif ($char[$i] =~ /\A \$ \{ \s* 0 \s* \} \z/oxms) {
            if ($ignorecase) {
                $char[$i] = '@{[Char::Earabic::ignorecase(' . $char[$i] . ')]}';
            }
        }

        # $$ --> $$
        elsif ($char[$i] =~ /\A \$\$ \z/oxms) {
        }

        # $1, $2, $3 --> $2, $3, $4 after s/// with multibyte anchoring
        # $1, $2, $3 --> $1, $2, $3 otherwise
        elsif ($char[$i] =~ /\A \$ ([1-9][0-9]*) \z/oxms) {
            $char[$i] = e_capture($1);
            if ($ignorecase) {
                $char[$i] = '@{[Char::Earabic::ignorecase(' . $char[$i] . ')]}';
            }
        }
        elsif ($char[$i] =~ /\A \$ \{ \s* ([1-9][0-9]*) \s* \} \z/oxms) {
            $char[$i] = e_capture($1);
            if ($ignorecase) {
                $char[$i] = '@{[Char::Earabic::ignorecase(' . $char[$i] . ')]}';
            }
        }

        # $$foo[ ... ] --> $ $foo->[ ... ]
        elsif ($char[$i] =~ /\A \$ ( \$ [A-Za-z_][A-Za-z0-9_]*(?: ::[A-Za-z_][A-Za-z0-9_]*)* ) ( \[ (?:$qq_bracket)*? \] ) \z/oxms) {
            $char[$i] = e_capture($1.'->'.$2);
            if ($ignorecase) {
                $char[$i] = '@{[Char::Earabic::ignorecase(' . $char[$i] . ')]}';
            }
        }

        # $$foo{ ... } --> $ $foo->{ ... }
        elsif ($char[$i] =~ /\A \$ ( \$ [A-Za-z_][A-Za-z0-9_]*(?: ::[A-Za-z_][A-Za-z0-9_]*)* ) ( \{ (?:$qq_brace)*? \} ) \z/oxms) {
            $char[$i] = e_capture($1.'->'.$2);
            if ($ignorecase) {
                $char[$i] = '@{[Char::Earabic::ignorecase(' . $char[$i] . ')]}';
            }
        }

        # $$foo
        elsif ($char[$i] =~ /\A \$ ( \$ [A-Za-z_][A-Za-z0-9_]*(?: ::[A-Za-z_][A-Za-z0-9_]*)* ) \z/oxms) {
            $char[$i] = e_capture($1);
            if ($ignorecase) {
                $char[$i] = '@{[Char::Earabic::ignorecase(' . $char[$i] . ')]}';
            }
        }

        # $`, ${`}, $PREMATCH, ${PREMATCH}, ${^PREMATCH} --> Char::Earabic::PREMATCH()
        elsif ($char[$i] =~ /\A ( \$` | \$\{`\} | \$ \s* PREMATCH  | \$ \s* \{ \s* PREMATCH  \s* \} | \$ \s* \{\^PREMATCH\}  ) \z/oxmsgc) {
            if ($ignorecase) {
                $char[$i] = '@{[Char::Earabic::ignorecase(Char::Earabic::PREMATCH())]}';
            }
            else {
                $char[$i] = '@{[Char::Earabic::PREMATCH()]}';
            }
        }

        # $&, ${&}, $MATCH, ${MATCH}, ${^MATCH} --> Char::Earabic::MATCH()
        elsif ($char[$i] =~ /\A ( \$& | \$\{&\} | \$ \s* MATCH     | \$ \s* \{ \s* MATCH     \s* \} | \$ \s* \{\^MATCH\}     ) \z/oxmsgc) {
            if ($ignorecase) {
                $char[$i] = '@{[Char::Earabic::ignorecase(Char::Earabic::MATCH())]}';
            }
            else {
                $char[$i] = '@{[Char::Earabic::MATCH()]}';
            }
        }

        # $POSTMATCH, ${POSTMATCH}, ${^POSTMATCH} --> Char::Earabic::POSTMATCH()
        elsif ($char[$i] =~ /\A (                 \$ \s* POSTMATCH | \$ \s* \{ \s* POSTMATCH \s* \} | \$ \s* \{\^POSTMATCH\} ) \z/oxmsgc) {
            if ($ignorecase) {
                $char[$i] = '@{[Char::Earabic::ignorecase(Char::Earabic::POSTMATCH())]}';
            }
            else {
                $char[$i] = '@{[Char::Earabic::POSTMATCH()]}';
            }
        }

        # ${ foo }
        elsif ($char[$i] =~ /\A \$ \s* \{ ( \s* [A-Za-z_][A-Za-z0-9_]*(?: ::[A-Za-z_][A-Za-z0-9_]*)* \s* ) \} \z/oxms) {
            if ($ignorecase) {
                $char[$i] = '@{[Char::Earabic::ignorecase(' . $char[$i] . ')]}';
            }
        }

        # ${ ... }
        elsif ($char[$i] =~ /\A \$ \s* \{ \s* ( .+ ) \s* \} \z/oxms) {
            $char[$i] = e_capture($1);
            if ($ignorecase) {
                $char[$i] = '@{[Char::Earabic::ignorecase(' . $char[$i] . ')]}';
            }
        }

        # $scalar or @array
        elsif ($char[$i] =~ /\A [\$\@].+ /oxms) {
            $char[$i] = e_string($char[$i]);
            if ($ignorecase) {
                $char[$i] = '@{[Char::Earabic::ignorecase(' . $char[$i] . ')]}';
            }
        }

        # quote character before ? + * {
        elsif (($i >= 1) and ($char[$i] =~ /\A [\?\+\*\{] \z/oxms)) {
            if ($char[$i-1] =~ /\A (?:\\[0-7]{2,3}|\\x[0-9-A-Fa-f]{1,2}) \z/oxms) {
            }
            elsif (($ope =~ /\A m? \z/oxms) and ($delimiter eq '?')) {
                my $char = $char[$i-1];
                if ($char[$i] eq '{') {
                    die __FILE__, qq{: "MULTIBYTE{n}" should be "(MULTIBYTE){n}" in m?? (and shift \$1,\$2,\$3,...) ($char){n}};
                }
                else {
                    die __FILE__, qq{: "MULTIBYTE$char[$i]" should be "(MULTIBYTE)$char[$i]" in m?? (and shift \$1,\$2,\$3,...) ($char)$char[$i]};
                }
            }
            else {
                $char[$i-1] = '(?:' . $char[$i-1] . ')';
            }
        }
    }

    # make regexp string
    $modifier =~ tr/i//d;
    if ($left_e > $right_e) {
        if (($ope =~ /\A m? \z/oxms) and ($delimiter eq '?')) {
            return join '', $ope, $delimiter, $anchor,        @char, '>]}' x ($left_e - $right_e),      $matched, $end_delimiter, $modifier;
        }
        else {
            return join '', $ope, $delimiter, $anchor, '(?:', @char, '>]}' x ($left_e - $right_e), ')', $matched, $end_delimiter, $modifier;
        }
    }
    if (($ope =~ /\A m? \z/oxms) and ($delimiter eq '?')) {
        return     join '', $ope, $delimiter, $anchor,        @char,                                    $matched, $end_delimiter, $modifier;
    }
    else {
        return     join '', $ope, $delimiter, $anchor, '(?:', @char,                               ')', $matched, $end_delimiter, $modifier;
    }
}

#
# double quote stuff
#
sub qq_stuff {
    my($delimiter,$end_delimiter,$stuff) = @_;

    # scalar variable or array variable
    if ($stuff =~ /\A [\$\@] /oxms) {
        return $stuff;
    }

    # quote by delimiter
    my %octet = map {$_ => 1} ($stuff =~ /\G ([\x00-\xFF]) /oxmsg);
    for my $char (qw( ! " $ % & * + - . / : = ? @ ^ ` | ~ ), "\x00".."\x1F", "\x7F", "\xFF") {
        next if $char eq $delimiter;
        next if $char eq $end_delimiter;
        if (not $octet{$char}) {
            return join '', 'qq', $char, $stuff, $char;
        }
    }
    return join '', 'qq', '<', $stuff, '>';
}

#
# escape regexp (m'', qr'', and m''b, qr''b)
#
sub e_qr_q {
    my($ope,$delimiter,$end_delimiter,$string,$modifier) = @_;
    $modifier ||= '';

