Changeset 23349 for main/trunk/greenstone2/perllib/plugins/ReadXMLFile.pm

Timestamp:

2010-11-26T12:09:53+13:00 (13 years ago)

Author:

davidb

Message:

More careful use of encoding parameter to $self->set_Source_metadata so it reflects the *filename* encoding that Greenstone has worked out, not the encoding for the *content* of the file, which of course could be completely different!

File:

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

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

@@ -213 +213 @@
     $self->{'filename_no_path'} = $filename_no_path;
     $self->{'processor'} = $processor;
+
     # this contains metadata passed in from running metadata_read with other plugins (eg from MetadataXMLPlugin)
     # we are also using it to store up any metadata found during parsing the XML, so that it can be added to the doc obj.
@@ -368 +369 @@
     my $self = shift(@_);
 
+    my $metadata = $self->{'metadata'};
+
     # create a new document
-    $self->{'doc_obj'} = new doc ($self->{'filename'}, "indexed_doc", $self->{'file_rename_method'});
-    $self->{'doc_obj'}->add_utf8_metadata($self->{'doc_obj'}->get_top_section(), "Plugin", "$self->{'plugin_type'}");
-    $self->set_Source_metadata($self->{'doc_obj'}, $self->{'filename_no_path'});
-    
-
+    my $doc_obj = $self->{'doc_obj'} = new doc ($self->{'filename'}, "indexed_doc", $self->{'file_rename_method'});
+
+    $doc_obj->add_utf8_metadata($doc_obj->get_top_section(), "Plugin", "$self->{'plugin_type'}");
+
+    my $filename_no_path = $self->{'filename_no_path'};
+    my $plugin_filename_encoding = $self->{'filename_encoding'};
+    my $filename_encoding = $self->deduce_filename_encoding($filename_no_path,$metadata,$plugin_filename_encoding);
+
+    $self->set_Source_metadata($doc_obj, $filename_no_path, $filename_encoding);
+    
     # do we want other auto metadata here (see BasePlugin.read_into_doc_obj)
 }