Changeset 9121

Timestamp:

2005-02-21T15:45:29+13:00 (19 years ago)

Author:

kjdon

Message:

changed encoding option to input_encoding so it gets the list of encodings from printusage, and auto detect encoding if set to auto.

File:

• trunk/gsdl/bin/script/explode_metadata_database.pl (modified) (3 diffs)

trunk/gsdl/bin/script/explode_metadata_database.pl

@@ -11 +11 @@
 use parsargv;
 use printusage;
+my $unicode_list =
+    [ { 'name' => "auto",
+    'desc' => "{BasPlug.input_encoding.auto}" },
+      { 'name' => "ascii",
+    'desc' => "{BasPlug.input_encoding.ascii}" },
+      { 'name' => "utf8",
+    'desc' => "{BasPlug.input_encoding.utf8}" },
+      { 'name' => "unicode",
+    'desc' => "{BasPlug.input_encoding.unicode}" } ];
 
 my $arguments = 
-    [ { 'name' => "encoding",
+    [ { 'name' => "input_encoding",
     'desc' => "{explode.encoding}",
-    'type' => "string",
+    'type' => "enum",
     'deft' => "iso_8859_1",
+    'list' => $unicode_list,
     'reqd' => "no" } ,
       { 'name' => "metadata_set",
@@ -44 +54 @@
     # Parse command line arguments
     if (!parsargv::parse(\@ARGV, 
-             'encoding/.*/iso_8859_1', \$encoding,
+             'input_encoding/.*/auto', \$encoding,
              'metadata_set/.*/', \$metadata_set,
              'plugin/.*/', \$plugin,
@@ -87 +97 @@
     # ...and initialize it
     $plugobj->init(1, "STDERR", "STDERR");
-
+    
+    if ($encoding eq "auto") {
+    $plugobj->{'input_encoding'} = $encoding;   
+    (my $language, $encoding) = $plugobj->textcat_get_language_encoding ($filename);
+    }
     # Use the plugin's read_file function to avoid duplicating code
     $plugobj->read_file($filename, $encoding, undef, \$text);