    $modifier =~ tr/p//d;
    if ($modifier =~ /([adlu])/oxms) {
        my $line = 0;
        for (my $i=0; my($package,$filename,$use_line,$subroutine) = caller($i); $i++) {
            if ($filename ne __FILE__) {
                $line = $use_line + (CORE::substr($_,0,pos($_)) =~ tr/\n//) + 1;
                last;
            }
        }
        die qq{Unsupported modifier "$1" used at line $line.\n};
    }

    $slash = 'div';

    # literal null string pattern
    if ($string eq '') {
        $modifier =~ tr/bB//d;
        $modifier =~ tr/i//d;
        return join '', $ope, $delimiter, $end_delimiter, $modifier;
    }

    # with /b /B modifier
    elsif ($modifier =~ tr/bB//d) {
        return e_qr_qb($ope,$delimiter,$end_delimiter,$string,$modifier);
    }

    # without /b /B modifier
    else {
        return e_qr_qt($ope,$delimiter,$end_delimiter,$string,$modifier);
    }
}

#
# escape regexp (m'', qr'')
#
sub e_qr_qt {
    my($ope,$delimiter,$end_delimiter,$string,$modifier) = @_;

    my $ignorecase = ($modifier =~ /i/oxms) ? 1 : 0;

    # split regexp
    my @char = $string =~ /\G(
        \[\:\^ [a-z]+ \:\] |
        \[\:   [a-z]+ \:\] |
        \[\^               |
        [\$\@\/\\]         |
        \\?    (?:$q_char)
    )/oxmsg;

    # unescape character
    for (my $i=0; $i <= $#char; $i++) {
        if (0) {
        }

        # open character class [...]
        elsif ($char[$i] eq '[') {
            my $left = $i;
            if ($char[$i+1] eq ']') {
                $i++;
            }
            while (1) {
                if (++$i > $#char) {
                    die __FILE__, ": Unmatched [] in regexp";
                }
                if ($char[$i] eq ']') {
                    my $right = $i;

                    # [...]
                    splice @char, $left, $right-$left+1, Char::Earabic::charlist_qr(@char[$left+1..$right-1], $modifier);

                    $i = $left;
                    last;
                }
            }
        }

        # open character class [^...]
        elsif ($char[$i] eq '[^') {
            my $left = $i;
            if ($char[$i+1] eq ']') {
                $i++;
            }
            while (1) {
                if (++$i > $#char) {
                    die __FILE__, ": Unmatched [] in regexp";
                }
                if ($char[$i] eq ']') {
                    my $right = $i;

                    # [^...]
                    splice @char, $left, $right-$left+1, Char::Earabic::charlist_not_qr(@char[$left+1..$right-1], $modifier);

                    $i = $left;
                    last;
                }
            }
        }

        # escape $ @ / and \
        elsif ($char[$i] =~ /\A [\$\@\/\\] \z/oxms) {
            $char[$i] = '\\' . $char[$i];
        }

        # rewrite character class or escape character
        elsif (my $char = character_class($char[$i],$modifier)) {
            $char[$i] = $char;
        }

        # /i modifier
        elsif ($ignorecase and ($char[$i] =~ /\A [\x00-\xFF] \z/oxms) and (Char::Earabic::uc($char[$i]) ne Char::Earabic::fc($char[$i]))) {
            if (CORE::length(Char::Earabic::fc($char[$i])) == 1) {
                $char[$i] = '['   . Char::Earabic::uc($char[$i])       . Char::Earabic::fc($char[$i]) . ']';
            }
            else {
                $char[$i] = '(?:' . Char::Earabic::uc($char[$i]) . '|' . Char::Earabic::fc($char[$i]) . ')';
            }
        }

        # quote character before ? + * {
        elsif (($i >= 1) and ($char[$i] =~ /\A [\?\+\*\{] \z/oxms)) {
            if ($char[$i-1] =~ /\A [\x00-\xFF] \z/oxms) {
            }
            else {
                $char[$i-1] = '(?:' . $char[$i-1] . ')';
            }
        }
    }

    $delimiter     = '/';
    $end_delimiter = '/';

    $modifier =~ tr/i//d;
    return join '', $ope, $delimiter, $anchor, '(?:', @char, ')', $matched, $end_delimiter, $modifier;
}

#
# escape regexp (m''b, qr''b)
#
sub e_qr_qb {
    my($ope,$delimiter,$end_delimiter,$string,$modifier) = @_;

    # split regexp
    my @char = $string =~ /\G(
        \\\\        |
        [\$\@\/\\]  |
        [\x00-\xFF]
    )/oxmsg;

    # unescape character
    for (my $i=0; $i <= $#char; $i++) {
        if (0) {
        }

        # remain \\
        elsif ($char[$i] eq '\\\\') {
        }

        # escape $ @ / and \
        elsif ($char[$i] =~ /\A [\$\@\/\\] \z/oxms) {
            $char[$i] = '\\' . $char[$i];
        }
    }

    $delimiter     = '/';
    $end_delimiter = '/';
    return join '', $ope, $delimiter, '(?:', @char, ')', $matched, $end_delimiter, $modifier;
}

#
# escape regexp (s/here//)
#
sub e_s1 {
    my($ope,$delimiter,$end_delimiter,$string,$modifier) = @_;
    $modifier ||= '';

    $modifier =~ tr/p//d;
    if ($modifier =~ /([adlu])/oxms) {
        my $line = 0;
        for (my $i=0; my($package,$filename,$use_line,$subroutine) = caller($i); $i++) {
            if ($filename ne __FILE__) {
                $line = $use_line + (CORE::substr($_,0,pos($_)) =~ tr/\n//) + 1;
                last;
            }
        }
        die qq{Unsupported modifier "$1" used at line $line.\n};
    }

    $slash = 'div';

    # literal null string pattern
    if ($string eq '') {
        $modifier =~ tr/bB//d;
        $modifier =~ tr/i//d;
        return join '', $ope, $delimiter, $end_delimiter, $modifier;
    }

    # /b /B modifier
    elsif ($modifier =~ tr/bB//d) {

        # choice again delimiter
        if ($delimiter =~ / [\@:] /oxms) {
            my @char = $string =~ /\G([\x00-\xFF])/oxmsg;
            my %octet = map {$_ => 1} @char;
            if (not $octet{')'}) {
                $delimiter     = '(';
                $end_delimiter = ')';
            }
            elsif (not $octet{'}'}) {
                $delimiter     = '{';
                $end_delimiter = '}';
            }
            elsif (not $octet{']'}) {
                $delimiter     = '[';
                $end_delimiter = ']';
            }
            elsif (not $octet{'>'}) {
                $delimiter     = '<';
                $end_delimiter = '>';
            }
            else {
                for my $char (qw( ! " $ % & * + - . / = ? ^ ` | ~ ), "\x00".."\x1F", "\x7F", "\xFF") {
                    if (not $octet{$char}) {
                        $delimiter     = $char;
                        $end_delimiter = $char;
                        last;
                    }
                }
            }
        }

        my $prematch = '';
        return join '', $ope, $delimiter, $prematch, '(?:', $string, ')', $matched, $end_delimiter, $modifier;
    }

    my $ignorecase = ($modifier =~ /i/oxms) ? 1 : 0;
    my $metachar = qr/[\@\\|[\]{^]/oxms;

    # split regexp
    my @char = $string =~ /\G(
        \\g \s* \{ \s* - \s* [1-9][0-9]* \s* \} |
        \\g \s* \{ \s*       [1-9][0-9]* \s* \} |
        \\g \s*              [1-9][0-9]*        |
        \\o\{                [0-7]+          \} |
        \\                   [1-9][0-9]*        |
        \\                   [0-7]{2,3}         |
        \\x\{                [0-9A-Fa-f]+    \} |
        \\x                  [0-9A-Fa-f]{1,2}   |
        \\c                  [\x40-\x5F]        |
        \\N\{                [^0-9\}][^\}]* \} |
        \\p\{                [^0-9\}][^\}]* \} |
        \\P\{                [^0-9\}][^\}]* \} |
        \\  (?:$q_char)                         |
        \$` | \$\{`\} | \$ \s* PREMATCH  | \$ \s* \{ \s* PREMATCH  \s* \} | \$ \s* \{\^PREMATCH\}  |
        \$& | \$\{&\} | \$ \s* MATCH     | \$ \s* \{ \s* MATCH     \s* \} | \$ \s* \{\^MATCH\}     |
                        \$ \s* POSTMATCH | \$ \s* \{ \s* POSTMATCH \s* \} | \$ \s* \{\^POSTMATCH\} |
        [\$\@] $qq_variable                     |
        \$ \s* \d+                              |
        \$ \s* \{ \s* \d+ \s* \}                |
        \$ \$ (?![\w\{])                        |
        \$ \s* \$ \s* $qq_variable              |
        \[\:\^ (?:alnum|alpha|ascii|blank|cntrl|digit|graph|lower|print|punct|space|upper|word|xdigit) :\] |
        \[\:   (?:alnum|alpha|ascii|blank|cntrl|digit|graph|lower|print|punct|space|upper|word|xdigit) :\] |
        \[\^                                    |
        \(\?                                    |
            (?:$q_char)
    )/oxmsg;

