Author image Flávio Soibelmann Glock


Pugs::Compiler::Regex - Compiler for Perl 6 Regex


    use Pugs::Compiler::Regex;
    use Pugs::Runtime::Match;

    $regex = Pugs::Compiler::Regex->compile('a*b');
    $match = $regex->match('aaab');
    print $match->(), "\n";
    print $match->from, "\n";
    print $match->to, "\n";

    package MyGrammar;
    $regex = Pugs::Compiler::Regex->compile(
        'a*', { ratchet => 1, continue => 1, sigspace => 1 }
    *my_match = $regex->code();
    $match = MyGrammar->my_match('aaaa');
    print "$match\n";

    package MyGrammar2;
    $regex = Pugs::Compiler::Regex->install(
        my_match => 'a*',
        { ratchet => 1, continue => 1, sigspace => 1 }
    $match = MyGrammar->my_match('aaaa');
    print "$match\n";


This class provides an implementation for Perl 6 regexes. It serves as a base class for Pugs::Compiler::Rule and Pugs::Compiler::Token.


$regex = Pugs::Compiler::Regex->compile($str, $params);

This method acts like a constructor, which returns a Pugs::Compiler::Regex object from the p6 regex specified in $str, or throws an exception on invalid rule syntax.

$params is an optional argument which specifies the following p6 regex modifiers:

grammar => 'Pugs::Grammar::Base'

Specify which namespace (Grammar) the rule belongs to. if grammar is not specified, then "Pugs::Grammar::Base" will be assumed.

continue => 0
c => 0

These modifiers cause the pattern to continue scanning from the string's current pos:

Note that in Perl 6 form


is roughlh equivalent to

    m:p/ .*? pattern /

Here is an example:

    package Foo;
        word => '\w+',
        { ratchet => 1, continue => 1 }
    $s = 'hello world';
    $match = Foo->word($s); # got 'hello'
    $match = Foo->word($s); # got 'world'

XXX Note that continue or c currently are not supported in non-ratchet mode.

ratchet => 0

Disable backtracking. Much faster. Defaults to 0. (Pugs::Compiler::Rule and Pugs::Compiler::Token have the default value of 1).

pos => undef
p => undef

Specify a string position to match. Starts from zero. Defaults to undef, which matches anywhere in the string.

sigspace => 0

Whitespace is significant. Defaults to 0 for Pugs::Compiler::Regex while 1 for Pugs::Compiler::Rule.

ignorecase => 0

Ignore character case. Defaults to 0 for Pugs::Compiler::Regex.


Return a string holding the Perl 5 code for reconstructing the current Pugs::Compiler::Regex object.

We are trying to make the perl5 method does something like serializing a Pugs::Compiler::Regex instance.

If you want the raw Perl 5 code generated by the various emitters, use $regex->{perl5} directly.

$regex->match($str, $grammar, $params)

Performs matching action on $str. Note that it's a thin wrapper around the p5 code compiled from the p6 regex with run-time modifier handling via the $params argument.

Here is an example:

    $regex = Pugs::Compiler::regex->compile('a*\w');
    my $match = $regex->match('aaa');
    print "Capture: $match";
    print "From: ", $match->from;
    print "To: ", $match->to;

Returns a closure (or an anonymous sub) which does the actual matching task. For example:

  $regex = Pugs::Compiler::Regex->compile('a|b', { ratchet => 1 });
  my $sub = $regex->code();
  my $match = $sub->('MyGrammar', 'aaa');

Or inserts the sub into the current package:

  package Foo;
  $regex = Pugs::Compiler::Regex->compile('a*');
  *match = $regex->code();
  my $match = __PACKAGE__->match('aaa');

Technically it's a thin wrapper around the match method.

$regex->install($name, @args_for_compile)

Installs the subroutine returned from the code method as a named subroutine using the name specified by $name;

If $name is fully qualified, then the corresponding package is used, otherwise the current package is assumed.

@args_for_compile are those arguments fed into the compile method.

It will croak if there's already a sub with the same name exists. If that's not what you want, use the reinstall method instead.

Here are some examples:

  package Bar;
  Pugs::Compiler::Regex->install(match => 'a*', {ratchet => 1});
  $match = Bar->match('aaa');

  # The following line dies with the message
  # "Can't install regex 'match' as 'Bar::match' which already
  # exists":
  Pugs::Compiler::Regex->install(match => 'b*');
$regex->reinstall($name, @args_for_compile)

Like the install method but it can replaces the named sub if it already exists:

  package Bar;
  Pugs::Compiler::Regex->install('match', 'a*', {ratchet => 1});
  Pugs::Compiler::Regex->reinstall('match', 'b*');
  $match = Bar->match('bbb'); # matched here



By default, the compile method will cache the compiled form (p5 source) of the p6 regex. The NoCache variable prevents any caching.


The Pugs Team <>.


The Perl 6 Rules Spec:


Copyright 2006 by Flavio Soibelmann Glock and others.

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