Changeset 15872 for gsdl/trunk/perllib/plugins/WordPlugin.pm

Timestamp:

2008-06-05T09:29:32+12:00 (16 years ago)

Author:

kjdon

Message:

plugin overhaul: plugins renamed to xxPlugin, and in some cases the names are made more sensible. They now use the new base plugins. Hopefully we have better code reuse. Some of the plugins still need work done as I didn't want to spend another month doing this before committing it. Alos, I haven't really tested anything yet...

File:

• gsdl/trunk/perllib/plugins/WordPlugin.pm (modified) (12 diffs)

gsdl/trunk/perllib/plugins/WordPlugin.pm

@@ -1 +1 @@
 ###########################################################################
 #
-# WordPlug.pm -- plugin for importing Microsoft Word documents
+# WordPlugin.pm -- plugin for importing Microsoft Word documents
 # A component of the Greenstone digital library software
 # from the New Zealand Digital Library Project at the 
@@ -25 +25 @@
 # 12/05/02 Added usage datastructure - John Thompson
 
-package WordPlug;
-
-use ConvertToPlug;
+package WordPlugin;
+
+use ConvertBinaryFile;
 use strict;
 no strict 'refs'; # allow filehandles to be variables and viceversa
 
 sub BEGIN {
-    @WordPlug::ISA = ('ConvertToPlug');
+    @WordPlugin::ISA = ('ConvertBinaryFile');
 }
 
 my $arguments =
     [ { 'name' => "process_exp",
-    'desc' => "{BasPlug.process_exp}",
+    'desc' => "{BasePlugin.process_exp}",
     'type' => "regexp",
     'deft' => &get_default_process_exp(),
     'reqd' => "no" },
       { 'name' => "description_tags",
-    'desc' => "{HTMLPlug.description_tags}",
+    'desc' => "{HTMLPlugin.description_tags}",
     'type' => "flag" }
       ];
 
-my $options = { 'name'     => "WordPlug",
-        'desc'     => "{WordPlug.desc}",
+my $options = { 'name'     => "WordPlugin",
+        'desc'     => "{WordPlugin.desc}",
         'abstract' => "no",
         'inherits' => "yes",
@@ -60 +60 @@
     if ($ENV{'GSDLOS'} =~ m/^windows$/i) {
     my $ws_arg = [ { 'name' => "windows_scripting",
-             'desc' => "{WordPlug.windows_scripting}",
+             'desc' => "{WordPlugin.windows_scripting}",
              'type' => "flag",
                  'reqd' => "no" },
@@ -67 +67 @@
              'deft' => "Title" },
                { 'name' => "level1_header",
-             'desc' => "{StructuredHTMLPlug.level1_header}",
+             'desc' => "{StructuredHTMLPlugin.level1_header}",
              'type' => "regexp",
              'reqd' => "no",
              'deft' => "" },
                { 'name' => "level2_header",
-             'desc' => "{StructuredHTMLPlug.level2_header}",
+             'desc' => "{StructuredHTMLPlugin.level2_header}",
              'type' => "regexp",
              'reqd' => "no",
              'deft' => "" },
                { 'name' => "level3_header",
-             'desc' => "{StructuredHTMLPlug.level3_header}",
+             'desc' => "{StructuredHTMLPlugin.level3_header}",
              'type' => "regexp",
              'reqd' => "no",
              'deft' => "" },
                { 'name' => "title_header",
-             'desc' => "{StructuredHTMLPlug.title_header}",
+             'desc' => "{StructuredHTMLPlugin.title_header}",
              'type' => "regexp",
              'reqd' => "no",
              'deft' => "" },
                { 'name' => "delete_toc",
-             'desc' => "{StructuredHTMLPlug.delete_toc}",
+             'desc' => "{StructuredHTMLPlugin.delete_toc}",
              'type' => "flag",
              'reqd' => "no",
@@ -94 +94 @@
              'modegli' => "3"},
                { 'name' => "toc_header",
-             'desc' => "{StructuredHTMLPlug.toc_header}",
+             'desc' => "{StructuredHTMLPlugin.toc_header}",
              'type' => "regexp",
              'reqd' => "no",
@@ -103 +103 @@
     }
 