    # choice again delimiter
    if ($delimiter =~ / [\@:] /oxms) {
        my %octet = map {$_ => 1} @char;
        if (not $octet{')'}) {
            $delimiter     = '(';
            $end_delimiter = ')';
        }
        elsif (not $octet{'}'}) {
            $delimiter     = '{';
            $end_delimiter = '}';
        }
        elsif (not $octet{']'}) {
            $delimiter     = '[';
            $end_delimiter = ']';
        }
        elsif (not $octet{'>'}) {
            $delimiter     = '<';
            $end_delimiter = '>';
        }
        else {
            for my $char (qw( ! " $ % & * + - . / = ? ^ ` | ~ ), "\x00".."\x1F", "\x7F", "\xFF") {
                if (not $octet{$char}) {
                    $delimiter     = $char;
                    $end_delimiter = $char;
                    last;
                }
            }
        }
    }

    # count '('
    my $parens = grep { $_ eq '(' } @char;

    my $left_e  = 0;
    my $right_e = 0;
    for (my $i=0; $i <= $#char; $i++) {

        # "\L\u" --> "\u\L"
        if (($char[$i] eq '\L') and ($char[$i+1] eq '\u')) {
            @char[$i,$i+1] = @char[$i+1,$i];
        }

        # "\U\l" --> "\l\U"
        elsif (($char[$i] eq '\U') and ($char[$i+1] eq '\l')) {
            @char[$i,$i+1] = @char[$i+1,$i];
        }

        # octal escape sequence
        elsif ($char[$i] =~ /\A \\o \{ ([0-7]+) \} \z/oxms) {
            $char[$i] = Char::Earabic::octchr($1);
        }

        # hexadecimal escape sequence
        elsif ($char[$i] =~ /\A \\x \{ ([0-9A-Fa-f]+) \} \z/oxms) {
            $char[$i] = Char::Earabic::hexchr($1);
        }

        # \N{CHARNAME} --> N\{CHARNAME}
        elsif ($char[$i] =~ /\A \\ (N) ( \{ ([^0-9\}][^\}]*) \} ) \z/oxms) {
            $char[$i] = $1 . '\\' . $2;
        }

        # \p{PROPERTY} --> p\{PROPERTY}
        elsif ($char[$i] =~ /\A \\ (p) ( \{ ([^0-9\}][^\}]*) \} ) \z/oxms) {
            $char[$i] = $1 . '\\' . $2;
        }

        # \P{PROPERTY} --> P\{PROPERTY}
        elsif ($char[$i] =~ /\A \\ (P) ( \{ ([^0-9\}][^\}]*) \} ) \z/oxms) {
            $char[$i] = $1 . '\\' . $2;
        }

        # \p, \P, \X --> p, P, X
        elsif ($char[$i] =~ /\A \\ ( [pPX] ) \z/oxms) {
            $char[$i] = $1;
        }

        if (0) {
        }

        # join separated multiple-octet
        elsif ($char[$i] =~ /\A (?: \\ [0-7]{2,3} | \\x [0-9A-Fa-f]{1,2}) \z/oxms) {
            if (   ($i+3 <= $#char) and (grep(/\A (?: \\ [0-7]{2,3} | \\x [0-9A-Fa-f]{1,2}) \z/oxms, @char[$i+1..$i+3]) == 3) and (eval(sprintf '"%s%s%s%s"', @char[$i..$i+3]) =~ /\A $q_char \z/oxms)) {
                $char[$i] .= join '', splice @char, $i+1, 3;
            }
            elsif (($i+2 <= $#char) and (grep(/\A (?: \\ [0-7]{2,3} | \\x [0-9A-Fa-f]{1,2}) \z/oxms, @char[$i+1..$i+2]) == 2) and (eval(sprintf '"%s%s%s"',   @char[$i..$i+2]) =~ /\A $q_char \z/oxms)) {
                $char[$i] .= join '', splice @char, $i+1, 2;
            }
            elsif (($i+1 <= $#char) and (grep(/\A (?: \\ [0-7]{2,3} | \\x [0-9A-Fa-f]{1,2}) \z/oxms, $char[$i+1      ]) == 1) and (eval(sprintf '"%s%s"',     @char[$i..$i+1]) =~ /\A $q_char \z/oxms)) {
                $char[$i] .= join '', splice @char, $i+1, 1;
            }
        }

        # open character class [...]
        elsif ($char[$i] eq '[') {
            my $left = $i;
            if ($char[$i+1] eq ']') {
                $i++;
            }
            while (1) {
                if (++$i > $#char) {
                    die __FILE__, ": Unmatched [] in regexp";
                }
                if ($char[$i] eq ']') {
                    my $right = $i;

                    # [...]
                    if (grep(/\A [\$\@]/oxms,@char[$left+1..$right-1]) >= 1) {
                        splice @char, $left, $right-$left+1, sprintf(q{@{[Char::Earabic::charlist_qr(%s,'%s')]}}, join(',', map {qq_stuff($delimiter,$end_delimiter,$_)} @char[$left+1..$right-1]), $modifier);
                    }
                    else {
                        splice @char, $left, $right-$left+1, Char::Earabic::charlist_qr(@char[$left+1..$right-1], $modifier);
                    }

                    $i = $left;
                    last;
                }
            }
        }

        # open character class [^...]
        elsif ($char[$i] eq '[^') {
            my $left = $i;
            if ($char[$i+1] eq ']') {
                $i++;
            }
            while (1) {
                if (++$i > $#char) {
                    die __FILE__, ": Unmatched [] in regexp";
                }
                if ($char[$i] eq ']') {
                    my $right = $i;

                    # [^...]
                    if (grep(/\A [\$\@]/oxms,@char[$left+1..$right-1]) >= 1) {
                        splice @char, $left, $right-$left+1, sprintf(q{@{[Char::Earabic::charlist_not_qr(%s,'%s')]}}, join(',', map {qq_stuff($delimiter,$end_delimiter,$_)} @char[$left+1..$right-1]), $modifier);
                    }
                    else {
                        splice @char, $left, $right-$left+1, Char::Earabic::charlist_not_qr(@char[$left+1..$right-1], $modifier);
                    }

                    $i = $left;
                    last;
                }
            }
        }

        # rewrite character class or escape character
        elsif (my $char = character_class($char[$i],$modifier)) {
            $char[$i] = $char;
        }

        # /i modifier
        elsif ($ignorecase and ($char[$i] =~ /\A [\x00-\xFF] \z/oxms) and (Char::Earabic::uc($char[$i]) ne Char::Earabic::fc($char[$i]))) {
            if (CORE::length(Char::Earabic::fc($char[$i])) == 1) {
                $char[$i] = '['   . Char::Earabic::uc($char[$i])       . Char::Earabic::fc($char[$i]) . ']';
            }
            else {
                $char[$i] = '(?:' . Char::Earabic::uc($char[$i]) . '|' . Char::Earabic::fc($char[$i]) . ')';
            }
        }

        # \u \l \U \L \F \Q \E
        elsif ($char[$i] =~ /\A [<>] \z/oxms) {
            if ($right_e < $left_e) {
                $char[$i] = '\\' . $char[$i];
            }
        }
        elsif ($char[$i] eq '\u') {
            $char[$i] = '@{[Char::Earabic::ucfirst qq<';
            $left_e++;
        }
        elsif ($char[$i] eq '\l') {
            $char[$i] = '@{[Char::Earabic::lcfirst qq<';
            $left_e++;
        }
        elsif ($char[$i] eq '\U') {
            $char[$i] = '@{[Char::Earabic::uc qq<';
            $left_e++;
        }
        elsif ($char[$i] eq '\L') {
            $char[$i] = '@{[Char::Earabic::lc qq<';
            $left_e++;
        }
        elsif ($char[$i] eq '\F') {
            $char[$i] = '@{[Char::Earabic::fc qq<';
            $left_e++;
        }
        elsif ($char[$i] eq '\Q') {
            $char[$i] = '@{[CORE::quotemeta qq<';
            $left_e++;
        }
        elsif ($char[$i] eq '\E') {
            if ($right_e < $left_e) {
                $char[$i] = '>]}';
                $right_e++;
            }
            else {
                $char[$i] = '';
            }
        }
        elsif ($char[$i] eq '\Q') {
            while (1) {
                if (++$i > $#char) {
                    last;
                }
                if ($char[$i] eq '\E') {
                    last;
                }
            }
        }
        elsif ($char[$i] eq '\E') {
        }

        # \0 --> \0
        elsif ($char[$i] =~ /\A \\ \s* 0 \z/oxms) {
        }

        # \g{N}, \g{-N}

        # P.108 Using Simple Patterns
        # in Chapter 7: In the World of Regular Expressions
        # of ISBN 978-0-596-52010-6 Learning Perl, Fifth Edition

        # P.221 Capturing
        # in Chapter 5: Pattern Matching
        # of ISBN 978-0-596-00492-7 Programming Perl 4th Edition.

