Changeset 15872 for gsdl/trunk/perllib/plugins/PDFPlugin.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/PDFPlugin.pm (modified) (12 diffs)

gsdl/trunk/perllib/plugins/PDFPlugin.pm

@@ -1 +1 @@
 ###########################################################################
 #
-# PDFPlug.pm -- reasonably with-it pdf plugin
+# PDFPlugin.pm -- reasonably with-it pdf plugin
 # A component of the Greenstone digital library software
 # from the New Zealand Digital Library Project at the 
@@ -23 +23 @@
 #
 ###########################################################################
-package PDFPlug;
-
-use ConvertToPlug;
+package PDFPlugin;
+
+use ConvertBinaryFile;
+use ReadTextFile;
 use unicode;
 use strict;
@@ -31 +32 @@
 
 sub BEGIN {
-    @PDFPlug::ISA = ('ConvertToPlug');
+    @PDFPlugin::ISA = ('ConvertBinaryFile', 'ReadTextFile');
 }
 
 my $convert_to_list =
     [ { 'name' => "auto",
-    'desc' => "{ConvertToPlug.convert_to.auto}" },
+    'desc' => "{ConvertBinaryFile.convert_to.auto}" },
       { 'name' => "html",
-    'desc' => "{ConvertToPlug.convert_to.html}" },
+    'desc' => "{ConvertBinaryFile.convert_to.html}" },
       { 'name' => "text",
-    'desc' => "{ConvertToPlug.convert_to.text}" },
+    'desc' => "{ConvertBinaryFile.convert_to.text}" },
       { 'name' => "pagedimg_jpg",
-    'desc' => "{ConvertToPlug.convert_to.pagedimg_jpg}"},
+    'desc' => "{ConvertBinaryFile.convert_to.pagedimg_jpg}"},
       { 'name' => "pagedimg_gif",
-    'desc' => "{ConvertToPlug.convert_to.pagedimg_gif}"},
+    'desc' => "{ConvertBinaryFile.convert_to.pagedimg_gif}"},
       { 'name' => "pagedimg_png",
-    'desc' => "{ConvertToPlug.convert_to.pagedimg_png}"}, 
+    'desc' => "{ConvertBinaryFile.convert_to.pagedimg_png}"}, 
       ];
 
@@ -53 +54 @@
     [
      { 'name' => "convert_to",
-       'desc' => "{ConvertToPlug.convert_to}",
+       'desc' => "{ConvertBinaryFile.convert_to}",
        'type' => "enum",
        'reqd' => "yes",
@@ -59 +60 @@
        'deft' => "html" },   
      { 'name' => "process_exp",
-       'desc' => "{BasPlug.process_exp}",
+       'desc' => "{BasePlugin.process_exp}",
        'type' => "regexp",
        'deft' => &get_default_process_exp(),
        'reqd' => "no" },
      { 'name' => "block_exp",
-       'desc' => "{BasPlug.block_exp}",
+       'desc' => "{BasePlugin.block_exp}",
        'type' => "regexp",
        'deft' => &get_default_block_exp() },
      { 'name' => "metadata_fields",
-       'desc' => "{HTMLPlug.metadata_fields}",
+       'desc' => "{HTMLPlugin.metadata_fields}",
        'type' => "string",
        'deft' => "" },
      { 'name' => "noimages",
-       'desc' => "{PDFPlug.noimages}",
+       'desc' => "{PDFPlugin.noimages}",
        'type' => "flag" },
      { 'name' => "allowimagesonly",
-       'desc' => "{PDFPlug.allowimagesonly}",
+       'desc' => "{PDFPlugin.allowimagesonly}",
        'type' => "flag" },
      { 'name' => "complex",
-       'desc' => "{PDFPlug.complex}",
+       'desc' => "{PDFPlugin.complex}",
        'type' => "flag" },
      { 'name' => "nohidden",
-       'desc' => "{PDFPlug.nohidden}",
+       'desc' => "{PDFPlugin.nohidden}",
        'type' => "flag" },
      { 'name' => "zoom",
-       'desc' => "{PDFPlug.zoom}",
+       'desc' => "{PDFPlugin.zoom}",
        'deft' => "2",
        'range' => "1,3", # actually the range is 0.5-3 
        'type' => "int" },
      { 'name' => "use_sections",
-       'desc' => "{PDFPlug.use_sections}",
+       'desc' => "{PDFPlugin.use_sections}",
        'type' => "flag" },
      { 'name' => "description_tags",
-       'desc' => "{HTMLPlug.description_tags}",
+       'desc' => "{HTMLPlugin.description_tags}",
        'type' => "flag" }
      ];
 