-    if(defined $arguments){ push(@{$hashArgOptLists->{"ArgList"}},@{$arguments});}
-    if(defined $options) { push(@{$hashArgOptLists->{"OptList"}},$options)};
-
-    my $self = new ConvertToPlug($pluginlist, $inputargs, $hashArgOptLists);
+    push(@{$hashArgOptLists->{"ArgList"}},@{$arguments});
+    push(@{$hashArgOptLists->{"OptList"}},$options);
+
+    my $self = new ConvertBinaryFile($pluginlist, $inputargs, $hashArgOptLists);
 
     if ($self->{'info_only'}) {
@@ -113 +113 @@
     }
 
-    #this is passed through to gsConvert.pl by ConvertToPlug.pm
+    $self->{'filename_extension'} = "doc";
+    $self->{'file_type'} = "Word";
+
+    #this is passed through to gsConvert.pl by ConvertBinaryFile.pm
     $self->{'convert_options'} = "-windows_scripting" if $self->{'windows_scripting'};
 
@@ -123 +126 @@
     my $secondary_plugin_options = $self->{'secondary_plugin_options'};
     if (defined $self->{'windows_scripting'}) { 
-    if (!defined $secondary_plugin_options->{'StructuredHTMLPlug'}){
-        $secondary_plugin_options->{'StructuredHTMLPlug'} = [];
-        my $structhtml_options = $secondary_plugin_options->{'StructuredHTMLPlug'};
+    if (!defined $secondary_plugin_options->{'StructuredHTMLPlugin'}){
+        $secondary_plugin_options->{'StructuredHTMLPlugin'} = [];
+        my $structhtml_options = $secondary_plugin_options->{'StructuredHTMLPlugin'};
         
-        # Instruct HTMLPlug (when eventually accessed through read_into_doc_obj)
+        # Instruct HTMLPlugin (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>");
@@ -142 +145 @@
     }
     }
-    if (!defined $secondary_plugin_options->{'HTMLPlug'}) {
-    $secondary_plugin_options->{'HTMLPlug'} = [];
-    }
-    if (!defined $secondary_plugin_options->{'TEXTPlug'}) {
-    $secondary_plugin_options->{'TEXTPlug'} = [];
-    }
-
-    my $html_options = $secondary_plugin_options->{'HTMLPlug'};
-    my $text_options = $secondary_plugin_options->{'TextPlug'};
-    my $structhtml_options = $secondary_plugin_options->{'StructuredHTMLPlug'};    
-    # wvWare will always produce html files encoded as utf-8, so make sure the secondary HTMLPlug knows this
+    if (!defined $secondary_plugin_options->{'HTMLPlugin'}) {
+    $secondary_plugin_options->{'HTMLPlugin'} = [];
+    }
+    if (!defined $secondary_plugin_options->{'TextPlugin'}) {
+    $secondary_plugin_options->{'TextPlugin'} = [];
+    }
+
+    my $html_options = $secondary_plugin_options->{'HTMLPlugin'};
+    my $text_options = $secondary_plugin_options->{'TextPlugin'};
+    my $structhtml_options = $secondary_plugin_options->{'StructuredHTMLPlugin'};    
+    # wvWare will always produce html files encoded as utf-8, so make sure the secondary HTMLPlugin knows this
     push(@$html_options,"-input_encoding", "utf8");
     push(@$html_options,"-extract_language") if $self->{'extract_language'};
     push(@$html_options, "-description_tags") if $self->{'description_tags'};
 
-    # Instruct HTMLPlug (when eventually accessed through read_into_doc_obj)
+    # Instruct HTMLPlugin (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>");
@@ -181 +184 @@
 }
 
-sub convert_post_process
+sub convert_post_process_old
 {
     my $self = shift (@_);
@@ -199 +202 @@
     # Write it out again!
     #$self->utf8_write_file (\$text, $conv_filename);
-}
-
-sub get_file_type {
-    my $self = shift (@_);
-    my $file_type = "Word";
-    return $file_type;
 }
 
@@ -230 +227 @@
 }
 
-# do plugin specific processing of doc_obj for HTML type
-sub process {
-    my $self = shift (@_);
-    my ($textref, $pluginfo, $base_dir, $file, $metadata, $doc_obj, $gli) = @_;
-
-    return $self->process_type("doc", $base_dir, $file, $doc_obj);
-}
 
 1;