        # \g{-1}, \g{-2}, \g{-3} --> \g{-1}, \g{-2}, \g{-3}
        elsif ($char[$i] =~ /\A \\g \s* \{ \s* - \s* ([1-9][0-9]*) \s* \} \z/oxms) {
        }

        # \g{1}, \g{2}, \g{3} --> \g{2}, \g{3}, \g{4} (only when multibyte anchoring is enable)
        elsif ($char[$i] =~ /\A \\g \s* \{ \s* ([1-9][0-9]*) \s* \} \z/oxms) {
        }

        # \g1, \g2, \g3 --> \g2, \g3, \g4 (only when multibyte anchoring is enable)
        elsif ($char[$i] =~ /\A \\g \s* ([1-9][0-9]*) \z/oxms) {
        }

        # \1, \2, \3 --> \2, \3, \4 (only when multibyte anchoring is enable)
        elsif ($char[$i] =~ /\A \\ \s* ([1-9][0-9]*) \z/oxms) {
        }

        # $0 --> $0
        elsif ($char[$i] =~ /\A \$ 0 \z/oxms) {
            if ($ignorecase) {
                $char[$i] = '@{[Char::Earabic::ignorecase(' . $char[$i] . ')]}';
            }
        }
        elsif ($char[$i] =~ /\A \$ \{ \s* 0 \s* \} \z/oxms) {
            if ($ignorecase) {
                $char[$i] = '@{[Char::Earabic::ignorecase(' . $char[$i] . ')]}';
            }
        }

        # $$ --> $$
        elsif ($char[$i] =~ /\A \$\$ \z/oxms) {
        }

        # $1, $2, $3 --> $2, $3, $4 after s/// with multibyte anchoring
        # $1, $2, $3 --> $1, $2, $3 otherwise
        elsif ($char[$i] =~ /\A \$ ([1-9][0-9]*) \z/oxms) {
            $char[$i] = e_capture($1);
            if ($ignorecase) {
                $char[$i] = '@{[Char::Earabic::ignorecase(' . $char[$i] . ')]}';
            }
        }
        elsif ($char[$i] =~ /\A \$ \{ \s* ([1-9][0-9]*) \s* \} \z/oxms) {
            $char[$i] = e_capture($1);
            if ($ignorecase) {
                $char[$i] = '@{[Char::Earabic::ignorecase(' . $char[$i] . ')]}';
            }
        }

        # $$foo[ ... ] --> $ $foo->[ ... ]
        elsif ($char[$i] =~ /\A \$ ( \$ [A-Za-z_][A-Za-z0-9_]*(?: ::[A-Za-z_][A-Za-z0-9_]*)* ) ( \[ (?:$qq_bracket)*? \] ) \z/oxms) {
            $char[$i] = e_capture($1.'->'.$2);
            if ($ignorecase) {
                $char[$i] = '@{[Char::Earabic::ignorecase(' . $char[$i] . ')]}';
            }
        }

        # $$foo{ ... } --> $ $foo->{ ... }
        elsif ($char[$i] =~ /\A \$ ( \$ [A-Za-z_][A-Za-z0-9_]*(?: ::[A-Za-z_][A-Za-z0-9_]*)* ) ( \{ (?:$qq_brace)*? \} ) \z/oxms) {
            $char[$i] = e_capture($1.'->'.$2);
            if ($ignorecase) {
                $char[$i] = '@{[Char::Earabic::ignorecase(' . $char[$i] . ')]}';
            }
        }

        # $$foo
        elsif ($char[$i] =~ /\A \$ ( \$ [A-Za-z_][A-Za-z0-9_]*(?: ::[A-Za-z_][A-Za-z0-9_]*)* ) \z/oxms) {
            $char[$i] = e_capture($1);
            if ($ignorecase) {
                $char[$i] = '@{[Char::Earabic::ignorecase(' . $char[$i] . ')]}';
            }
        }

        # $`, ${`}, $PREMATCH, ${PREMATCH}, ${^PREMATCH} --> Char::Earabic::PREMATCH()
        elsif ($char[$i] =~ /\A ( \$` | \$\{`\} | \$ \s* PREMATCH  | \$ \s* \{ \s* PREMATCH  \s* \} | \$ \s* \{\^PREMATCH\}  ) \z/oxmsgc) {
            if ($ignorecase) {
                $char[$i] = '@{[Char::Earabic::ignorecase(Char::Earabic::PREMATCH())]}';
            }
            else {
                $char[$i] = '@{[Char::Earabic::PREMATCH()]}';
            }
        }

        # $&, ${&}, $MATCH, ${MATCH}, ${^MATCH} --> Char::Earabic::MATCH()
        elsif ($char[$i] =~ /\A ( \$& | \$\{&\} | \$ \s* MATCH     | \$ \s* \{ \s* MATCH     \s* \} | \$ \s* \{\^MATCH\}     ) \z/oxmsgc) {
            if ($ignorecase) {
                $char[$i] = '@{[Char::Earabic::ignorecase(Char::Earabic::MATCH())]}';
            }
            else {
                $char[$i] = '@{[Char::Earabic::MATCH()]}';
            }
        }

        # $POSTMATCH, ${POSTMATCH}, ${^POSTMATCH} --> Char::Earabic::POSTMATCH()
        elsif ($char[$i] =~ /\A (                 \$ \s* POSTMATCH | \$ \s* \{ \s* POSTMATCH \s* \} | \$ \s* \{\^POSTMATCH\} ) \z/oxmsgc) {
            if ($ignorecase) {
                $char[$i] = '@{[Char::Earabic::ignorecase(Char::Earabic::POSTMATCH())]}';
            }
            else {
                $char[$i] = '@{[Char::Earabic::POSTMATCH()]}';
            }
        }

        # ${ foo }
        elsif ($char[$i] =~ /\A \$ \s* \{ ( \s* [A-Za-z_][A-Za-z0-9_]*(?: ::[A-Za-z_][A-Za-z0-9_]*)* \s* ) \} \z/oxms) {
            if ($ignorecase) {
                $char[$i] = '@{[Char::Earabic::ignorecase(' . $char[$i] . ')]}';
            }
        }

        # ${ ... }
        elsif ($char[$i] =~ /\A \$ \s* \{ \s* ( .+ ) \s* \} \z/oxms) {
            $char[$i] = e_capture($1);
            if ($ignorecase) {
                $char[$i] = '@{[Char::Earabic::ignorecase(' . $char[$i] . ')]}';
            }
        }

        # $scalar or @array
        elsif ($char[$i] =~ /\A [\$\@].+ /oxms) {
            $char[$i] = e_string($char[$i]);
            if ($ignorecase) {
                $char[$i] = '@{[Char::Earabic::ignorecase(' . $char[$i] . ')]}';
            }
        }

        # quote character before ? + * {
        elsif (($i >= 1) and ($char[$i] =~ /\A [\?\+\*\{] \z/oxms)) {
            if ($char[$i-1] =~ /\A (?:\\[0-7]{2,3}|\\x[0-9-A-Fa-f]{1,2}) \z/oxms) {
            }
            else {
                $char[$i-1] = '(?:' . $char[$i-1] . ')';
            }
        }
    }

    # make regexp string
    my $prematch = '';
    $modifier =~ tr/i//d;
    if ($left_e > $right_e) {
        return join '', $ope, $delimiter, $prematch, '(?:', @char, '>]}' x ($left_e - $right_e), ')', $matched, $end_delimiter, $modifier;
    }
    return     join '', $ope, $delimiter, $prematch, '(?:', @char,                               ')', $matched, $end_delimiter, $modifier;
}

