Changeset 23154

Timestamp:

2010-10-18T13:50:10+13:00 (14 years ago)

Author:

kjdon

Message:

store a hash of all doc oids, then check against this hash when asked to classify something, so that we don't classify anything twice, as this leads to the document being in the list 4 times

File:

• main/trunk/greenstone2/perllib/classify/List.pm (modified) (7 diffs)

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

@@ -273 +273 @@
     }
 
-    # An empty array for the document OIDs
+    # An empty array for the document/section OIDs that we are classifying
     $self->{'OIDs'} = [];
-
+    # A hash for all the doc ids that we have seen, so we don't classify something twice
+    $self->{'all_doc_OIDs'} = {};
     return bless $self, $class;
 }
@@ -292 +293 @@
     my ($doc_obj) = @_;
 
+    if (defined $self->{'all_doc_OIDs'}->{$doc_obj->get_OID()}) {
+    print STDERR "Warning, List classifier has already seen document ".$doc_obj->get_OID().", not classifying again\n";
+    return;
+    }
+    $self->{'all_doc_OIDs'}->{$doc_obj->get_OID()} = 1;
     # If "-classify_sections" is set, classify every section of the document
     if ($self->{'classify_sections'}) {
@@ -304 +310 @@
     $self->classify_section($doc_obj, $doc_obj->get_OID(), $doc_obj->get_top_section());
     }
-}
-
+    
+}
 
 sub classify_section
@@ -344 +350 @@
    
     # Otherwise, include this section in the classifier
+    
     push(@{$self->{'OIDs'}}, $section_OID);
 
@@ -419 +426 @@
     my @OIDs = @{shift(@_)};
     my $classifier_node = shift(@_);
-    # print STDERR "\nAdding AZ list for " . $classifier_node->{'Title'} . "\n";
-
+    
     my $metadata_group = $metadata_groups[0];
-    # print STDERR "Processing metadata group: " . $metadata_group . "\n";
-    # print STDERR "Number of OID values: " . @OIDs . "\n";
-
+    
     if (!defined($self->{$metadata_group . ".list"})) {
     print STDERR "Warning: No metadata values assigned to $metadata_group.\n";
@@ -444 +448 @@
     }
     }
-    # 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)
@@ -778 +782 @@
     {
     my @OIDs = @{$metadata_value_to_OIDs_hash_ref->{$metadata_value}};
-
     # If there is only one item and 'bookshelf_type' is not always (ie. never or duplicate_only), add the item to the list
     if (@OIDs == 1 && $self->{$metadata_group . ".bookshelf_type"} ne "always") {