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

package Text::Dokuwiki;

use strict;
use warnings;
use feature qw/ switch /;
use utf8;

use Data::Dumper;

sub new {
  my ($class) = @_;
  my $self = {};

  return bless($self, $class);
}

sub _parse_list {
  my ($self, $lines) = @_;
  my @lists = ();
  my @stack = ( \@lists );
  my $types = {curr => '', last => ''};
  my $level = {curr => 0,  last => 0};

  foreach my $line (@{ $lines }) {
    $line =~ m/^(\s+)/o;
    my ($ident, $dot, $rest) = ($line =~ m/^((?:\s{2})+)([\*-])\s*(.+)/);
    $level->{last} = $level->{curr};
    $level->{curr} = $ident =~ tr/ / /;

    $types->{last} = $types->{curr};
    $types->{curr} = ($dot eq '-') ? 'ol' : 'ul';

    if ($level->{curr} == $level->{last} and
        $types->{curr} ne $types->{last}) {
      pop @stack;
      my $list = [$types->{curr} => {}];
      push @{ $stack[-1] }, (@stack > 1) ? [li => {}, $list] : $list;
      push @stack, $list;
    }
    if ($level->{curr} >  $level->{last}) {
      my $list = [$types->{curr} => {}];
      push @{ $stack[-1] }, (@stack > 1) ? [li => {}, $list] : $list;
      push @stack, $list;
    }
    if ($level->{curr} <  $level->{last}) {
      pop @stack;
    }
    push @{ $stack[-1] }, [li => {}, $rest];
  }
  pop @stack while @stack;

  return @lists;
}

sub _parse_table {
  my ($self, $lines) = @_;
  my ($i, $j, $colspan) = (0, 0, 0);
  my @rows = ();

  foreach my $line (@{ $lines }) {
    my @row = ();
    for ($j = 0; ;$j++) {
      last if ($line eq '|' or $line eq '^');
      next if ($line !~ m/^(([\|\^])([^\|\^]*))/o);
      my $attrs = {};
      my ($all, $key, $value) = ($1, $2, $3);
      $line = substr($line, length($all));
      my $type = ($key eq '^') ? 'th' : 'td';
      my ($lpad, $content, $rpad) = ($value =~ m/^(\s*)(.*?)(\s*)$/o);
      if ($line =~ m/^([\|\^]{2,})/o) {
        # colspan detected;
        $attrs->{colspan} = $colspan = length($1);
      } elsif (index($content, ":::") >= 0) {
        # rowspan detected
        for (my $k = $i - 1; $k >= 0; $k--) {         # k is idx of prev rows
          my $cellptr = $rows[$k][$j + 2];            # +2 for (tr => {})
          next if (index($cellptr->[2], ":::") >= 0); # also rowspan
          $cellptr->[1]->{rowspan} //= 1;             # init attr, if missing
          $cellptr->[1]->{rowspan}  += 1;             # incr value
          last;
        }
        $attrs->{skip} = 1;
      } elsif ($colspan > 1) {
        # colspan-eaten column
        $attrs->{skip} = 1;
        $colspan -= 1;
      }
      if ($lpad eq '') {
        $attrs->{align} = ($rpad eq '') ? 'center' : 'left';
      } else {
        $attrs->{align} = ($rpad eq '') ? 'right' : 'center';
      }
      push @row, [$type => $attrs, $content];
    }
    push @rows, [tr => {}, @row];
    $i++;
  }

  return [table => {}, @rows];
}

sub parse {
  my ($self, $text) = @_;
  my ($tree, $mode, $attrs, $buf) = ([], 'text', '', '');

  my @lines = split /\r?\n/o, $text;
  my $linenum = 0;

  foreach my $line (@lines) {
    $linenum++;

    given ($mode) {
      when (m!block/(file|code|nowiki)!o) {
        if ($line =~ m{^\s*</$1>}o) {
          if ($1 eq 'file') {
            my $dt = [dt => {}, 0 => $attrs->{file}];
            my $dd = [dt => {}, 0 => [pre => {class => $attrs->{class}}, 0 => $buf]];
            push @{ $tree }, [dl => {class => 'file'}, [$dt, $dd]];
          } elsif ($1 eq 'nowiki') {
            push @{ $tree }, [pre  => {}, 0 => $buf];
          } else {
            push @{ $tree }, [code => {class => $attrs->{class}}, 0 => $buf];
          }
          ($buf, $mode, $attrs) = ('', '', {}); next;
        }
        $buf .= $line . "\n";
      }
      when ("code") {
        if ($line =~ m/^\s{2}(.+)/o) {
          $buf .= $line . "\n"; next;
        } else {
          push @{ $tree }, [pre => {}, 0 => $buf];
          ($buf, $mode, $attrs) = ('', '', {}); continue;
        }
      }
      when ("list") {
        if ($line =~ m/^\s{2}([\*-])\s+(.+)/o) {
          push @{ $buf }, [li => {}, 0 => $2];
          next;
        } else {
          push @{ $tree }, [ul => {}, @$buf]; # TODO: lost list type
          ($buf, $mode, $attrs) = ('', '', {}); continue;
        }
      }
      when ("table") {
        ...
      }
    }


    given ($line) {
      # header
      when (m/^\s?(={2,6}) (.+) \g{1}\s*/o) {
        my $level = $1 =~ tr/=/=/;
        $level = 7 - $level; # invert
        push @{ $tree }, ["h$level" => {}, 0 => $2];
        next;
      }
      # code/file block
      when (m/^\s?<(code|file)(?:\s+(\S+)\s+(\S+))?>\s*$/o) {
        $mode = "block/$1";
        $attrs = ($2) ? {class => $2, file => $3} : {};
        next;
      }
      # nowiki block
      when (m/\s?<nowiki>/o) {
        $mode = "block/nowiki";
        next;
      }
      # lists
      when (m/^\s{2}([\*-])\s+(.+)/o) {
        $mode = 'list';
        $buf  = [];
        push @{ $buf }, [li => {}, 0 => $2];
        next;
      }
      # quotes
      when (m/^\s?(>)+\s*(.+)/o) {
        my $level = $1 =~ tr/>/>/;
        push @{ $tree }, [blockquote => {level => $level}, 0 => $2];
        next;
      }
      # table
      when (m/^\s?[\|\^]/o) {
        $mode = 'table';
        $buf = $line . "\n"; # render later
        next;
      }
      # code idented with two spaces
      when (m/^\s{2}(\S.+)/o) {
        $mode = 'code';
        $buf = $line . "\n";
        next;
      }
      # nonempty line
      when (m/^\s?(\S.+)/o) {
        push @{ $tree }, [p => {}, 0 => $1];
        next;
      }
      # empty lines;
      when (m/^\s*$/) {
        push @{ $tree }, [br => {}];
        $mode = '';
        next;
      }
      # catchall
      default {
        printf "Unmatched % 3d: %s\n", $linenum, $line;
        continue;
      }
    }
  }

  return $tree;
}

1;

__END__

[div => {class => 'block'},           # <div class='block'>
  0 => 'text1',                       #   Hello!
  [strong => {}, 0 => 'bold text'],   #   <strong>user</strong>
  0 => ', this is converted text.',   #   , this is converted text.
  [br => {}],                         #   <br/>
  [p => {}, 0 => 'Second paragraph']  #   <p>Second paragraph</p>
]                                     # </div>