Changeset 22597 for main/trunk/greenstone2/perllib/plugins/PDFPlugin.pm

Timestamp:

2010-08-10T14:31:53+12:00 (14 years ago)

Author:

kjdon

Message:

code tidy up. rearranged how convertbinaryfile plugins set up their secondary plugins - now only set up the options for the one they are using. all subclass specific code moved out of convertbinaryfile.new into the appropriate plugin file.

File:

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

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

@@ -119 +119 @@
     push(@{$hashArgOptLists->{"OptList"}},$options);
 
-    my @arg_array = @$inputargs;
     my $self = new ConvertBinaryFile($pluginlist, $inputargs, $hashArgOptLists);
     
@@ -138 +137 @@
     $self->{'convert_options'} .= " -pdf_allow_images_only" if $self->{"allowimagesonly"};
 
+    # check convert_to
+    if ($self->{'convert_to'} eq "text" && $ENV{'GSDLOS'} =~ /^windows$/i) {
+    print STDERR "Windows does not support pdf to text. PDFs will be converted to HTML instead\n";
+    $self->{'convert_to'} = "html";
+    }
+    elsif ($self->{'convert_to'} eq "auto") {
+    # choose html ?? is this the best option
+    $self->{'convert_to'} = "html";
+    }
+    # set convert_to_plugin and convert_to_ext
+    $self->ConvertBinaryFile::set_standard_convert_settings();
+
+    my $secondary_plugin_name = $self->{'convert_to_plugin'};
     my $secondary_plugin_options = $self->{'secondary_plugin_options'};
 
-    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'} =~ m/(pagedimage|pagedimg).*/i) {
-    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+)?');
-        push(@$pagedimg_options, "-screenviewsize", "1000");
-        push(@$pagedimg_options, "-enable_cache");
-    }
-    }
-    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") {
-#   $self->{'input_encoding'} = "utf8";
-#    }
-
-    # if pdftohtml is always producing utf8, then htmlplug always needs this option
-    push(@$html_options,"-input_encoding", "utf8");
-    push(@$html_options,"-extract_language") if $self->{'extract_language'};
-
-    push(@$html_options, "-processing_tmp_files");
-    push(@$pagedimg_options, "-processing_tmp_files");
-
-    # Instruct HTMLPlug (when eventually accessed through read_into_doc_obj) 
-    # to extract these metadata fields from the HEAD META fields
-    my $required_metadata;
-    if (defined $self->{'metadata_fields'} && $self->{'metadata_fields'} =~ /\S/) {
-    push(@$html_options,"-metadata_fields",$self->{'metadata_fields'});
-    } else {
-    push(@$html_options,"-metadata_fields","Title,GENERATOR,date,author<Creator>");
-    }
-    if (defined $self->{'metadata_field_separator'} && $self->{'metadata_field_separator'} =~ /\S/) {
-    push(@$html_options,"-metadata_field_separator",$self->{'metadata_field_separator'});
-    }
-        
-    if ($self->{'use_sections'} || $self->{'description_tags'}) {
-    $self->{'description_tags'} = 1;
-    push(@$html_options,"-description_tags");
-    }
+    if (!defined $secondary_plugin_options->{$secondary_plugin_name}) {
+    $secondary_plugin_options->{$secondary_plugin_name} = [];
+    }
+    my $specific_options = $secondary_plugin_options->{$secondary_plugin_name};
 
     # following title_sub removes "Page 1" added by pdftohtml, and a leading
     # "1", which is often the page number at the top of the page. Bad Luck
     # if your document title actually starts with "1 " - is there a better way?
-    push(@$html_options , "-title_sub", '^(Page\s+\d+)?(\s*1\s+)?');
-    push(@$text_options , "-title_sub", '^(Page\s+\d+)?(\s*1\s+)?');
-
+    push(@$specific_options , "-title_sub", '^(Page\s+\d+)?(\s*1\s+)?');
     my $associate_tail_re = $self->{'associate_tail_re'};
     if ((defined $associate_tail_re) && ($associate_tail_re ne "")) {
-    push(@$html_options, "-associate_tail_re", $associate_tail_re);
-    push(@$text_options, "-associate_tail_re", $associate_tail_re);
-    push(@$pagedimg_options, "-associate_tail_re", $associate_tail_re) if defined $pagedimg_options;
-    }
-
-    push(@$html_options, "-file_rename_method", "none");
-    push(@$text_options, "-file_rename_method", "none");
-    push(@$pagedimg_options, "-file_rename_method", "none") if defined $pagedimg_options;
+    push(@$specific_options, "-associate_tail_re", $associate_tail_re);
+    }
+    push(@$specific_options, "-file_rename_method", "none");
+    
+    if ($secondary_plugin_name eq "HTMLPlugin") {
+    # pdftohtml always produces utf8
+    push(@$specific_options, "-input_encoding", "utf8");
+    push(@$specific_options, "-extract_language") if $self->{'extract_language'};
+    push(@$specific_options, "-processing_tmp_files");
+    # Instruct HTMLPlug (when eventually accessed through read_into_doc_obj) 
+    # to extract these metadata fields from the HEAD META fields
+    if (defined $self->{'metadata_fields'} && $self->{'metadata_fields'} =~ /\S/) {
+        push(@$specific_options,"-metadata_fields",$self->{'metadata_fields'});
+    } else {
+        push(@$specific_options,"-metadata_fields","Title,GENERATOR,date,author<Creator>");
+    }
+    if (defined $self->{'metadata_field_separator'} && $self->{'metadata_field_separator'} =~ /\S/) {
+        push(@$specific_options,"-metadata_field_separator",$self->{'metadata_field_separator'});
+    }
+    if ($self->{'use_sections'} || $self->{'description_tags'}) {
+        $self->{'description_tags'} = 1;
+        push(@$specific_options, "-description_tags");
+    }
+    }
+    elsif ($secondary_plugin_name eq "PagedImagePlugin") {
+    push(@$specific_options, "-screenviewsize", "1000");
+    push(@$specific_options, "-enable_cache");
+    push(@$specific_options, "-processing_tmp_files");
+    }
 
     $self = bless $self, $class;