Changeset 8795

Timestamp:

2004-12-14T12:31:58+13:00 (19 years ago)

Author:

kjdon

Message:

if use_sections is on, now we are a bit more relaxed about what the sections are split on. used to be <a name=, now its <[Aa] name=, and instead of matching \d+ for the title, we use \"?\w+\"?, cos sometimes you get <a name="outline">
And since we can now have non-digit titles in the document, we explicitly set gsdlthistype to be paged, otherwise greenstone will treat it as hierarchical if there are non-digit titles.

File:

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

trunk/gsdl/perllib/plugins/PDFPlug.pm

@@ -150 +150 @@
 
     # we have "<a name=1></a>" etc for each page
-    my @sections = split('<a name=', $$textref);
+    # it may be <A name=
+    my @sections = split('<[Aa] name=', $$textref);
 
     if (scalar (@sections) == 1) { #only one section - no split!
@@ -161 +162 @@
     
     my $title = $sections[0];
-    $title =~ s/^\d+>//; # specific for pdftohtml...
+    $title =~ s/^\"?\w+\"?>//; # specific for pdftohtml...
     $title =~ s/<\/([^>]+)><\1>//g; # (eg) </b><b> - no space
     $title =~ s/<[^>]*>/ /g;
@@ -183 +184 @@
     # add metadata per section...
     foreach my $section (@sections) {
-        $section =~ s@^(\d+)></a>@@; # leftover from split expression...
+        # section names are not always just digits, may be like "outline"
+        $section =~ s@^\"?(\w+)\"?></a>@@; # leftover from split expression...
 
         $title = $1; # Greenstone does magic if sections are titled digits
         if (! defined($title) ) {
         print STDERR "no title: $section\n";
+        $title = " "; # get rid of the undefined warning in next line
         }
         my $newsection = "<!-- from PDFPlug -->\n<!-- <Section>\n";
@@ -237 +240 @@
 
     # Add NumPages metadata (we have "<a name=1>" etc for each page)
-    my @pages = ($$textref =~ /\<a name=\d+\>/ig);
+    my @pages = ($$textref =~ /\<[Aa] name=\"?\w+\"?>/ig);
     $doc_obj->add_utf8_metadata($cursection, "NumPages", scalar(@pages));
-
+    
+    if ($self->{'use_sections'} && $self->{'converted_to'} eq "HTML") {
+    # we explicitly make it a paged document, cos greenstone won't get it
+    # right if any section has an empty title, or one with letters in it
+    $doc_obj->set_utf8_metadata_element ($cursection, "gsdlthistype", "Paged");
+    }
     return $result;
 }