#
# escape regexp (s'here'' or s'here''b)
#
sub e_s1_q {
    my($ope,$delimiter,$end_delimiter,$string,$modifier) = @_;
    $modifier ||= '';

    $modifier =~ tr/p//d;
    if ($modifier =~ /([adlu])/oxms) {
        my $line = 0;
        for (my $i=0; my($package,$filename,$use_line,$subroutine) = caller($i); $i++) {
            if ($filename ne __FILE__) {
                $line = $use_line + (CORE::substr($_,0,pos($_)) =~ tr/\n//) + 1;
                last;
            }
        }
        die qq{Unsupported modifier "$1" used at line $line.\n};
    }

    $slash = 'div';

    # literal null string pattern
    if ($string eq '') {
        $modifier =~ tr/bB//d;
        $modifier =~ tr/i//d;
        return join '', $ope, $delimiter, $end_delimiter, $modifier;
    }

    # with /b /B modifier
    elsif ($modifier =~ tr/bB//d) {
        return e_s1_qb($ope,$delimiter,$end_delimiter,$string,$modifier);
    }

    # without /b /B modifier
    else {
        return e_s1_qt($ope,$delimiter,$end_delimiter,$string,$modifier);
    }
}

#
# escape regexp (s'here'')
#
sub e_s1_qt {
    my($ope,$delimiter,$end_delimiter,$string,$modifier) = @_;

    my $ignorecase = ($modifier =~ /i/oxms) ? 1 : 0;

    # split regexp
    my @char = $string =~ /\G(
        \[\:\^ [a-z]+ \:\] |
        \[\:   [a-z]+ \:\] |
        \[\^               |
        [\$\@\/\\]         |
        \\?    (?:$q_char)
    )/oxmsg;

    # unescape character
    for (my $i=0; $i <= $#char; $i++) {
        if (0) {
        }

        # open character class [...]
        elsif ($char[$i] eq '[') {
            my $left = $i;
            if ($char[$i+1] eq ']') {
                $i++;
            }
            while (1) {
                if (++$i > $#char) {
                    die __FILE__, ": Unmatched [] in regexp";
                }
                if ($char[$i] eq ']') {
                    my $right = $i;

                    # [...]
                    splice @char, $left, $right-$left+1, Char::Earabic::charlist_qr(@char[$left+1..$right-1], $modifier);

                    $i = $left;
                    last;
                }
            }
        }

        # open character class [^...]
        elsif ($char[$i] eq '[^') {
            my $left = $i;
            if ($char[$i+1] eq ']') {
                $i++;
            }
            while (1) {
                if (++$i > $#char) {
                    die __FILE__, ": Unmatched [] in regexp";
                }
                if ($char[$i] eq ']') {
                    my $right = $i;

                    # [^...]
                    splice @char, $left, $right-$left+1, Char::Earabic::charlist_not_qr(@char[$left+1..$right-1], $modifier);

                    $i = $left;
                    last;
                }
            }
        }

        # escape $ @ / and \
        elsif ($char[$i] =~ /\A [\$\@\/\\] \z/oxms) {
            $char[$i] = '\\' . $char[$i];
        }

        # rewrite character class or escape character
        elsif (my $char = character_class($char[$i],$modifier)) {
            $char[$i] = $char;
        }

        # /i modifier
        elsif ($ignorecase and ($char[$i] =~ /\A [\x00-\xFF] \z/oxms) and (Char::Earabic::uc($char[$i]) ne Char::Earabic::fc($char[$i]))) {
            if (CORE::length(Char::Earabic::fc($char[$i])) == 1) {
                $char[$i] = '['   . Char::Earabic::uc($char[$i])       . Char::Earabic::fc($char[$i]) . ']';
            }
            else {
                $char[$i] = '(?:' . Char::Earabic::uc($char[$i]) . '|' . Char::Earabic::fc($char[$i]) . ')';
            }
        }

        # quote character before ? + * {
        elsif (($i >= 1) and ($char[$i] =~ /\A [\?\+\*\{] \z/oxms)) {
            if ($char[$i-1] =~ /\A [\x00-\xFF] \z/oxms) {
            }
            else {
                $char[$i-1] = '(?:' . $char[$i-1] . ')';
            }
        }
    }

    $modifier =~ tr/i//d;
    $delimiter     = '/';
    $end_delimiter = '/';
    my $prematch = '';
    return join '', $ope, $delimiter, $prematch, '(?:', @char, ')', $matched, $end_delimiter, $modifier;
}

#
# escape regexp (s'here''b)
#
sub e_s1_qb {
    my($ope,$delimiter,$end_delimiter,$string,$modifier) = @_;

    # split regexp
    my @char = $string =~ /\G(
        \\\\        |
        [\$\@\/\\]  |
        [\x00-\xFF]
    )/oxmsg;

    # unescape character
    for (my $i=0; $i <= $#char; $i++) {
        if (0) {
        }

        # remain \\
        elsif ($char[$i] eq '\\\\') {
        }

        # escape $ @ / and \
        elsif ($char[$i] =~ /\A [\$\@\/\\] \z/oxms) {
            $char[$i] = '\\' . $char[$i];
        }
    }

    $delimiter     = '/';
    $end_delimiter = '/';
    my $prematch = '';
    return join '', $ope, $delimiter, $prematch, '(?:', @char, ')', $matched, $end_delimiter, $modifier;
}

#
# escape regexp (s''here')
#
sub e_s2_q {
    my($ope,$delimiter,$end_delimiter,$string) = @_;

    $slash = 'div';

    my @char = $string =~ / \G (\\\\|[\$\@\/\\]|$q_char) /oxmsg;
    for (my $i=0; $i <= $#char; $i++) {
        if (0) {
        }

        # not escape \\
        elsif ($char[$i] =~ /\A \\\\ \z/oxms) {
        }

        # escape $ @ / and \
        elsif ($char[$i] =~ /\A [\$\@\/\\] \z/oxms) {
            $char[$i] = '\\' . $char[$i];
        }
    }

    return join '', $ope, $delimiter, @char,   $end_delimiter;
}

#
# escape regexp (s/here/and here/modifier)
#
sub e_sub {
    my($variable,$delimiter1,$pattern,$end_delimiter1,$delimiter2,$replacement,$end_delimiter2,$modifier) = @_;
    $modifier ||= '';

    $modifier =~ tr/p//d;
    if ($modifier =~ /([adlu])/oxms) {
        my $line = 0;
        for (my $i=0; my($package,$filename,$use_line,$subroutine) = caller($i); $i++) {
            if ($filename ne __FILE__) {
                $line = $use_line + (CORE::substr($_,0,pos($_)) =~ tr/\n//) + 1;
                last;
            }
        }
        die qq{Unsupported modifier "$1" used at line $line.\n};
    }

    if ($variable eq '') {
        $variable      = '$_';
        $bind_operator = ' =~ ';
    }

    $slash = 'div';

    # P.128 Start of match (or end of previous match): \G
    # P.130 Advanced Use of \G with Perl
    # in Chapter 3: Overview of Regular Expression Features and Flavors
    # P.312 Iterative Matching: Scalar Context, with /g
    # in Chapter 7: Perl
    # of ISBN 0-596-00289-0 Mastering Regular Expressions, Second edition

    # P.181 Where You Left Off: The \G Assertion
    # in Chapter 5: Pattern Matching
    # of ISBN 0-596-00027-8 Programming Perl Third Edition.

