Changeset 3369

Timestamp:

2002-08-20T17:09:03+12:00 (22 years ago)

Author:

sjboddie

Message:

HTMLPlug will no longer prevent metadata extraction when the
-description_tags option is set unless it actually finds at least one
<Section> tag in the document. This allows documents with <section> tags
and those without to be built in the same collection without getting an
unexpected lack of metadata for the latter.

File:

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

trunk/gsdl/perllib/plugins/HTMLPlug.pm

@@ -81 +81 @@
                           occur. Note that by setting this option you
                           implicitly set -no_metadata, as all metadata should
-                          be included within the <Section> tags. Also,
-                          '-keep_head' will have no effect when this option
-                          is set.\n";
+                          be included within the <Section> tags (this is only
+                          true for documents that actually contain <Section> tags
+                          however). Also, '-keep_head' will have no effect when 
+                          this option is set, regardless of whether a document
+                          contains Section tags.\n";
 }
 
@@ -186 +188 @@
         my $comment = $2;
         if (defined $text) {
+        # text before a comment - note that getting to here
+        # doesn't necessarily mean there are Section tags in
+        # the document
         $self->process_section(\$text, $base_dir, $file, $doc_obj, $cursection);
         }
@@ -231 +236 @@
         print $outhandle "HTMLPlug: WARNING: $file appears to contain no Section tags so\n";
         print $outhandle "          will be processed as a single section document\n";
+        
+        # go ahead and process single-section document
         $self->process_section($textref, $base_dir, $file, $doc_obj, $cursection);
+
+        # if document contains no Section tags we'll go ahead
+        # and extract metadata (this won't have been done
+        # above as the -description_tags option prevents it)
+        $self->extract_metadata (\$doc_obj->get_text($cursection), $metadata, $doc_obj, $cursection) 
+            unless $self->{'no_metadata'};
+
         } else {
         print $outhandle "HTMLPlug: WARNING: $file contains the following text outside\n";
@@ -245 +259 @@
         print $outhandle " ($text)\n";
         }
+    } elsif (!$found_something) {
+
+        # may get to here if document contained no valid Section
+        # tags but did contain some comments. The text will have
+        # been processed already but we should print the warning
+        # as above and extract metadata
+        print $outhandle "HTMLPlug: WARNING: $file appears to contain no Section tags so\n";
+        print $outhandle "          will be processed as a single section document\n";
+
+        $self->extract_metadata (\$doc_obj->get_text($cursection), $metadata, $doc_obj, $cursection) 
+        unless $self->{'no_metadata'};
     }