-my $options = { 'name'     => "PDFPlug",
-        'desc'     => "{PDFPlug.desc}",
+my $options = { 'name'     => "PDFPlugin",
+        'desc'     => "{PDFPlugin.desc}",
         'abstract' => "no",
         'inherits' => "yes",
@@ -111 +112 @@
     push(@$inputargs,'^(Page\s+\d+)?(\s*1\s+)?');
 
-    if(defined $arguments){ push(@{$hashArgOptLists->{"ArgList"}},@{$arguments});}
-    if(defined $options) { push(@{$hashArgOptLists->{"OptList"}},$options)};
+    push(@{$hashArgOptLists->{"ArgList"}},@{$arguments});
+    push(@{$hashArgOptLists->{"OptList"}},$options);
 
     my @arg_array = @$inputargs;
-    my $self = new ConvertToPlug($pluginlist, $inputargs, $hashArgOptLists);
+    my $self = new ConvertBinaryFile($pluginlist, $inputargs, $hashArgOptLists);
     
     if ($self->{'info_only'}) {
@@ -122 +123 @@
     }
 
-    # these are passed through to gsConvert.pl by ConvertToPlug.pm
+    $self->{'filename_extension'} = "pdf";
+    $self->{'file_type'} = "PDF";
+
+    # these are passed through to gsConvert.pl by ConvertBinaryFile.pm
     my $zoom = $self->{"zoom"};
     $self->{'convert_options'} = "-pdf_zoom $zoom";
@@ -132 +136 @@
     my $secondary_plugin_options = $self->{'secondary_plugin_options'};
 
-    if (!defined $secondary_plugin_options->{'HTMLPlug'}) {
-    $secondary_plugin_options->{'HTMLPlug'} = [];
-    }
-    if (!defined $secondary_plugin_options->{'TEXTPlug'}) {
-    $secondary_plugin_options->{'TEXTPlug'} = [];
+    if (!defined $secondary_plugin_options->{'HTMLPlugin'}) {
+    $secondary_plugin_options->{'HTMLPlugin'} = [];
+    }
+    if (!defined $secondary_plugin_options->{'TextPlugin'}) {
+    $secondary_plugin_options->{'TextPlugin'} = [];
     }
     if (defined $self->{'convert_to'} && $self->{'convert_to'} =~ /pagedimg.*/i) {
-    if (!defined $secondary_plugin_options->{'PagedImgPlug'}){
-        $secondary_plugin_options->{'PagedImgPlug'} = [];
-        my $pagedimg_options = $secondary_plugin_options->{'PagedImgPlug'}; 
+    if (!defined $secondary_plugin_options->{'PagedImagePlugin'}){
+        $secondary_plugin_options->{'PagedImagePlugin'} = [];
+        my $pagedimg_options = $secondary_plugin_options->{'PagedImagePlugin'}; 
         push(@$pagedimg_options, "-title_sub", '^(Page\s+\d+)?(\s*1\s+)?');
     }
     }
-    my $html_options = $secondary_plugin_options->{'HTMLPlug'};
-    my $text_options = $secondary_plugin_options->{'TEXTPlug'};
-    my $pagedimg_options = $secondary_plugin_options->{'PagedImgPlug'};
+    my $html_options = $secondary_plugin_options->{'HTMLPlugin'};
+    my $text_options = $secondary_plugin_options->{'TextPlugin'};
+    my $pagedimg_options = $secondary_plugin_options->{'PagedImagePlugin'};
    
     if ($self->{'input_encoding'} eq "auto") {
@@ -227 +231 @@
     && $self->{'converted_to'} eq "HTML") {
 
-    print $outhandle "PDFPlug: Calculating sections...\n";
+    print $outhandle "PDFPlugin: Calculating sections...\n";
 
     # we have "<a name=1></a>" etc for each page
@@ -236 +240 @@
 
     if (scalar (@sections) == 1) { #only one section - no split!
-        print $outhandle "PDFPlug: warning - no sections found\n";
+        print $outhandle "PDFPlugin: warning - no sections found\n";
     } else {
         $top_section .= shift @sections; # keep HTML header etc as top_section
@@ -274 +278 @@
         $title = " "; # get rid of the undefined warning in next line
         }
-        my $newsection = "<!-- from PDFPlug -->\n<!-- <Section>\n";
+        my $newsection = "<!-- from PDFPlugin -->\n<!-- <Section>\n";
         $newsection .= "<Metadata name=\"Title\">" . $title
         . "</Metadata>\n--><p>\n";
@@ -296 +300 @@
 sub process {
     my $self = shift (@_);
-    my ($textref, $pluginfo, $base_dir, $file, $metadata, $doc_obj, $gli) = @_;
+    my ($pluginfo, $base_dir, $file, $metadata, $doc_obj, $gli) = @_;
 
     my $result = $self->process_type("pdf",$base_dir,$file,$doc_obj);