    # P.220 Where You Left Off: The \G Assertion
    # in Chapter 5: Pattern Matching
    # of ISBN 978-0-596-00492-7 Programming Perl 4th Edition.

    my $e_modifier = $modifier =~ tr/e//d;
    my $r_modifier = $modifier =~ tr/r//d;

    my $my = '';
    if ($variable =~ s/\A \( \s* ( (?: local \b | my \b | our \b | state \b )? .+ ) \) \z/$1/oxms) {
        $my = $variable;
        $variable =~ s/ (?: local \b | my \b | our \b | state \b ) \s* //oxms;
        $variable =~ s/ = .+ \z//oxms;
    }

    (my $variable_basename = $variable) =~ s/ [\[\{].* \z//oxms;
    $variable_basename =~ s/ \s+ \z//oxms;

    # quote replacement string
    my $e_replacement = '';
    if ($e_modifier >= 1) {
        $e_replacement = e_qq('', '', '', $replacement);
        $e_modifier--;
    }
    else {
        if ($delimiter2 eq "'") {
            $e_replacement = e_s2_q('qq', '/',         '/',             $replacement);
        }
        else {
            $e_replacement = e_qq  ('qq', $delimiter2, $end_delimiter2, $replacement);
        }
    }

    my $sub = '';

    # with /r
    if ($r_modifier) {
        if (0) {
        }

        # s///gr without multibyte anchoring
        elsif ($modifier =~ /g/oxms) {
            $sub = sprintf(
                #                        1                        2   3                                 4   5
                q,

                $variable,                                                       #  1
                ($delimiter1 eq "'") ?                                           #  2
                e_s1_q('m', $delimiter1, $end_delimiter1, $pattern, $modifier) : #  :
                e_s1  ('m', $delimiter1, $end_delimiter1, $pattern, $modifier),  #  :
                $s_matched,                                                      #  3
                $e_replacement,                                                  #  4
                '$Char::Arabic::re_r=eval $Char::Arabic::re_r; ' x $e_modifier,                  #  5
            );
        }

        # s///r
        else {

            my $prematch = q{$`};

            $sub = sprintf(
                #  1     2          3                                 4   5  6                    7
                q<(%s =~ %s) ? eval{%s local $^W=0; local $Char::Arabic::re_r=%s; %s"%s$Char::Arabic::re_r$'" } : %s>,

                $variable,                                                       #  1
                ($delimiter1 eq "'") ?                                           #  2
                e_s1_q('m', $delimiter1, $end_delimiter1, $pattern, $modifier) : #  :
                e_s1  ('m', $delimiter1, $end_delimiter1, $pattern, $modifier),  #  :
                $s_matched,                                                      #  3
                $e_replacement,                                                  #  4
                '$Char::Arabic::re_r=eval $Char::Arabic::re_r; ' x $e_modifier,                  #  5
                $prematch,                                                       #  6
                $variable,                                                       #  7
            );
        }

        # $var !~ s///r doesn't make sense
        if ($bind_operator =~ / !~ /oxms) {
            $sub = q{die("$0: Using !~ with s///r doesn't make sense"), } . $sub;
        }
    }

    # without /r
    else {
        if (0) {
        }

        # s///g without multibyte anchoring
        elsif ($modifier =~ /g/oxms) {
            $sub = sprintf(
                #                                 1     2   3                                 4   5 6                         7                                                 8
                q,

                $variable,                                                       #  1
                ($delimiter1 eq "'") ?                                           #  2
                e_s1_q('m', $delimiter1, $end_delimiter1, $pattern, $modifier) : #  :
                e_s1  ('m', $delimiter1, $end_delimiter1, $pattern, $modifier),  #  :
                $s_matched,                                                      #  3
                $e_replacement,                                                  #  4
                '$Char::Arabic::re_r=eval $Char::Arabic::re_r; ' x $e_modifier,                  #  5
                $variable,                                                       #  6
                $variable,                                                       #  7
                ($bind_operator =~ / !~ /oxms) ? '!' : '',                       #  8
            );
        }

        # s///
        else {

            my $prematch = q{$`};

            $sub = sprintf(

                ($bind_operator =~ / =~ /oxms) ?

                #  1 2 3          4                                 5   6 7   8
                q<(%s%s%s) ? eval{%s local $^W=0; local $Char::Arabic::re_r=%s; %s%s="%s$Char::Arabic::re_r$'"; 1 } : undef> :

                #  1 2 3              4                                 5   6 7   8
                q<(%s%s%s) ? 1 : eval{%s local $^W=0; local $Char::Arabic::re_r=%s; %s%s="%s$Char::Arabic::re_r$'"; undef }>,

                $variable,                                                       #  1
                $bind_operator,                                                  #  2
                ($delimiter1 eq "'") ?                                           #  3
                e_s1_q('m', $delimiter1, $end_delimiter1, $pattern, $modifier) : #  :
                e_s1  ('m', $delimiter1, $end_delimiter1, $pattern, $modifier),  #  :
                $s_matched,                                                      #  4
                $e_replacement,                                                  #  5
                '$Char::Arabic::re_r=eval $Char::Arabic::re_r; ' x $e_modifier,                  #  6
                $variable,                                                       #  7
                $prematch,                                                       #  8
            );
        }
    }

    # (my $foo = $bar) =~ s///   -->   (my $foo = $bar, eval { ... })[1]
    if ($my ne '') {
        $sub = "($my, $sub)[1]";
    }

    # clear s/// variable
    $sub_variable = '';
    $bind_operator = '';

    return $sub;
}

#
# escape regexp of split qr//
#
sub e_split {
    my($ope,$delimiter,$end_delimiter,$string,$modifier) = @_;
    $modifier ||= '';

    $modifier =~ tr/p//d;
    if ($modifier =~ /([adlu])/oxms) {
        my $line = 0;
        for (my $i=0; my($package,$filename,$use_line,$subroutine) = caller($i); $i++) {
            if ($filename ne __FILE__) {
                $line = $use_line + (CORE::substr($_,0,pos($_)) =~ tr/\n//) + 1;
                last;
            }
        }
        die qq{Unsupported modifier "$1" used at line $line.\n};
    }

    $slash = 'div';

    # /b /B modifier
    if ($modifier =~ tr/bB//d) {
        return join '', 'split', $ope, $delimiter, $string, $end_delimiter, $modifier;
    }

    my $ignorecase = ($modifier =~ /i/oxms) ? 1 : 0;
    my $metachar = qr/[\@\\|[\]{^]/oxms;

    # split regexp
    my @char = $string =~ /\G(
        \\o\{ [0-7]+           \}  |
        \\    [0-7]{2,3}           |
        \\x\{ [0-9A-Fa-f]+     \}  |
        \\x   [0-9A-Fa-f]{1,2}     |
        \\c   [\x40-\x5F]          |
        \\N\{ [^0-9\}][^\}]* \} |
        \\p\{ [^0-9\}][^\}]* \} |
        \\P\{ [^0-9\}][^\}]* \} |
        \\  (?:$q_char)            |
        \$` | \$\{`\} | \$ \s* PREMATCH  | \$ \s* \{ \s* PREMATCH  \s* \} | \$ \s* \{\^PREMATCH\}  |
        \$& | \$\{&\} | \$ \s* MATCH     | \$ \s* \{ \s* MATCH     \s* \} | \$ \s* \{\^MATCH\}     |
                        \$ \s* POSTMATCH | \$ \s* \{ \s* POSTMATCH \s* \} | \$ \s* \{\^POSTMATCH\} |
        [\$\@] $qq_variable        |
        \$ \s* \d+                 |
        \$ \s* \{ \s* \d+ \s* \}   |
        \$ \$ (?![\w\{])           |
        \$ \s* \$ \s* $qq_variable |
        \[\:\^ (?:alnum|alpha|ascii|blank|cntrl|digit|graph|lower|print|punct|space|upper|word|xdigit) :\] |
        \[\:   (?:alnum|alpha|ascii|blank|cntrl|digit|graph|lower|print|punct|space|upper|word|xdigit) :\] |
        \[\^                       |
        \(\?                       |
            (?:$q_char)
    )/oxmsg;

    my $left_e  = 0;
    my $right_e = 0;
    for (my $i=0; $i <= $#char; $i++) {

        # "\L\u" --> "\u\L"
        if (($char[$i] eq '\L') and ($char[$i+1] eq '\u')) {
            @char[$i,$i+1] = @char[$i+1,$i];
        }

        # "\U\l" --> "\l\U"
        elsif (($char[$i] eq '\U') and ($char[$i+1] eq '\l')) {
            @char[$i,$i+1] = @char[$i+1,$i];
        }

        # octal escape sequence
        elsif ($char[$i] =~ /\A \\o \{ ([0-7]+) \} \z/oxms) {
            $char[$i] = Char::Earabic::octchr($1);
        }

        # hexadecimal escape sequence
        elsif ($char[$i] =~ /\A \\x \{ ([0-9A-Fa-f]+) \} \z/oxms) {
            $char[$i] = Char::Earabic::hexchr($1);
        }

        # \N{CHARNAME} --> N\{CHARNAME}
        elsif ($char[$i] =~ /\A \\ (N) ( \{ ([^0-9\}][^\}]*) \} ) \z/oxms) {
            $char[$i] = $1 . '\\' . $2;
        }

        # \p{PROPERTY} --> p\{PROPERTY}
        elsif ($char[$i] =~ /\A \\ (p) ( \{ ([^0-9\}][^\}]*) \} ) \z/oxms) {
            $char[$i] = $1 . '\\' . $2;
        }

