Changeset 17066 for gsdl/trunk/perllib/plugins/OAIPlugin.pm

Timestamp:

2008-08-28T17:20:44+12:00 (16 years ago)

Author:

ak19

Message:

OAIPlugin now works again: 1. needs to inherit from ReadTextFile as well in order to have access to the method textcat_get_language_encoding; 2. if dc is not specified as a prefix in each metadata element tag, looks inside the containing/wrapper element to see if that is set to dc and if so, uses that as the prefix.

File:

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

gsdl/trunk/perllib/plugins/OAIPlugin.pm

@@ -34 +34 @@
 
 use ReadXMLFile;
+use ReadTextFile; # needed for subroutine textcat_get_language_encoding
 
 sub BEGIN {
-    @OAIPlugin::ISA = ('ReadXMLFile');
+    @OAIPlugin::ISA = ('ReadXMLFile', 'ReadTextFile');
 }
 
@@ -46 +47 @@
     'reqd' => "no",
     'deft' => &get_default_process_exp() },
+      { 'name' => "input_encoding", # needed by subroutine textcat_get_language_encoding
+    'desc' => "{ReadTextFile.input_encoding}",
+    'type' => "enum",
+    'list' => $ReadTextFile::encoding_plus_auto_list,
+    'reqd' => "no" ,
+    'deft' => "auto" }
       ];
 
@@ -410 +417 @@
     my($namespace,$top_level_prefix) = ($wrapper_metadata_xml =~ m/^(.*?):(.*?)$/);
 
-    if ($top_level_prefix !~ /dc$/) {
+    # sometimes, the dc namespace is not specified as the prefix in each element (like <dc:title>)
+    # but is rather defined in the wrapper element containing the various dc meta elements,
+    # like <dc><title></title><creator></creator></dc>.
+    # In such a case, we use this wrapper element as the top_level_prefix
+    if(!defined $top_level_prefix && defined $wrapper_metadata_xml && $wrapper_metadata_xml =~ m/dc$/) {
+        $top_level_prefix = $wrapper_metadata_xml;
+    }
+
+    if ($top_level_prefix !~ m/dc$/) {
         print $outhandle "Warning: OAIPlugin currently only designed for Dublin Core (or variant) metadata\n";
         print $outhandle "         This recorded metadata section '$top_level_prefix' does not appear to match.\n";
@@ -434 +449 @@
         {
         $metaname = "$top_level_prefix.$metaname";
+#       print STDERR "*** metaname = $metaname\tmetavalue = $metavalue\n";
         }
         $metaname =~ s/\.(.)/\.\u$1/;