Changeset 16642 for gsdl/trunk/perllib/plugins/ReadTextFile.pm

Timestamp:

2008-08-04T12:47:36+12:00 (16 years ago)

Author:

kjdon

Message:

separate_cjk option and code moved to CJKTextSegmenter, and used by AutoExtractMetadata

File:

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

gsdl/trunk/perllib/plugins/ReadTextFile.pm

@@ -34 +34 @@
 use encodings;
 use unicode;
-use cnseg;
 use textcat;
 use doc;
-eval "require diagnostics"; # some perl distros (eg mac) don't have this
 use ghtml;
 use gsprintf 'gsprintf';
@@ -73 +71 @@
     'type' => "string",
     'deft' => "en",
-    'reqd' => "no" },
-     { 'name' => "separate_cjk",
-    'desc' => "{ReadTextFile.separate_cjk}",
-    'type' => "flag",
-    'reqd' => "no"},
- ];
+    'reqd' => "no" }
+      ];
 
 
@@ -226 +220 @@
     $reader->set_encoding ($encoding);
     $reader->read_file ($textref);
-        #Now segments chinese if the separate_cjk option is set
-    if ($self->{'separate_cjk'}) {
-        # segment the Chinese words
-        $$textref = &cnseg::segment($$textref);
-    }
     }
     close FILE;