General syntax for requesting patterns

To access a particular pattern, %RE is treated as a hierarchical hash of hashes (of hashes...), with each successive key being an identifier. For example, to access the pattern that matches real numbers, you specify:

        $RE{num}{real}

and to access the pattern that matches integers:

        $RE{num}{int}

Deeper layers of the hash are used to specify flags: arguments that modify the resulting pattern in some way. The keys used to access these layers are prefixed with a minus sign and may have a value; if a value is given, it's done by using a multidimensional key. For example, to access the pattern that matches base-2 real numbers with embedded commas separating groups of three digits (e.g. 10,101,110.110101101):

        $RE{num}{real}{-base => 2}{-sep => ','}{-group => 3}

Through the magic of Perl, these flag layers may be specified in any order (and even interspersed through the identifier keys!) so you could get the same pattern with:

        $RE{num}{real}{-sep => ','}{-group => 3}{-base => 2}

or:

        $RE{num}{-base => 2}{real}{-group => 3}{-sep => ','}

or even:

        $RE{-base => 2}{-group => 3}{-sep => ','}{num}{real}

etc.

Note, however, that the relative order of amongst the identifier keys is significant. That is:

        $RE{list}{set}

would not be the same as:

        $RE{set}{list}

Flag syntax

In versions prior to 2.113, flags could also be written as "{"-flag=value"}". This no longer works, although "{"-flag$;value"}" still does. However, "{-flag => 'value'}" is the preferred syntax.

Universal flags

Normally, flags are specific to a single pattern. However, there is two flags that all patterns may specify.

"-keep"

By default, the patterns provided by %RE contain no capturing parentheses. However, if the "-keep" flag is specified (it requires no value) then any significant substrings that the pattern matches are captured. For example:

        if ($str =~ $RE{num}{real}{-keep}) {
                $number   = $1;
                $whole    = $3;
                $decimals = $5;
        }
    

Special care is needed if a "kept" pattern is interpolated into a larger regular expression, as the presence of other capturing parentheses is likely to change the "number variables" into which significant substrings are saved.

See also "Adding new regular expressions", which describes how to create new patterns with "optional" capturing brackets that respond to "-keep".

"-i"

Some patterns or subpatterns only match lowercase or uppercase letters. If one wants the do case insensitive matching, one option is to use the "/i" regexp modifier, or the special sequence "(?i)". But if the functional interface is used, one does not have this option. The "-i" switch solves this problem; by using it, the pattern will do case insensitive matching.

OO interface and inline matching/substitution

The patterns returned from %RE are objects, so rather than writing:

        if ($str =~ /$RE{some}{pattern}/ ) {...}

you can write:

        if ( $RE{some}{pattern}->matches($str) ) {...}

For matching this would seem to have no great advantage apart from readability (but see below).

For substitutions, it has other significant benefits. Frequently you want to perform a substitution on a string without changing the original. Most people use this:

        $changed = $original;
        $changed =~ s/$RE{some}{pattern}/$replacement/;

The more adept use:

        ($changed = $original) =~ s/$RE{some}{pattern}/$replacement/;

Regexp::Common allows you do write this:

        $changed = $RE{some}{pattern}->subs($original=>$replacement);

Apart from reducing precedence-angst, this approach has the added advantages that the substitution behaviour can be optimized from the regular expression, and the replacement string can be provided by default (see "Adding new regular expressions").

For example, in the implementation of this substitution:

        $cropped = $RE{ws}{crop}->subs($uncropped);

the default empty string is provided automatically, and the substitution is optimized to use:

        $uncropped =~ s/^\s+//;
        $uncropped =~ s/\s+$//;

rather than:

        $uncropped =~ s/^\s+|\s+$//g;

Subroutine-based interface

The hash-based interface was chosen because it allows regexes to be effortlessly interpolated, and because it also allows them to be "curried". For example:

        my $num = $RE{num}{int};
        my $commad     = $num->{-sep=>','}{-group=>3};
        my $duodecimal = $num->{-base=>12};

However, the use of tied hashes does make the access to Regexp::Common patterns slower than it might otherwise be. In contexts where impatience overrules laziness, Regexp::Common provides an additional subroutine-based interface.

For each (sub-)entry in the %RE hash ($RE{key1}{key2}{etc}), there is a corresponding exportable subroutine: "RE_key1_key2_etc()". The name of each subroutine is the underscore-separated concatenation of the non-flag keys that locate the same pattern in %RE. Flags are passed to the subroutine in its argument list. Thus:

        use Regexp::Common qw( RE_ws_crop RE_num_real RE_profanity );
        $str =~ RE_ws_crop() and die "Surrounded by whitespace";
        $str =~ RE_num_real(-base=>8, -sep=>" ") or next;
        $offensive = RE_profanity(-keep);
        $str =~ s/$offensive/$bad{$1}++; "<expletive deleted>"/ge;

Note that, unlike the hash-based interface (which returns objects), these subroutines return ordinary "qr"'d regular expressions. Hence they do not curry, nor do they provide the OO match and substitution inlining described in the previous section.

It is also possible to export subroutines for all available patterns like so:

        use Regexp::Common 'RE_ALL';

Or you can export all subroutines with a common prefix of keys like so:

        use Regexp::Common 'RE_num_ALL';

which will export "RE_num_int" and "RE_num_real" (and if you have create more patterns who have first key num, those will be exported as well). In general, RE_key1_..._keyn_ALL will export all subroutines whose pattern names have first keys key1 ... keyn.

