* Text::Dokuwiki::Parser : use regex parser instead per-line

lib/Text/Dokuwiki/Parser.pm

@@ -4,6 +4,8 @@ use strict;
 use warnings;
 use utf8;
 
+use Text::Dokuwiki::Regexps;
+
 sub new {
   my ($class) = @_;
   my $self = {
@@ -224,108 +226,54 @@ sub _parse_text {
 
 sub parse {
   my ($self, $text) = @_;
-  my ($mode, $attrs, $buf, @tree) = ('text', '', '');
-
-  my @lines = split /\r?\n/o, $text;
-  my $linenum = 0;
-
-  foreach my $line (@lines) {
-    $linenum++;
-
-    if ($mode =~ m!block/(file|code|nowiki)!o) {
-      my $tag = $1;
-      $buf ||= [];
-      if ($line =~ m{^\s*</$tag>}o) {
-        $buf = join("\n", @{ $buf });
-        if ($tag eq 'file') {
-          my $dt = [dt => {}, $attrs->{file}];
-          my $dd = [dt => {}, [pre => {class => $attrs->{class}}, $buf]];
+
+  my $rx = $Text::Dokuwiki::Regexps::regexps;
+  my (@tree);
+  my $len = length($text);
+  pos($text) = 0;
+
+  while ($len - pos($text) > 0) {
+    my $pos = pos($text);
+    if (0) { # TODO: MACRO
+    } elsif ($text =~ m/\G$rx->{header}/cgi) {
+      my $level = length($+{line});
+      $level = 7 - $level; # invert
+      push @tree, ["h$level" => {}, $+{header}];
+    } elsif ($text =~ m/\G$rx->{codeblock}/cgi) {
+      if ($+{tag} eq 'file' and $+{filename}) {
+        my $attrs = $+{syntax} ? {class => $+{syntax}} : {};
+        my $dt = [dt => {}, $+{filename}];
+        my $dd = [dt => {}, [pre => $attrs, $+{block}]];
         push @tree, [dl => {class => 'file'}, [$dt, $dd]];
-        } elsif ($tag eq 'nowiki') {
-          push @tree, [pre  => {}, $buf];
+      } elsif ($+{tag} eq 'file') {
+        push @tree, [pre => {class => 'file'}, $+{block}];
       } else {
-          push @tree, [code => {class => $attrs->{class}}, $buf];
-        }
-        ($buf, $mode, $attrs) = ('', '', {}); next;
-      }
-      push @{ $buf }, $line;
-      next;
-    } elsif ($mode eq 'code') {
-      if ($line =~ m/^\s{2}(.+)/o) {
-        $buf .= $line . "\n";
+        my $attrs = $+{syntax} ? {class => $+{syntax}} : {};
+        push @tree, [code => $attrs, [pre => {}, $+{block}]];
+      }
+    } elsif ($text =~ m/\G$rx->{table}/cgi) {
+      push @tree, $self->_parse_table($+{table});
+    } elsif ($text =~ m/\G$rx->{list}/cgi) {
+      my $list = $+{list};
+      chomp $list;
+      push @tree, $self->_parse_list([ split(/\r?\n/, $list) ]);
+    } elsif ($text =~ m/\G$rx->{blockquote}/cgi) {
+      push @tree, [blockquote => {}, $+{block}];
+    } elsif ($text =~ m/\G$rx->{pre}/cgi) {
+      push @tree, [pre => {}, $+{block}];
+    } elsif ($text =~ m/\G$rx->{paragraph}/cgi) {
+      my $text = $+{text};
+      chomp $text;
+      push @tree, [p => {}, $self->_parse_text($text)];
+    } elsif ($text =~ m/\G$rx->{emptyline}/cgi) {
       next;
+    } else {
+      my $msg = "Unmatched: '" . substr($text, pos($text)) . "'\n";
+      die($msg);
     }
-      push @tree, [pre => {}, $buf];
-      ($buf, $mode, $attrs) = ('', '', {});
-    } elsif ($mode eq 'list') {
-      if ($line =~ m/^(\s{2})+([\*-])\s+(.+)/o) {
-        push @{ $buf }, $line;
-        next;
-      }
-      push @tree, $self->_parse_list($buf);
-      ($buf, $mode, $attrs) = ('', '', {});
-    } elsif ($mode eq 'table') {
-      if ($line =~ m/^\s?[\|\^]/o) {
-        push @{ $buf }, $line;
-        next;
-      }
-      push @tree, $self->_parse_table($buf);
-      ($buf, $mode, $attrs) = ('', '', {});
-    }
-
-    # macro
-    if ($line =~ m/~~NO(TOC|CACHE)~~/) {
-      warn "ignored macro `$line` at line $linenum\n";
-      next;
-    } elsif ($line =~ m/^\s?(={2,6}) (.+) \g{1}\s*/o) {
-      # header
-      my $level = $1 =~ tr/=/=/;
-      $level = 7 - $level; # invert
-      push @tree, ["h$level" => {}, $2];
-      next;
-    } elsif ($line =~ m/^\s?<(code|file)(?:\s+(\S+)\s+(\S+))?>\s*$/o) {
-      # code/file block
-      $mode = "block/$1";
-      $attrs = ($2) ? {class => $2, file => $3} : {};
-      next;
-    } elsif ($line =~ m/\s?<nowiki>/o) {
-      # nowiki block
-      $mode = "block/nowiki";
-      next;
-    } elsif ($line =~ m/^(\s{2})+([\*-])\s+(.+)/o) {
-      # lists
-      $mode = 'list';
-      $buf  = [];
-      push @{ $buf }, $line;
-      next;
-    } elsif ($line =~ m/^\s?(>)+\s*(.+)/o) {
-      # quotes
-      my $level = $1 =~ tr/>/>/;
-      push @tree, [blockquote => {level => $level}, $2];
-      next;
-    } elsif ($line =~ m/^\s?[\|\^]/o) {
-      # table
-      $mode = 'table';
-      $buf = [];
-      push @{ $buf }, $line;
-      next;
-    } elsif ($line =~ m/^\s{2}(\S.+)/o) {
-      # code idented with two spaces
-      $mode = 'code';
-      $buf = $line . "\n";
-      next;
-    } elsif ($line =~ m/^\s?(\S.+)/o) {
-      # nonempty line
-      push @tree, [p => {}, $self->_parse_text($1)];
-      next;
-    } elsif ($line =~ m/^\s*$/o) {
-      # empty lines
-      push @tree, [br => {}];
-      $mode = '';
-      next;
+    if (pos($text) <= $pos) {
+      die("parser failed, abort\n");
     }
-    # catchall
-    printf "Unmatched % 3d: %s\n", $linenum, $line;
   }
 
   if (scalar @{ $self->{footnotes} }) {