Changeset 2819

Timestamp:

2001-11-05T22:49:46+13:00 (22 years ago)

Author:

sjboddie

Message:

Altered HTMLPlug's description_tags option a bit so it should now also
work for plugins derived from HTMLPlug (i.e. RTFPlug, WordPlug, PDFPlug,
PSPlug etc). Tested briefly with RTFPlug and WordPlug and it seems ok.

File:

• trunk/gsdl/perllib/plugins/HTMLPlug.pm (modified) (4 diffs)

trunk/gsdl/perllib/plugins/HTMLPlug.pm

@@ -156 +156 @@
     if ($self->{'description_tags'}) {
 
+    my $opencom = '(?:<!--|&lt;!(?:&mdash;|&#151;|--))';
+    my $closecom = '(?:-->|(?:&mdash;|&#151;|--)&gt;)';
+    my $lt = '(?:<|&lt;)';
+    my $gt = '(?:>|&gt;)';
+    my $quot = '(?:"|&quot;|&rdquo;|&ldquo;)';
+
     my $found_something = 0; my $top = 1;
-    while ($$textref =~ s/^(.*?)<!--(.*?)-->//s) {
+    while ($$textref =~ s/^(.*?)$opencom(.*?)$closecom//s) {
         my $text = $1;
         my $comment = $2;
@@ -163 +169 @@
         $self->process_section(\$text, $base_dir, $file, $doc_obj, $cursection);
         }
-        while ($comment =~ s/<([^>]+)>//s) {
+        while ($comment =~ s/$lt(.*?)$gt//s) {
+        
         my $tag = $1;
         if ($tag eq "Section") {
@@ -172 +179 @@
             $found_something = 1;
             $cursection = $doc_obj->get_parent_section ($cursection);
-        } elsif ($tag =~ /^Metadata name=\"([^\"]+)\"/s) {
+        } elsif ($tag =~ /^Metadata name=$quot(.*?)$quot/s) {
             my $metaname = $1;
-            $comment =~ s/^(.*?)<\/Metadata>//s;
+            $comment =~ s/^(.*?)$lt\/Metadata$gt//s;
             my $metavalue = $1;
             $metavalue =~ s/^\s+//;
             $metavalue =~ s/\s+$//;
+                    # assume that no metadata value intentionally includes
+                    # carriage returns or HTML tags (if they're there they
+                    # were probably introduced when converting to HTML from
+                    # some other format).
+            $metavalue =~ s/[\cJ\cM]/ /sg;
+            $metavalue =~ s/<[^>]+>//sg;
+            $metavalue =~ s/\s+/ /sg;
             $doc_obj->set_utf8_metadata_element($cursection, $metaname, $metavalue);
+        } elsif ($tag eq "Description" || $tag eq "/Description") {
+            # do nothing with containing Description tags
+        } else {
+            # simple HTML tag (probably created by the conversion
+            # to HTML from some other format) - we'll ignore it and
+            # hope for the best ;-)
         }
         }
@@ -197 +217 @@
         print $outhandle "          of the final closing </Section> tag. This text will\n";
         print $outhandle "          be ignored.";
+        my ($text);
         if (length($$textref) > 30) {
             $text = substr($$textref, 0, 30) . "...";
+        } else {
+            $text = $$textref;
         }
         $text =~ s/\n/ /isg;