Changeset 3720 for trunk/gsdl/perllib/plugins/PDFPlug.pm

Timestamp:

2003-01-28T16:22:26+13:00 (21 years ago)

Author:

sjboddie

Message:

Added options to PDFPlug to take advantage of the improvements in
version 0.34 of pdftohtml. It now works much better for non latin
input documents (producing UTF-8 encoded HTML).

File:

• trunk/gsdl/perllib/plugins/PDFPlug.pm (modified) (4 diffs)

trunk/gsdl/perllib/plugins/PDFPlug.pm

@@ -40 +40 @@
             'desc' => "Files matching this regular expression will be blocked from being passed to any later plugins in the list. This has no real effect other than to prevent lots of warning messages about input files you don't care about. Each plugin might have a default block_exp. e.g. by default HTMLPlug blocks any files with .gif, .jpg, .jpeg, .png or .css file extensions.",
             'type' => 'string',
-            'deft' =>  q^^ }
+            'deft' =>  q^^ },
+          { 'name' => "noimages",
+            'desc' =>  "Don't attempt to extract images from PDF.",
+            'type' => "flag" },
+          { 'name' => "complex",
+            'desc' =>  "Create more complex output. With this option set the output html will look much more like the original PDF file. For this to function properly you Ghostscript installed (for *nix gs should be on your path while for windows you must have gswin32c.exe on your path).",
+            'type' => "flag" },
+          { 'name' => "zoom",
+            'desc' =>  "The factor by which to zoomthe PDF for output (this is only useful if -complex is set).",
+            'type' => "int" }
           ];
 
@@ -50 +59 @@
 sub new {
     my $class = shift (@_);
+
+    my ($noimages, $complex, $zoom);
+    
+    if (!parsargv::parse(\@_, 
+             q^noimages^, \$noimages,
+             q^complex^, \$complex,
+             q^zoom/\d+/2^, \$zoom,
+             "allow_extra_options")) {
+
+    print STDERR "\nIncorrect options passed to PDFPlug, check your collect.cfg configuration file\n";
+    &print_usage();
+    die "\n";
+    }
 
     # following title_sub removes "Page 1" added by pdftohtml, and a leading
@@ -81 +103 @@
     my $option_list = $self->{'option_list'};
     push( @{$option_list}, $options );
+
+    # these are passed through to gsConvert.pl by ConvertToPlug.pm
+    $self->{'convert_options'} = "-pdf_zoom $zoom";
+    $self->{'convert_options'} .= " -pdf_complex" if $complex;
+    $self->{'convert_options'} .= " -pdf_ignore_images" if $noimages;
+
+    # pdftohtml will always produce html files encoded as utf-8
+    if ($self->{'input_encoding'} eq "auto") {
+    $self->{'input_encoding'} = "utf8";
+    $self->{'extract_language'} = 1;
+    }
     
     return bless $self, $class;
@@ -89 +122 @@
     print STDERR "\n  usage: plugin PDFPlug [options]\n\n";
     print STDERR "  options:\n";
-    print STDERR "   -convert_to (html|text) plugin converts to TEXT or HTML\n";
-    print STDERR "                           (default html)\n";
-    print STDERR "   -use_sections     create a separate section for each page of the PDF file.\n\n";
+    print STDERR "   -convert_to (html|text) Convert to TEXT or HTML (default html)\n";
+    print STDERR "   -use_sections     Create a separate section for each page\n";
+    print STDERR "                     of the PDF file.\n";
+    print STDERR "   -noimages         Don't attempt to extract images from PDF.\n";
+    print STDERR "   -complex          Create more complex output. With this option\n";
+    print STDERR "                     set the output html will look much more like\n";
+    print STDERR "                     the original PDF file. For this to function\n";
+    print STDERR "                     properly you Ghostscript installed (for *nix\n";
+    print STDERR "                     gs should be on your path while for windows\n";
+    print STDERR "                     you must have gswin32c.exe on your path).\n";
+    print STDERR "   -zoom             The factor by which to zoomthe PDF for output\n";
+    print STDERR "                     (this is only useful if -complex is set).\n\n";
 }