        # \P{PROPERTY} --> P\{PROPERTY}
        elsif ($char[$i] =~ /\A \\ (P) ( \{ ([^0-9\}][^\}]*) \} ) \z/oxms) {
            $char[$i] = $1 . '\\' . $2;
        }

        # \p, \P, \X --> p, P, X
        elsif ($char[$i] =~ /\A \\ ( [pPX] ) \z/oxms) {
            $char[$i] = $1;
        }

        if (0) {
        }

        # join separated multiple-octet
        elsif ($char[$i] =~ /\A (?: \\ [0-7]{2,3} | \\x [0-9A-Fa-f]{1,2}) \z/oxms) {
            if (   ($i+3 <= $#char) and (grep(/\A (?: \\ [0-7]{2,3} | \\x [0-9A-Fa-f]{1,2}) \z/oxms, @char[$i+1..$i+3]) == 3) and (eval(sprintf '"%s%s%s%s"', @char[$i..$i+3]) =~ /\A $q_char \z/oxms)) {
                $char[$i] .= join '', splice @char, $i+1, 3;
            }
            elsif (($i+2 <= $#char) and (grep(/\A (?: \\ [0-7]{2,3} | \\x [0-9A-Fa-f]{1,2}) \z/oxms, @char[$i+1..$i+2]) == 2) and (eval(sprintf '"%s%s%s"',   @char[$i..$i+2]) =~ /\A $q_char \z/oxms)) {
                $char[$i] .= join '', splice @char, $i+1, 2;
            }
            elsif (($i+1 <= $#char) and (grep(/\A (?: \\ [0-7]{2,3} | \\x [0-9A-Fa-f]{1,2}) \z/oxms, $char[$i+1      ]) == 1) and (eval(sprintf '"%s%s"',     @char[$i..$i+1]) =~ /\A $q_char \z/oxms)) {
                $char[$i] .= join '', splice @char, $i+1, 1;
            }
        }

        # open character class [...]
        elsif ($char[$i] eq '[') {
            my $left = $i;
            if ($char[$i+1] eq ']') {
                $i++;
            }
            while (1) {
                if (++$i > $#char) {
                    die __FILE__, ": Unmatched [] in regexp";
                }
                if ($char[$i] eq ']') {
                    my $right = $i;

                    # [...]
                    if (grep(/\A [\$\@]/oxms,@char[$left+1..$right-1]) >= 1) {
                        splice @char, $left, $right-$left+1, sprintf(q{@{[Char::Earabic::charlist_qr(%s,'%s')]}}, join(',', map {qq_stuff($delimiter,$end_delimiter,$_)} @char[$left+1..$right-1]), $modifier);
                    }
                    else {
                        splice @char, $left, $right-$left+1, Char::Earabic::charlist_qr(@char[$left+1..$right-1], $modifier);
                    }

                    $i = $left;
                    last;
                }
            }
        }

        # open character class [^...]
        elsif ($char[$i] eq '[^') {
            my $left = $i;
            if ($char[$i+1] eq ']') {
                $i++;
            }
            while (1) {
                if (++$i > $#char) {
                    die __FILE__, ": Unmatched [] in regexp";
                }
                if ($char[$i] eq ']') {
                    my $right = $i;

                    # [^...]
                    if (grep(/\A [\$\@]/oxms,@char[$left+1..$right-1]) >= 1) {
                        splice @char, $left, $right-$left+1, sprintf(q{@{[Char::Earabic::charlist_not_qr(%s,'%s')]}}, join(',', map {qq_stuff($delimiter,$end_delimiter,$_)} @char[$left+1..$right-1]), $modifier);
                    }
                    else {
                        splice @char, $left, $right-$left+1, Char::Earabic::charlist_not_qr(@char[$left+1..$right-1], $modifier);
                    }

                    $i = $left;
                    last;
                }
            }
        }

        # rewrite character class or escape character
        elsif (my $char = character_class($char[$i],$modifier)) {
            $char[$i] = $char;
        }

        # P.794 29.2.161. split
        # in Chapter 29: Functions
        # of ISBN 0-596-00027-8 Programming Perl Third Edition.

        # P.951 split
        # in Chapter 27: Functions
        # of ISBN 978-0-596-00492-7 Programming Perl 4th Edition.

        # said "The //m modifier is assumed when you split on the pattern /^/",
        # but perl5.008 is not so. Therefore, this software adds //m.
        # (and so on)

        # split(m/^/) --> split(m/^/m)
        elsif (($char[$i] eq '^') and ($modifier !~ /m/oxms)) {
            $modifier .= 'm';
        }

        # /i modifier
        elsif ($ignorecase and ($char[$i] =~ /\A [\x00-\xFF] \z/oxms) and (Char::Earabic::uc($char[$i]) ne Char::Earabic::fc($char[$i]))) {
            if (CORE::length(Char::Earabic::fc($char[$i])) == 1) {
                $char[$i] = '['   . Char::Earabic::uc($char[$i])       . Char::Earabic::fc($char[$i]) . ']';
            }
            else {
                $char[$i] = '(?:' . Char::Earabic::uc($char[$i]) . '|' . Char::Earabic::fc($char[$i]) . ')';
            }
        }

        # \u \l \U \L \F \Q \E
        elsif ($char[$i] =~ /\A ([<>]) \z/oxms) {
            if ($right_e < $left_e) {
                $char[$i] = '\\' . $char[$i];
            }
        }
        elsif ($char[$i] eq '\u') {
            $char[$i] = '@{[Char::Earabic::ucfirst qq<';
            $left_e++;
        }
        elsif ($char[$i] eq '\l') {
            $char[$i] = '@{[Char::Earabic::lcfirst qq<';
            $left_e++;
        }
        elsif ($char[$i] eq '\U') {
            $char[$i] = '@{[Char::Earabic::uc qq<';
            $left_e++;
        }
        elsif ($char[$i] eq '\L') {
            $char[$i] = '@{[Char::Earabic::lc qq<';
            $left_e++;
        }
        elsif ($char[$i] eq '\F') {
            $char[$i] = '@{[Char::Earabic::fc qq<';
            $left_e++;
        }
        elsif ($char[$i] eq '\Q') {
            $char[$i] = '@{[CORE::quotemeta qq<';
            $left_e++;
        }
        elsif ($char[$i] eq '\E') {
            if ($right_e < $left_e) {
                $char[$i] = '>]}';
                $right_e++;
            }
            else {
                $char[$i] = '';
            }
        }
        elsif ($char[$i] eq '\Q') {
            while (1) {
                if (++$i > $#char) {
                    last;
                }
                if ($char[$i] eq '\E') {
                    last;
                }
            }
        }
        elsif ($char[$i] eq '\E') {
        }

        # $0 --> $0
        elsif ($char[$i] =~ /\A \$ 0 \z/oxms) {
            if ($ignorecase) {
                $char[$i] = '@{[Char::Earabic::ignorecase(' . $char[$i] . ')]}';
            }
        }
        elsif ($char[$i] =~ /\A \$ \{ \s* 0 \s* \} \z/oxms) {
            if ($ignorecase) {
                $char[$i] = '@{[Char::Earabic::ignorecase(' . $char[$i] . ')]}';
            }
        }

        # $$ --> $$
        elsif ($char[$i] =~ /\A \$\$ \z/oxms) {
        }

        # $1, $2, $3 --> $2, $3, $4 after s/// with multibyte anchoring
        # $1, $2, $3 --> $1, $2, $3 otherwise
        elsif ($char[$i] =~ /\A \$ ([1-9][0-9]*) \z/oxms) {
            $char[$i] = e_capture($1);
            if ($ignorecase) {
                $char[$i] = '@{[Char::Earabic::ignorecase(' . $char[$i] . ')]}';
            }
        }
        elsif ($char[$i] =~ /\A \$ \{ \s* ([1-9][0-9]*) \s* \} \z/oxms) {
            $char[$i] = e_capture($1);
            if ($ignorecase) {
                $char[$i] = '@{[Char::Earabic::ignorecase(' . $char[$i] . ')]}';
            }
        }

        # $$foo[ ... ] --> $ $foo->[ ... ]
        elsif ($char[$i] =~ /\A \$ ( \$ [A-Za-z_][A-Za-z0-9_]*(?: ::[A-Za-z_][A-Za-z0-9_]*)* ) ( \[ (?:$qq_bracket)*? \] ) \z/oxms) {
            $char[$i] = e_capture($1.'->'.$2);
            if ($ignorecase) {
                $char[$i] = '@{[Char::Earabic::ignorecase(' . $char[$i] . ')]}';
            }
        }

