libtext-dokuwiki-perl/lib/Text/Dokuwiki/Regexps.pm

package Text::Dokuwiki::Regexps;

use strict;
use warnings;
use utf8;

my $SP  = '[\ \t]';
my $EOL = '\r?\n';
our $regexps = {};

$regexps->{header} = qr/
(?:
  ^                 # line start
  $SP?              # maybe one leading space
  (?<line> ={2,6} ) # header marks
  \s*               # leading spaces
  (?<header> .+ )   # header text (at least one non-empty character, but least small piece)
  $SP*              # trailing spaces
  (?: \g{line} )    # header mark (exactly same as above)
  (?: $EOL | $)     # newline or file end
)
/mx;

$regexps->{codeblock} = qr%
^     # first line
$SP?  # maybe one leading space
<     # opening tag start
(?<tag> code | file ) # tag name
(?:    # optional block
  (?: $SP+ (?<syntax>\S+) )?
  (?: $SP+ (?<filename>\S+) )
)?
>     # opening tag end
$SP*  # maybe any number of trailing spaces
$EOL  # end of first line

(?<block>  # body of block
  (?:
    ^      # line start
    .*     # contents
    $EOL   # line end (note: no $)
  )*?      # don't be greedy, other blocks may follow
)          # block body end

^             # last line
$SP?          # maybe one leading space
</\g{tag}>    # the same tag as in first line
$SP*          # maybe any number of trailing spaces
(?: $EOL | $) # end of last line
%mx;

$regexps->{table} = qr/
(?<table>
  (?:
    ^              # line start
    $SP?           # maybe one leading space
    [|^]           # at next char is '|' (td) or '^' (th)
    .*             # rest of line
    [|^]           # ends with '|' or '^'
    $SP*           # maybe one or more trailing space(s)
    (?: $EOL | $ ) # newline or file end
  )+               # one or more such lines
)
/mx;

$regexps->{pre} = qr/
(?<block>
  (?:
    ^              # line start
    $SP {2}        # at least two spaces
    .*             # rest of line
    (?: $EOL | $ ) # newline or file end
  )+               # one or more lines
)
/mx;

$regexps->{blockquote} = qr/
(?<block>
  (?:
    ^              # line start
    $SP?           # maybe one leading space
    [>]{1,}        # quote marker(s)
    .*             # rest of line
    (?: $EOL | $ ) # newline or file end
  )+               # one or more lines
)
/mx;

$regexps->{list} = qr/
(?<list>
  (?:
    ^              # line start
    $SP {2,}       # two or more spaces
    [*-]           # start marker of the list item
    .+             # rest of line
    (?: $EOL | $ ) # newline or file end
  )+               # one or more lines
)
/mx;

$regexps->{paragraph} = qr/
(?<text>
  (?:
    ^              # line start
    $SP?           # optional leading space
    \S.*           # at least one printable symbol
    (?: $EOL | $ ) # newline or file end
  )+               # one or more lines
)
/mx;

$regexps->{emptyline} = qr{
(?:
  ^                # line start
  (?: $SP* )       # zero or more spaces
  (?: $EOL | $ )   # newline or file end
)*
}mx;

$regexps->{hrule} = qr/
(?:
  ^                # line start
  (?: -{4,} )      # four or more dashes
  (?: $EOL | $ )   # newline or file end
)
/mx;

1;