libregexp-assemble-perl 0.35-8 / usr / share / doc / libregexp-assemble-perl / examples / naive

#!/usr/bin/perl -w
#
# naive - generate a regular expression that will lex regular expressions
# it is naive in the sense that it can't lex certain patterns, such as a
# pattern that contains nested parentheses (e.g. /ab(?:cd(?ef)?gh)+ij/ )
#
# The emitted regular expression is the default pattern that Regexp::Assemble
# uses to pick apart a string into tokens suitable for being assembled. If it
# isn't sufficiently sophisticated, you will have to supply your own lexer.
#
# Copyright (C) David Landgren 2004-2005

use strict;

my $directive = q{[bABCEGLQUXZ]};

my $punct  = q{[^\\w]};

my $meta   = q{[aefnrtdDwWsS]};
my $ctrl   = q{c.};
my $octal  = q{0\\d{2}};
my $hex    = q{x(?:[\\da-fA-F]{2}|{[\\da-fA-F]{4}})};
my $named  = q{N\\{\w+\\}};
my $prop   = q{[Pp](?:\\{\w+\\}|.)};

my $single = q{[^\\w\\/{|}-]};

my $modifiable = qq{$punct|$meta|$ctrl|$octal|$hex|$named|$prop};

my $modifier   = q{(?:[*+?]\\??|\\{\\d+(?:,\\d*)?\\}\\??)?};

my $backslash  = qq{\\\\(?:$directive|[lu].|(?:$modifiable)$modifier)};

my $class  = q{\\[.*?(?<!\\\\)\\]};
my $group  = q{\\(.*?(?<!\\\\)\\)};
# \$Default_Lexer = qr/(?![[(\\\\]).$modifier|$backslash|$class$modifier|$group$modifier/;

print <<PATTS;
# The following patterns were generated with eg/naive
\$Default_Lexer = qr/(?![[(\\\\]).$modifier|$backslash|$class$modifier|$group$modifier/;

\$Single_Char   = qr/^(?:\\\\(?:$meta|$ctrl|$single|$octal|$hex)|[^\$^])\$/;
PATTS

# my brain hurts just looking at this