        # $$foo{ ... } --> $ $foo->{ ... }
        elsif ($char[$i] =~ /\A \$ ( \$ [A-Za-z_][A-Za-z0-9_]*(?: ::[A-Za-z_][A-Za-z0-9_]*)* ) ( \{ (?:$qq_brace)*? \} ) \z/oxms) {
            $char[$i] = e_capture($1.'->'.$2);
            if ($ignorecase) {
                $char[$i] = '@{[Char::Earabic::ignorecase(' . $char[$i] . ')]}';
            }
        }

        # $$foo
        elsif ($char[$i] =~ /\A \$ ( \$ [A-Za-z_][A-Za-z0-9_]*(?: ::[A-Za-z_][A-Za-z0-9_]*)* ) \z/oxms) {
            $char[$i] = e_capture($1);
            if ($ignorecase) {
                $char[$i] = '@{[Char::Earabic::ignorecase(' . $char[$i] . ')]}';
            }
        }

        # $`, ${`}, $PREMATCH, ${PREMATCH}, ${^PREMATCH} --> Char::Earabic::PREMATCH()
        elsif ($char[$i] =~ /\A ( \$` | \$\{`\} | \$ \s* PREMATCH  | \$ \s* \{ \s* PREMATCH  \s* \} | \$ \s* \{\^PREMATCH\}  ) \z/oxmsgc) {
            if ($ignorecase) {
                $char[$i] = '@{[Char::Earabic::ignorecase(Char::Earabic::PREMATCH())]}';
            }
            else {
                $char[$i] = '@{[Char::Earabic::PREMATCH()]}';
            }
        }

        # $&, ${&}, $MATCH, ${MATCH}, ${^MATCH} --> Char::Earabic::MATCH()
        elsif ($char[$i] =~ /\A ( \$& | \$\{&\} | \$ \s* MATCH     | \$ \s* \{ \s* MATCH     \s* \} | \$ \s* \{\^MATCH\}     ) \z/oxmsgc) {
            if ($ignorecase) {
                $char[$i] = '@{[Char::Earabic::ignorecase(Char::Earabic::MATCH())]}';
            }
            else {
                $char[$i] = '@{[Char::Earabic::MATCH()]}';
            }
        }

        # $POSTMATCH, ${POSTMATCH}, ${^POSTMATCH} --> Char::Earabic::POSTMATCH()
        elsif ($char[$i] =~ /\A (                 \$ \s* POSTMATCH | \$ \s* \{ \s* POSTMATCH \s* \} | \$ \s* \{\^POSTMATCH\} ) \z/oxmsgc) {
            if ($ignorecase) {
                $char[$i] = '@{[Char::Earabic::ignorecase(Char::Earabic::POSTMATCH())]}';
            }
            else {
                $char[$i] = '@{[Char::Earabic::POSTMATCH()]}';
            }
        }

        # ${ foo }
        elsif ($char[$i] =~ /\A \$ \s* \{ ( \s* [A-Za-z_][A-Za-z0-9_]*(?: ::[A-Za-z_][A-Za-z0-9_]*)* \s* ) \} \z/oxms) {
            if ($ignorecase) {
                $char[$i] = '@{[Char::Earabic::ignorecase(' . $1 . ')]}';
            }
        }

        # ${ ... }
        elsif ($char[$i] =~ /\A \$ \s* \{ \s* ( .+ ) \s* \} \z/oxms) {
            $char[$i] = e_capture($1);
            if ($ignorecase) {
                $char[$i] = '@{[Char::Earabic::ignorecase(' . $char[$i] . ')]}';
            }
        }

        # $scalar or @array
        elsif ($char[$i] =~ /\A [\$\@].+ /oxms) {
            $char[$i] = e_string($char[$i]);
            if ($ignorecase) {
                $char[$i] = '@{[Char::Earabic::ignorecase(' . $char[$i] . ')]}';
            }
        }

        # quote character before ? + * {
        elsif (($i >= 1) and ($char[$i] =~ /\A [\?\+\*\{] \z/oxms)) {
            if ($char[$i-1] =~ /\A (?:\\[0-7]{2,3}|\\x[0-9-A-Fa-f]{1,2}) \z/oxms) {
            }
            else {
                $char[$i-1] = '(?:' . $char[$i-1] . ')';
            }
        }
    }

    # make regexp string
    $modifier =~ tr/i//d;
    if ($left_e > $right_e) {
        return join '', 'Char::Earabic::split', $ope, $delimiter, @char, '>]}' x ($left_e - $right_e), $end_delimiter, $modifier;
    }
    return     join '', 'Char::Earabic::split', $ope, $delimiter, @char,                               $end_delimiter, $modifier;
}

#
# escape regexp of split qr''
#
sub e_split_q {
    my($ope,$delimiter,$end_delimiter,$string,$modifier) = @_;
    $modifier ||= '';

    $modifier =~ tr/p//d;
    if ($modifier =~ /([adlu])/oxms) {
        my $line = 0;
        for (my $i=0; my($package,$filename,$use_line,$subroutine) = caller($i); $i++) {
            if ($filename ne __FILE__) {
                $line = $use_line + (CORE::substr($_,0,pos($_)) =~ tr/\n//) + 1;
                last;
            }
        }
        die qq{Unsupported modifier "$1" used at line $line.\n};
    }

    $slash = 'div';

    # /b /B modifier
    if ($modifier =~ tr/bB//d) {
        return join '', 'split', $ope, $delimiter, $string, $end_delimiter, $modifier;
    }

    my $ignorecase = ($modifier =~ /i/oxms) ? 1 : 0;

    # split regexp
    my @char = $string =~ /\G(
        \[\:\^ [a-z]+ \:\] |
        \[\:   [a-z]+ \:\] |
        \[\^               |
        \\?    (?:$q_char)
    )/oxmsg;

    # unescape character
    for (my $i=0; $i <= $#char; $i++) {
        if (0) {
        }

        # open character class [...]
        elsif ($char[$i] eq '[') {
            my $left = $i;
            if ($char[$i+1] eq ']') {
                $i++;
            }
            while (1) {
                if (++$i > $#char) {
                    die __FILE__, ": Unmatched [] in regexp";
                }
                if ($char[$i] eq ']') {
                    my $right = $i;

                    # [...]
                    splice @char, $left, $right-$left+1, Char::Earabic::charlist_qr(@char[$left+1..$right-1], $modifier);

                    $i = $left;
                    last;
                }
            }
        }

        # open character class [^...]
        elsif ($char[$i] eq '[^') {
            my $left = $i;
            if ($char[$i+1] eq ']') {
                $i++;
            }
            while (1) {
                if (++$i > $#char) {
                    die __FILE__, ": Unmatched [] in regexp";
                }
                if ($char[$i] eq ']') {
                    my $right = $i;

                    # [^...]
                    splice @char, $left, $right-$left+1, Char::Earabic::charlist_not_qr(@char[$left+1..$right-1], $modifier);

                    $i = $left;
                    last;
                }
            }
        }

        # rewrite character class or escape character
        elsif (my $char = character_class($char[$i],$modifier)) {
            $char[$i] = $char;
        }

        # split(m/^/) --> split(m/^/m)
        elsif (($char[$i] eq '^') and ($modifier !~ /m/oxms)) {
            $modifier .= 'm';
        }

        # /i modifier
        elsif ($ignorecase and ($char[$i] =~ /\A [\x00-\xFF] \z/oxms) and (Char::Earabic::uc($char[$i]) ne Char::Earabic::fc($char[$i]))) {
            if (CORE::length(Char::Earabic::fc($char[$i])) == 1) {
                $char[$i] = '['   . Char::Earabic::uc($char[$i])       . Char::Earabic::fc($char[$i]) . ']';
            }
            else {
                $char[$i] = '(?:' . Char::Earabic::uc($char[$i]) . '|' . Char::Earabic::fc($char[$i]) . ')';
            }
        }

        # quote character before ? + * {
        elsif (($i >= 1) and ($char[$i] =~ /\A [\?\+\*\{] \z/oxms)) {
            if ($char[$i-1] =~ /\A [\x00-\xFF] \z/oxms) {
            }
            else {
                $char[$i-1] = '(?:' . $char[$i-1] . ')';
            }
        }
    }

    $modifier =~ tr/i//d;
    return join '', 'Char::Earabic::split', $ope, $delimiter, @char, $end_delimiter, $modifier;
}

1;

__END__