Changeset 12896

Timestamp:

2006-09-28T13:40:40+12:00 (18 years ago)

Author:

mdewsnip

Message:

More tidying up of the code.

File:

• trunk/gsdl/perllib/classify/GenericList.pm (modified) (4 diffs)

trunk/gsdl/perllib/classify/GenericList.pm

@@ -174 +174 @@
 
 
+# Called for each document in the collection
 sub classify
 {
@@ -179 +180 @@
     my $doc_obj = shift(@_);
 
-    my $doc_OID = $doc_obj->get_OID();
-
+    # If "-classify_sections" is set, classify every section of the document
     if ($self->{'classify_sections'}) {
     my $section = $doc_obj->get_next_section($doc_obj->get_top_section());
     while (defined $section) {
-        $self->classify_section($doc_obj, $doc_OID . ".$section", $section);
+        $self->classify_section($doc_obj, $doc_obj->get_OID() . ".$section", $section);
         $section = $doc_obj->get_next_section($section);
     }
     }
+    # Otherwise just classify the top document section
     else {
-    $self->classify_section($doc_obj, $doc_OID, $doc_obj->get_top_section());
+    $self->classify_section($doc_obj, $doc_obj->get_OID(), $doc_obj->get_top_section());
     }
 }
@@ -198 +199 @@
     my $self = shift(@_);
     my $doc_obj = shift(@_);
-    my $doc_OID = shift(@_);
+    my $section_OID = shift(@_);
     my $section = shift(@_);
 
     my @metadata_groups = @{$self->{'metadata_groups'}};
 
-    # Only classify the document if it has a value for one of the metadata elements in the first group
-    foreach my $firstmetagroupelem (split(/\;/, $metadata_groups[0])) {
-    my $firstmetagroupelemvalue = $doc_obj->get_metadata_element($section, $firstmetagroupelem);
-    if (defined($firstmetagroupelemvalue) && $firstmetagroupelemvalue ne "") {
-        push(@{$self->{'OIDs'}}, $doc_OID);
-
-        # Create a hash for the metadata values of each metadata element we're interested in
-        my %metagroupsdone = ();
-        foreach my $metadata_group (@metadata_groups, @{$self->{'sort_leaf_nodes_using_metadata_groups'}}) {
-        # Take care not to do a metadata group more than once
-        unless ($metagroupsdone{$metadata_group}) {
-            foreach my $metaelem (split(/\;/, $metadata_group)) {
-            my @metavalues = @{$doc_obj->get_metadata($section, $metaelem)};
-            foreach my $metavalue (@metavalues) {
-                # Strip leading and trailing whitespace
-                $metavalue =~ s/^\s*//;
-                $metavalue =~ s/\s*$//;
-                push(@{$self->{$metadata_group . ".list"}->{$doc_OID}}, $metavalue);
-            }
-            last if (@metavalues > 0);
-            }
-
-            $metagroupsdone{$metadata_group} = 1;
+    # Only classify the section if it has a value for one of the metadata elements in the first group
+    my $classify_section = 0;
+    my $first_metadata_group = $metadata_groups[0];
+    foreach my $first_metadata_group_element (split(/\;/, $first_metadata_group)) {
+    my $first_metadata_group_element_value = $doc_obj->get_metadata_element($section, $first_metadata_group_element);
+    if (defined($first_metadata_group_element_value) && $first_metadata_group_element_value ne "") {
+        # This section must be included in the classifier
+        $classify_section = 1;
+        last;
+    }
+    }
+
+    # We're not classifying this section because it doesn't have the required metadata
+    return if (!$classify_section);
+
+    # Otherwise, include this section in the classifier
+    push(@{$self->{'OIDs'}}, $section_OID);
+
+    # Create a hash for the metadata values of each metadata element we're interested in
+    my %metadata_groups_done = ();
+    foreach my $metadata_group (@metadata_groups, @{$self->{'sort_leaf_nodes_using_metadata_groups'}}) {
+    # Take care not to do a metadata group more than once
+    unless ($metadata_groups_done{$metadata_group}) {
+        foreach my $metadata_element (split(/\;/, $metadata_group)) {
+        my @metadata_values = @{$doc_obj->get_metadata($section, $metadata_element)};
+        foreach my $metadata_value (@metadata_values) {
+            # Strip leading and trailing whitespace
+            $metadata_value =~ s/^\s*//;
+            $metadata_value =~ s/\s*$//;
+            push(@{$self->{$metadata_group . ".list"}->{$section_OID}}, $metadata_value);
         }
-        }
-
-        last;
+        last if (@metadata_values > 0);
+        }
+
+        $metadata_groups_done{$metadata_group} = 1;
     }
     }
@@ -239 +248 @@
     my $self = shift(@_);
 
-    # The metadata elements to classify by
+    # The metadata groups to classify by
     my @metadata_groups = @{$self->{'metadata_groups'}};
     my $first_metadata_group = $metadata_groups[0];
 
-    # The OID values of the documents to include in the classification
+    # The OID values of the documents to include in the classifier
     my @OIDs = @{$self->{'OIDs'}};
 
-    # The root node of the classification hierarchy
+    # Create the root node of the classification hierarchy
     my %classifier_node = ( 'thistype' => "Invisible",
                 'childtype' => $self->{$first_metadata_group . ".list_type"},