Changeset 14940 for gsdl/trunk/perllib/plugins/OAIPlug.pm

Timestamp:

2008-01-11T14:09:55+13:00 (15 years ago)

Author:

dmn

Message:

updates for qualified DC

File:

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

gsdl/trunk/perllib/plugins/OAIPlug.pm

@@ -326 +326 @@
 
 
+sub remap_dcterms_metadata
+{
+    my $self = shift(@_);
+
+    my ($metaname) = @_;
+
+    my $dcterm_mapping = {
+    "alternative" => "dc.title",
+    "tableOfContents" => "dc.description",
+    "abstract" => "dc.description",
+    "created" => "dc.date",
+    "valid" => "dc.date",
+    "available" => "dc.date",
+    "issued" => "dc.date",
+    "modified" => "dc.date",
+    "dateAccepted" => "dc.date",
+    "dateCopyrighted" => "dc.date",
+    "dateSubmitted" => "dc.date",
+    "extent" => "dc.format",
+    "medium" => "dc.format",
+    "isVersionOf" => "dc.relation",
+    "hasVersion" => "dc.relation",
+    "isReplacedBy" => "dc.relation",
+    "replaces" => "dc.relation",
+    "isRequiredBy" => "dc.relation",
+    "requires" => "dc.relation",
+    "isPartOf" => "dc.relation",
+    "hasPart" => "dc.relation",
+    "isReferencedBy" => "dc.relation",
+    "references" => "dc.relation",
+    "isFormatOf" => "dc.relation",
+    "hasFormat" => "dc.relation",
+    "conformsTo" => "dc.relation",
+    "spatial" => "dc.coverage",
+    "temporal" => "dc.coverage",
+    "audience" => "dc.any",
+    "accrualMethod" => "dc.any",
+    "accrualPeriodicity" => "dc.any",
+    "accrualPolicy" => "dc.any",
+    "instructionalMethod" => "dc.any",
+    "provenance" => "dc.any",
+    "rightsHolder" => "dc.any",
+    "mediator" => "audience",
+    "educationLevel" => "audience",
+    "accessRights" => "dc.rights",
+    "license" => "dc.rights",
+    "bibliographicCitation" => "dc.identifier"
+    };
+
+    my ($prefix,$name) = ($metaname =~ m/^(.*?)\.(.*?)$/);
+    
+    if ($prefix eq "dcterms")
+    {
+    if (defined $dcterm_mapping->{$name}) 
+    {
+        return $dcterm_mapping->{$name}."^".$name;
+    }
+
+    }
+    return $metaname; # didn't get a match, return param passed in unchanged
+}
 
 
@@ -340 +401 @@
     {
     my $metadata_text = $1;
-    $metadata_text =~ s/^.*?<(oai_dc:)?dc.*?>(.*?)<\/(oai_dc:)?dc>.*?/$2/s;
-
-    while ($metadata_text =~ m/<(.*?)>(.*?)<\/(.*?)>(.*)/s)
+
+    # locate and remove outermost tag (ignoring any attribute information in top-level tag)
+    my ($wrapper_metadata_xml,$inner_metadata_text) = ($metadata_text =~ m/<([^ >]+).*?>(.*?)<\/\1>/s);
+
+    # split tag into namespace and tag name
+    my($namespace,$top_level_prefix) = ($wrapper_metadata_xml =~ m/^(.*?):(.*?)$/);
+
+    if ($top_level_prefix !~ /dc$/) {
+        print $outhandle "Warning: OAIPlug 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";
+        print $outhandle "         Metadata assumed to be in form: <prefix:tag>value</prefix:tag> and will be converted\n";
+        print $outhandle "         into Greenstone metadata as prefix.tag = value\n";
+    }
+
+    while ($inner_metadata_text =~ m/<([^ >]+).*?>(.*?)<\/(.*?)>(.*)/s)
     {
         # if URL given for document as identifier metadata, store it ...
@@ -349 +422 @@
         my $metaname = $1;
         my $metavalue = $2;
-        $metadata_text = $4;
-        
-        $metaname =~ s/^(dc:)?(.)/\u$2/;
-
-        if ($metaname eq "Identifier")
+        $inner_metadata_text = $4;
+        
+        # $metaname =~ s/^(dc:)?(.)/\u$2/; # strip of optional prefix and uppercase first letter
+        $metaname =~ s/:/\./;
+        if ($metaname !~ m/\./)
         {
-        # name clashes with GSDL reserved metadata name for hash id
-        $metaname = "URL";
+        $metaname = "$top_level_prefix.$metaname";
         }
+
+
+        $metaname = $self->remap_dcterms_metadata($metaname);
+
+#       if ($metaname eq "Identifier")
+#       {
+#       # name clashes with GSDL reserved metadata name for hash id
+#       $metaname = "URL";
+#       }
 
         if (defined $metadata->{$metaname})