Changeset 6482

Timestamp:

2004-01-14T12:14:32+13:00 (20 years ago)

Author:

jrm21

Message:

use the lowercase name of the metadata has the hash key, and then do a
hash lookup for each item rather than a search of the keys.
(This is MUCH faster, especially for large collections).

File:

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

trunk/gsdl/perllib/classify/AZCompactList.pm

@@ -485 +485 @@
         }
 
-        $self->{'classifiers'}->{$metavalue} 
+        # use the lower case, for speed of lookup.
+        my $meta_lc=lc($metavalue);
+        $self->{'classifiers'}->{$meta_lc}
         = { 'classifyobj'   => $listclassobj,
             'formattednode' => $formatted_node };
@@ -510 +512 @@
 
     # find metavalue in list of sub-classifiers
-    my $found = 0;
-    my $node_name;
-    foreach $node_name (keys %{$self->{'classifiers'}}) 
+    # check if we have a key (lower case) for this metadata value
+    my $node_name=lc($metavalue);
+    if (exists $self->{'classifiers'}->{$node_name})
     {
-        $resafe_node_name = $node_name;
-        # escape chars that mean something to perl...
-        $resafe_node_name =~ s/([\\\(\)\[\]\{\}\^\$\.\+\*\?\|])/\\$1/g;
-        if ($metavalue =~ m/^$resafe_node_name$/i) 
-        {
-        my ($doc_obj,$date) = @{$self->{'reclassify'}->{$doc_OID}};
-
-        ## date appears to not be used in classifier call ####
-
-        if ($doc_OID =~ m/^[^\.]*\.([\d\.]+)$/)
-        {
-            my $section=$1;
-            if ($self->{'doclevel'} =~ m/^top/i) { # toplevel
-            $self->{'classifiers'}->{$node_name}->{'classifyobj'}
-            ->classify($doc_obj, "Section=$section");
-            } else { # section level
-            # Thanks to Don Gourley for this...
-            # classify can't handle multi-level section
-            $self->{'classifiers'}->{$node_name}->{'classifyobj'}
-            ->classify_section($section, $doc_obj, $date);
-            }
-        }
-        else
-        {
+        my ($doc_obj,$date) = @{$self->{'reclassify'}->{$doc_OID}};
+
+        ## date appears to not be used in classifier call ####
+
+        if ($doc_OID =~ m/^[^\.]*\.([\d\.]+)$/)
+        {
+        my $section=$1;
+        if ($self->{'doclevel'} =~ m/^top/i) { # toplevel
             $self->{'classifiers'}->{$node_name}->{'classifyobj'}
-            ->classify($doc_obj); 
-        }
-        
-        $found = 1;
-        last;
-        }
-    }
-    
-    if (!$found)
-    {
+            ->classify($doc_obj, "Section=$section");
+        } else { # section level
+            # Thanks to Don Gourley for this...
+            # classify can't handle multi-level section
+            $self->{'classifiers'}->{$node_name}->{'classifyobj'}
+            ->classify_section($section, $doc_obj, $date);
+        }
+        }
+        else
+        {
+        $self->{'classifiers'}->{$node_name}->{'classifyobj'}
+        ->classify($doc_obj); 
+        }
+    } else { # this key is not in the hash
         my $outhandle=$self->{outhandle};
         print $outhandle "Warning: AZCompactList::reclassify ";
@@ -730 +720 @@
     my $title = $self->{'reclassifylist'}->{$classification};
     $title =~ s/&(.){2,4};//g; # remove any HTML special chars
-### $title =~ s/^\s+//g; # remove a leading spaces
-### $title =~ s/^_+//g; # remove a leading underscores
-    $title =~ s/^\W+//g;
-### $title =~ s/^(\'|\`|\"|\:|\()//g; # remove any opening punctutation
+    $title =~ s/^\W+//g; # remove leading non-word chars
 
 # only want first character for classification
-    $title =~ m/^(.)/; $title=$1;
+    $title =~ m/^(.)/;
+    if ($1) {
+        $title=$1;
+    } else {
+        print STDERR "no first character found for \"$title\" - \"" .
+        $self->{'reclassifylist'}->{$classification} . "\"\n";
+    }
     $title =~ tr/[a-z]/[A-Z]/;