Changeset 10428

Timestamp:

2005-08-05T15:21:41+12:00 (19 years ago)

Author:

chi

Message:

Modification of the way passing argument and option lists for the secondary plugin. Also, add an
option (extracted_word_metadata) to retrieve metadata based on user-defined fields from the html document
converted by VB scripting.

File:

• trunk/gsdl/perllib/plugins/WordPlug.pm (modified) (6 diffs)

trunk/gsdl/perllib/plugins/WordPlug.pm

@@ -94 +94 @@
              'type' => "regexp",
              'reqd' => "no",
-             'deft' => "" }];
+             'deft' => "" },
+               { 'name' => "extracted_word_metadata_fields",
+             'type' => "string",
+             'deft' => "Title"}];
     
     push(@$arguments,@$ws_arg);
@@ -101 +104 @@
     if(defined $arguments){ push(@{$hashArgOptLists->{"ArgList"}},@{$arguments});}
     if(defined $options) { push(@{$hashArgOptLists->{"OptList"}},$options)};
-
-    #foreach my $arg (@{$hashArgOptLists->{"ArgList"}})
-    #{
-    #   print STDERR "**** Word Argument =$arg->{'name'}\n";
-    #}
-    
+    my @arg_array = @$inputargs;
     my $self = (defined $hashArgOptLists)? new ConvertToPlug($pluginlist,$inputargs,$hashArgOptLists): new ConvertToPlug($pluginlist,$inputargs);
 
@@ -115 +113 @@
     if (defined $self->{'windows_scripting'}) { 
     if (!defined $secondary_plugin_options->{'StructuredHTMLPlug'}){
-        $secondary_plugin_options->{'StructuredHTMLPlug'} =[];
+        $secondary_plugin_options->{'StructuredHTMLPlug'} = [];
         my $structhtml_options = $secondary_plugin_options->{'StructuredHTMLPlug'};
+        if ($self->{'input_encoding'} eq "auto") {
+        $self->{'input_encoding'} = "utf8";
+        $self->{'extract_language'} = 1;
+        push(@$structhtml_options,"-input_encoding", "utf8");
+        push(@$structhtml_options,"-extract_language");
+                
+        # Instruct HTMLPlug (when eventually accessed through read_into_doc_obj)
+        # to extract these metadata fields from the HEAD META fields
+        push(@$structhtml_options,"-metadata_fields","Title,GENERATOR,date,author<Creator>");
+        push(@$structhtml_options , "-title_sub", '^(Page\s+\d+)?(\s*1\s+)?');
+        }    
     }
     }
@@ -139 +148 @@
     # to extract these metadata fields from the HEAD META fields
     push(@$html_options,"-metadata_fields","Title,GENERATOR,date,author<Creator>");
-    #push(@$html_options , "-title_sub", '^(Page\s+\d+)?(\s*1\s+)?');
+    push(@$html_options , "-title_sub", '^(Page\s+\d+)?(\s*1\s+)?');
     }
     
     $self = bless $self, $class;
-    $self->load_secondary_plugins($class,$secondary_plugin_options);
+    $self->load_secondary_plugins($class,$secondary_plugin_options,$hashArgOptLists);
 
     return bless $self;
@@ -171 +180 @@
 
     # Write it out again!
+    $self->utf8_write_file (\$text, $conv_filename);
 }
 
@@ -206 +216 @@
     my $self = shift (@_);
     my ($textref, $pluginfo, $base_dir, $file, $metadata, $doc_obj, $gli) = @_;
-   
+    
     return $self->process_type("doc", $base_dir, $file, $doc_obj);
 }