Changeset 16644

Timestamp:

2008-08-04T12:48:59+12:00 (16 years ago)

Author:

kjdon

Message:

now uses CJKTextSegmenter to add segmentation functionality to text documents

File:

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

gsdl/trunk/perllib/plugins/AutoExtractMetadata.pm

@@ -40 +40 @@
 use DateExtractor;
 use GISExtractor;
+use CJKTextSegmenter;
+
 
 sub BEGIN {
-    @AutoExtractMetadata::ISA = ( 'BasePlugin', 'AcronymExtractor', 'KeyphraseExtractor', 'EmailAddressExtractor', 'DateExtractor', 'GISExtractor' );
+    @AutoExtractMetadata::ISA = ( 'BasePlugin', 'AcronymExtractor', 'KeyphraseExtractor', 'EmailAddressExtractor', 'DateExtractor', 'CJKTextSegmenter','GISExtractor' );
 }
 
@@ -76 +78 @@
     new DateExtractor($pluginlist, $inputargs, $hashArgOptLists);
     new GISExtractor($pluginlist, $inputargs, $hashArgOptLists);
- 
+    new CJKTextSegmenter($pluginlist, $inputargs, $hashArgOptLists);
     my $self = new BasePlugin($pluginlist, $inputargs, $hashArgOptLists);
 
@@ -119 +121 @@
     $self->extract_date_metadata($doc_obj);
     $self->extract_gis_metadata($doc_obj);
+    $self->separate_cjk_text($doc_obj);
 
 }