Changeset 14934 for gsdl/trunk/perllib/basebuilder.pm

Timestamp:

2007-12-20T21:53:14+13:00 (16 years ago)

Author:

davidb

Message:

Changes to allow statistic calculations for metadata coverage, i.e. for this docment which metadata set prefixes are used, which fields within those prefixes are used, and how many times. This is then agregated over the all documents and the summary stored as collection level metadata.

File:

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

gsdl/trunk/perllib/basebuilder.pm

@@ -368 +368 @@
     $self->{'buildproc'}->zero_reset(); 
 
+    $self->{'buildproc'}->{'mdprefix_fields'} = {};
+
     if ($self->{'keepold'}) {
     # create flat classify structure, ready for new docs to be added
@@ -377 +379 @@
 
    
+    &plugin::read ($self->{'pluginfo'}, $self->{'source_dir'}, 
+           "", {}, $self->{'buildproc'}, $self->{'maxdocs'},0, $self->{'gli'});
+
     # this has changed to only output collection meta if its 
     # not in the config file
     $self->output_collection_meta($handle);
-    &plugin::read ($self->{'pluginfo'}, $self->{'source_dir'}, 
-           "", {}, $self->{'buildproc'}, $self->{'maxdocs'},0, $self->{'gli'});
     
     # output classification information
@@ -562 +565 @@
 }
 
-# default is to output an empty [collection] entry
+
+
+sub output_collection_meta_start {
+    my $self = shift(@_);
+    my ($handle) = @_;
+    
+    print $handle "[collection]\n";
+
+}  
+
+sub output_collection_meta_sets {
+    my $self = shift(@_);
+    my ($handle) = @_;
+
+    my $mdprefix_fields = $self->{'buildproc'}->{'mdprefix_fields'};
+
+    foreach my $prefix (keys %$mdprefix_fields)
+    {
+    print $handle "<metadataset>$prefix\n";
+
+    foreach my $field (keys %{$mdprefix_fields->{$prefix}})
+    {
+        my $val = $mdprefix_fields->{$prefix}->{$field};
+
+        print $handle "<metadatalist-$prefix>$field\n";
+        print $handle "<metadatafreq-$prefix-$field>$val\n";
+    }
+
+    }
+
+}  
+
+sub output_collection_meta_end {
+    my $self = shift(@_);
+    my ($handle) = @_;
+    
+    print $handle ('-' x 70) . "\n";;
+
+}  
+
+
+# default is to output the metadata sets (prefixes) used in collection
+
 sub output_collection_meta {
     my $self = shift(@_);
     my ($handle) = @_;
-    
-    print $handle "[collection]\n". ('-' x 70) . "\n";;
+
+    $self->output_collection_meta_start($handle);
+    $self->output_collection_meta_sets($handle);
+    $self->output_collection_meta_end($handle);
 
 }