Changeset 24012 for main/trunk/greenstone2

Timestamp:

2011-05-11T16:47:52+12:00 (13 years ago)

Author:

ak19

Message:

Bugfix to the way List deals with subclassification levels specified by the forward slash token. The bug was that when sub classifications were not specified for a document, it would never appear (even in a classification level higher up, which was specified for it). Now such docs appears under the lowest classification level specified for them. E.g. dc.Creator/dc.Date. If Date is not specified for doc x even if Creator is, then x used to not appear under dc.Creator either. Now it appears under dc.Creator.

File:

• main/trunk/greenstone2/perllib/classify/List.pm (modified) (6 diffs)

main/trunk/greenstone2/perllib/classify/List.pm

@@ -142 +142 @@
     die "Error: No metadata fields specified for List.\n";
     }
-    my @metadata_groups = split(/\//, $self->{'metadata'});
+
+    my @metadata_groups = split(/[\/]/, $self->{'metadata'});
     $self->{'metadata_groups'} = \@metadata_groups;
 
@@ -148 +149 @@
     if (!$self->{'buttonname'}) {
     my $first_metadata_group = $metadata_groups[0];
-    my $first_metadata_element = (split(/\;|,/, $first_metadata_group))[0];
+    my $first_metadata_element = (split(/[\;|,\/]/, $first_metadata_group))[0]; 
     $self->{'buttonname'} = $self->generate_title_from_metadata($first_metadata_element);
     }
@@ -451 +452 @@
     }
     }
-     #print STDERR "Number of distinct values: " . scalar(keys %metadata_value_to_OIDs_hash) . "\n";
+    #print STDERR "Number of distinct values: " . scalar(keys %metadata_value_to_OIDs_hash) . "\n";
 
     # Partition the values (if necessary)
@@ -495 +496 @@
     my @metadata_values_in_bucket = ();
     my $num_items_in_bucket = 0;
-    foreach my $metadata_value (@sortedmetadata_values) {       
+    foreach my $metadata_value (@sortedmetadata_values) {
         my $metadata_valuepartition = substr($metadata_value, 0, 1);
         if ($metadata_valuepartition ne $last_partition) {
@@ -801 +802 @@
     }
     # Otherwise create a sublist (bookshelf) for the metadata value
-    else {          
+    else {
         my %child_classifier_node = ( 'Title' => $self->convert_unicode_string_to_utf8_string($metadata_value),
                       'childtype' => "VList",
@@ -807 +808 @@
                       'contains' => [] );
 
+        #@OIDs = $self->sort_leaf_items(\@OIDs);
         # If there are metadata elements remaining, recursively apply the process
         if (@metadata_groups > 0) {
-        my $next_metadata_group = $metadata_groups[0];
+        my $next_metadata_group = $metadata_groups[0];      
         $child_classifier_node{'childtype'} = $self->{$next_metadata_group . ".list_type"};
-        $self->add_level(\@metadata_groups, \@OIDs, \%child_classifier_node);
+
+        # separate metadata into those that below in the next/sub-metadata_group
+        # and those that below at the current level's metadata_group
+
+        my $OID_to_metadata_values_hash_ref = $self->{$next_metadata_group . ".list"};
+        my @current_level_OIDs = ();
+        my @next_level_OIDs = ();
+        foreach my $OID (@OIDs)
+        {
+            if ($OID_to_metadata_values_hash_ref->{$OID}) {
+            push(@next_level_OIDs, $OID);
+            } else {
+            push(@current_level_OIDs, $OID);
+            }
+        }
+        # recursively process those docs belonging to the sub-metadata_group
+        $self->add_level(\@metadata_groups, \@next_level_OIDs, \%child_classifier_node);
+
+        # For those docs that don't belong in the sub/next_metadata_group, but which belong
+        # at this level, just add the documents as children of this list at the current level
+        @current_level_OIDs = $self->sort_leaf_items(\@current_level_OIDs);
+        foreach my $current_level_OID (@current_level_OIDs) {
+            my $offset = $self->metadata_offset($metadata_group, $current_level_OID, $metadata_value);
+            push(@{$child_classifier_node{'contains'}}, { 'OID' => $current_level_OID , 'offset' => $offset });
+        }       
         }
         # Otherwise just add the documents as children of this list