Changeset 23335 for main/trunk/greenstone2/perllib/plugins/ReadTextFile.pm

Timestamp:

2010-11-19T13:29:29+13:00 (13 years ago)

Author:

davidb

Message:

Work done on improving handing of filenames when the actualy filename encoding used is not necesarrily known. Tested for Linux. Work currently includes some debug statements that will be removed once testing for Windows and Mac is done.

File:

• main/trunk/greenstone2/perllib/plugins/ReadTextFile.pm (modified) (2 diffs)

main/trunk/greenstone2/perllib/plugins/ReadTextFile.pm

@@ -128 +128 @@
 
     # Do encoding stuff
-    my ($language, $encoding) = $self->textcat_get_language_encoding ($filename_full_path);
+    my ($language, $content_encoding) = $self->textcat_get_language_encoding ($filename_full_path);
     if ($self->{'verbosity'} > 2) {
-    print $outhandle "ReadTextFile: reading $file as ($encoding,$language)\n";
+    print $outhandle "ReadTextFile: reading $file as ($content_encoding,$language)\n";
     }
 
@@ -140 +140 @@
     $doc_obj->add_utf8_metadata($top_section, "Plugin", "$self->{'plugin_type'}");
     $doc_obj->add_utf8_metadata($top_section, "FileSize", (-s $filename_full_path));
-    $self->set_Source_metadata($doc_obj, $filename_no_path, $encoding);
+
+    my $filename_encoding = $self->deduce_filename_encoding($file,$metadata);
+    $self->set_Source_metadata($doc_obj, $filename_no_path, $filename_encoding);
 
     $doc_obj->add_utf8_metadata($top_section, "Language", $language);
-    $doc_obj->add_utf8_metadata($top_section, "Encoding", $encoding);
+    $doc_obj->add_utf8_metadata($top_section, "Encoding", $content_encoding);
     
     # read in file ($text will be in utf8)
     my $text = "";
-    $self->read_file ($filename_full_path, $encoding, $language, \$text);
+    $self->read_file ($filename_full_path, $content_encoding, $language, \$text);
 
     if (!length ($text)) {