Adding new regular expressions

You can add your own regular expressions to the %RE hash at run-time, using the exportable "pattern" subroutine. It expects a hash-like list of key/value pairs that specify the behaviour of the pattern. The various possible argument pairs are:

"name => [ @list ]"

A required argument that specifies the name of the pattern, and any flags it may take, via a reference to a list of strings. For example:

         pattern name => [qw( line of -char )],
                 # other args here
                 ;
    

This specifies an entry $RE{line}{of}, which may take a "-char" flag.

Flags may also be specified with a default value, which is then used whenever the flag is specified without an explicit value (but not when the flag is omitted). For example:

         pattern name => [qw( line of -char=_ )],
                 # default char is '_'
                 # other args here
                 ;
    

"create => $sub_ref_or_string"

A required argument that specifies either a string that is to be returned as the pattern:

        pattern name    => [qw( line of underscores )],
                create  => q/(?:^_+$)/
                ;
    

or a reference to a subroutine that will be called to create the pattern:

        pattern name    => [qw( line of -char=_ )],
                create  => sub {
                                my ($self, $flags) = @_;
                                my $char = quotemeta $flags->{-char};
                                return '(?:^$char+$)';
                            },
                ;
    

If the subroutine version is used, the subroutine will be called with three arguments: a reference to the pattern object itself, a reference to a hash containing the flags and their values, and a reference to an array containing the non-flag keys.

Whatever the subroutine returns is stringified as the pattern.

No matter how the pattern is created, it is immediately postprocessed to include or exclude capturing parentheses (according to the value of the "-keep" flag). To specify such "optional" capturing parentheses within the regular expression associated with "create", use the notation "(?k:...)". Any parentheses of this type will be converted to "(...)" when the "-keep" flag is specified, or "(?:...)" when it is not. It is a Regexp::Common convention that the outermost capturing parentheses always capture the entire pattern, but this is not enforced.

"match => $sub_ref"

An optional argument that specifies a subroutine that is to be called when the "$RE{...}->matches(...)" method of this pattern is invoked.

The subroutine should expect two arguments: a reference to the pattern object itself, and the string to be matched against.

It should return the same types of values as a "m/.../" does.

     pattern name    => [qw( line of -char )],
             create  => sub {...},
             match   => sub {
                             my ($self, $str) = @_;
                             $str !~ /[^$self->{flags}{-char}]/;
                        },
             ;
    

"subs => $sub_ref"

An optional argument that specifies a subroutine that is to be called when the "$RE{...}->subs(...)" method of this pattern is invoked.

The subroutine should expect three arguments: a reference to the pattern object itself, the string to be changed, and the value to be substituted into it. The third argument may be "undef", indicating the default substitution is required.

The subroutine should return the same types of values as an "s/.../.../" does.

For example:

     pattern name    => [ 'lineof', '-char=_' ],
             create  => sub {...},
             subs    => sub {
                          my ($self, $str, $ignore_replacement) = @_;
                          $_[1] =~ s/^$self->{flags}{-char}+$//g;
                        },
             ;
    

Note that such a subroutine will almost always need to modify $_[1] directly.

"version => $minimum_perl_version"

If this argument is given, it specifies the minimum version of perl required to use the new pattern. Attempts to use the pattern with earlier versions of perl will generate a fatal diagnostic.

Loading specific sets of patterns.

By default, all the sets of patterns listed below are made available. However, it is possible to indicate which sets of patterns should be made available - the wanted sets should be given as arguments to "use". Alternatively, it is also possible to indicate which sets of patterns should not be made available - those sets will be given as argument to the "use" statement, but are preceded with an exclaimation mark. The argument no_defaults indicates none of the default patterns should be made available. This is useful for instance if all you want is the "pattern()" subroutine.

Examples:

 use Regexp::Common qw /comment number/;  # Comment and number patterns.
 use Regexp::Common qw /no_defaults/;     # Don't load any patterns.
 use Regexp::Common qw /!delimited/;      # All, but delimited patterns.

It's also possible to load your own set of patterns. If you have a module "Regexp::Common::my_patterns" that makes patterns available, you can have it made available with

 use Regexp::Common qw /my_patterns/;

Note that the default patterns will still be made available - only if you use no_defaults, or mention one of the default sets explicitly, the non mentioned defaults aren't made available.

List of available patterns

The patterns listed below are currently available. Each set of patterns has its own manual page describing the details. For each pattern set named name, the manual page Regexp::Common::name describes the details.

Currently available are:

Regexp::Common::balanced

Provides regexes for strings with balanced parenthesized delimiters.

Regexp::Common::comment

Provides regexes for comments of various languages (43 languages currently).

Regexp::Common::delimited

Provides regexes for delimited strings.

Regexp::Common::lingua

Provides regexes for palindromes.

Regexp::Common::list

Provides regexes for lists.

Regexp::Common::net

Provides regexes for IPv4, IPv6, and MAC addresses.

Regexp::Common::number

Provides regexes for numbers (integers and reals).

Regexp::Common::profanity

Provides regexes for profanity.

Regexp::Common::whitespace

Provides regexes for leading and trailing whitespace.

Regexp::Common::zip

Provides regexes for zip codes.

Forthcoming patterns and features

Future releases of the module will also provide patterns for the following:

        * email addresses 
        * HTML/XML tags
        * more numerical matchers,
        * mail headers (including multiline ones),
        * more URLS
        * telephone numbers of various countries
        * currency (universal 3 letter format, Latin-1, currency names)
        * dates
        * binary formats (e.g. UUencoded, MIMEd)

If you have other patterns or pattern generators that you think would be generally useful, please send them to the maintainer -- preferably as source code using the "pattern" subroutine. Submissions that include a set of tests will be especially welcome.