Changeset 10442

Timestamp:

2005-08-09T10:44:32+12:00 (19 years ago)

Author:

chi

Message:

To retrieve encoding information for the HTML file generated from word2html vb script.

File:

• trunk/gsdl/perllib/plugins/BasPlug.pm (modified) (1 diff)

trunk/gsdl/perllib/plugins/BasPlug.pm

@@ -1006 +1006 @@
     }
     }
-   
-
+    
+    # VB scripting generated Word to HTML file
+    if ($text =~ /charset=(.*)[\"]/ig){
+    my $vbhtml_encoding = $1;
+    $vbhtml_encoding =~ s/-+/_/g;
+    $self->{'input_encoding'} = $vbhtml_encoding;
+    }
+    
     # remove <title>stuff</title> -- as titles tend often to be in English
     # for foreign language documents