Changeset 10724

Timestamp:

2005-10-14T13:37:43+13:00 (19 years ago)

Author:

chi

Message:

Add an option-metadata_fields to allow user-specified metadata fields to be extracted from PDF document.

File:

• trunk/gsdl/perllib/plugins/PDFPlug.pm (modified) (3 diffs)

trunk/gsdl/perllib/plugins/PDFPlug.pm

@@ -67 +67 @@
     'type' => "regexp",
     'deft' => &get_default_block_exp() },
-      { 'name' => "noimages",
-    'desc' => "{PDFPlug.noimages}",
+       { 'name' => "metadata_fields",
+     'type' => "string",
+     'deft' => "" },
+       { 'name' => "noimages",
+     'desc' => "{PDFPlug.noimages}",
     'type' => "flag" },
       { 'name' => "allowimagesonly",
@@ -156 +159 @@
     # Instruct HTMLPlug (when eventually accessed through read_into_doc_obj) 
     # to extract these metadata fields from the HEAD META fields
-    push(@$html_options,"-metadata_fields","Title,GENERATOR,date,author<Creator>");
-    
+    my $required_metadata;
+    if (defined $self->{'metadata_fields'} && $self->{'metadata_fields'} =~ /\S/) {
+    push(@$html_options,"-metadata_fields",$self->{'metadata_fields'});
+    } else {
+    push(@$html_options,"-metadata_fields","Title,GENERATOR,date,author<Creator>");
+    }
+    #push(@$html_options,"-metadata_fields","Title,GENERATOR,date,author<Creator>");
+        
     if ($self->{'use_sections'} || $self->{'description_tags'}) {
     $self->{'description_tags'} = 1;
@@ -287 +296 @@
     # and fix the capitalisation of 'date'
     my $cursection = $doc_obj->get_top_section();
-
     foreach my $datemeta (@{$doc_obj->get_metadata($cursection, "date")}) {
     $doc_obj->delete_metadata($cursection, "date", $datemeta);