Changeset 10441 for trunk/gsdl/perllib/plugins/WordPlug.pm

Timestamp:

2005-08-09T10:43:13+12:00 (19 years ago)

Author:

chi

Message:

Modifications for pushing required option and argument lists to StructuredHTMLPlug, also comments out
ensure_utf8() as it seems messing up the document text with UTF8 encoding.

File:

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

trunk/gsdl/perllib/plugins/WordPlug.pm

@@ -104 +104 @@
     if(defined $arguments){ push(@{$hashArgOptLists->{"ArgList"}},@{$arguments});}
     if(defined $options) { push(@{$hashArgOptLists->{"OptList"}},$options)};
-    my @arg_array = @$inputargs;
+
     my $self = (defined $hashArgOptLists)? new ConvertToPlug($pluginlist,$inputargs,$hashArgOptLists): new ConvertToPlug($pluginlist,$inputargs);
 
@@ -120 +120 @@
         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+)?');
-        }    
+        # 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+)?');
+        push (@$structhtml_options, "-description_tags") if $self->{'windows_scripting'};
+        push (@$structhtml_options, "-checkout_toc") if $self->{'checkout_toc'};
+        push (@$structhtml_options, "-toc_header", $self->{'toc_header'}) if $self->{'toc_header'};
+        push (@$structhtml_options, "-tof_header", $self->{'tof_header'}) if $self->{'tof_header'};
+        push (@$structhtml_options, "-title_header", $self->{'title_header'}) if $self->{'title_header'};
+        push (@$structhtml_options, "-level1_header", $self->{'level1_header'}) if $self->{'level1_header'};
+        push (@$structhtml_options, "-level2_header", $self->{'level2_header'})if $self->{'level2_header'};
+        push (@$structhtml_options, "-level3_header", $self->{'level3_header'}) if $self->{'level3_header'};
+        push (@$structhtml_options, "-extracted_word_metadata_fields", $self->{'extracted_word_metadata_fields'}) if $self->{'extracted_word_metadata_fields'};
     }
     }
@@ -144 +153 @@
     push(@$html_options,"-input_encoding", "utf8");
     push(@$html_options,"-extract_language");
-
-    # Instruct HTMLPlug (when eventually accessed through read_into_doc_obj)
-    # to extract these metadata fields from the HEAD META fields
-    push(@$html_options,"-metadata_fields","Title,GENERATOR,date,author<Creator>");
-    push(@$html_options , "-title_sub", '^(Page\s+\d+)?(\s*1\s+)?');
-    }
-    
+    }
+
+    # Instruct HTMLPlug (when eventually accessed through read_into_doc_obj)
+    # to extract these metadata fields from the HEAD META fields
+    push(@$html_options,"-metadata_fields","Title,GENERATOR,date,author<Creator>");
+    push(@$html_options , "-title_sub", '^(Page\s+\d+)?(\s*1\s+)?');
+       
     $self = bless $self, $class;
     $self->load_secondary_plugins($class,$secondary_plugin_options,$hashArgOptLists);
@@ -169 +178 @@
 
     my $outhandle=$self->{'outhandle'};
-
+     
     my ($language, $encoding) = $self->textcat_get_language_encoding ($conv_filename);
 
@@ -177 +186 @@
 
     # turn any high bytes that aren't valid utf-8 into utf-8.
-    unicode::ensure_utf8(\$text);
-
+    #unicode::ensure_utf8(\$text);
+    
     # Write it out again!
-    $self->utf8_write_file (\$text, $conv_filename);
+    #$self->utf8_write_file (\$text, $conv_filename);
 }
 
@@ -216 +225 @@
     my $self = shift (@_);
     my ($textref, $pluginfo, $base_dir, $file, $metadata, $doc_obj, $gli) = @_;
-    
+
     return $self->process_type("doc", $base_dir, $file, $doc_obj);
 }