origami 1.2.1 → 1.2.2

data/README

@@ -27,7 +27,7 @@ INSTALL
 
 DIRECTORIES
 
-  ``origami/``
+  ``lib/``
     Core scripts used to parse a PDF file. All objects and features are
     provided here.
 

data/bin/gui/hexview.rb

@@ -49,7 +49,7 @@ module PDFWalker
 
         @valuebuffer.create_tag( "HexView",
           :weight => Pango::WEIGHT_BOLD, 
-          :foreground => "black", 
+          #:foreground => "black", 
           :family => "Courier", 
           :scale => Pango::AttrScale::LARGE
         )

data/bin/gui/menu.rb

@@ -152,14 +152,14 @@ module PDFWalker
         :Sensitive => true,
         :Callback => lambda { |widget, viewer, path|
           stm = viewer.model.get_value(viewer.model.get_iter(path), viewer.class::OBJCOL)
-          w,h = stm[:Width], stm[:Height]
-          bpp = stm[:BitsPerComponent] || 8
-          bpr = (bpp >> 3) * 3 * w
+          w,h = stm.Width, stm.Height
+          bpc = stm.BitsPerComponent || 8
+          bpr = (bpc >> 3) * 3 * w
           data = stm.data
 
           begin
             imgview = ImgViewer.new
-            if stm[:Filter] == :DCTDecode or (stm[:Filter].is_a?(Array) and stm[:Filter][0] == :DCTDecode)
+            if stm.Filter == :DCTDecode or (stm.Filter.is_a?(Array) and stm.Filter[0] == :DCTDecode)
               imgview.show_compressed_img data
             else
               imgview.show_raw_img data, w, h, bpp, bpr

data/bin/gui/textview.rb

@@ -33,9 +33,7 @@ module PDFWalker
     private
     
     def create_objectview
-      
       @objectview = ObjectView.new(self)
-      
     end
     
     class ObjectView < Notebook
@@ -56,14 +54,18 @@ module PDFWalker
         @pdfpanel.add_with_viewport @pdfview
         append_page(@pdfpanel, Label.new("PDF Code"))
         
-        @pdfbuffer.create_tag("Object", { :weight => Pango::WEIGHT_BOLD, :foreground => "darkblue", :family => "Courier", :scale => Pango::AttrScale::LARGE })
+        @pdfbuffer.create_tag("Object", 
+          :weight => Pango::WEIGHT_BOLD, 
+          :foreground => "darkblue", 
+          :family => "Courier", 
+          :scale => Pango::AttrScale::LARGE
+        )
         
       end
       
       def load(object)
       
         begin
-        
           self.clear
         
           pdftag = "Object"

data/bin/gui/treeview.rb

@@ -59,7 +59,7 @@ module PDFWalker
     FGCOL = 4
     BGCOL = 5
     
-    @@appearance = Hash.new({ :Color => "black", :Weight => Pango::WEIGHT_NORMAL, :Style => Pango::STYLE_NORMAL })
+    @@appearance = Hash.new(:Weight => Pango::WEIGHT_NORMAL, :Style => Pango::STYLE_NORMAL)
     
     attr_reader :parent
     
@@ -368,7 +368,7 @@ module PDFWalker
     
     def reset_appearance
       
-      @@appearance[:Filename] = {:Color => "black", :Weight => Pango::WEIGHT_BOLD, :Style => Pango::STYLE_NORMAL}
+      @@appearance[:Filename] = {:Weight => Pango::WEIGHT_BOLD, :Style => Pango::STYLE_NORMAL}
       @@appearance[:Header] = {:Color => "darkgreen", :Weight => Pango::WEIGHT_BOLD, :Style => Pango::STYLE_NORMAL}
       @@appearance[:Revision] = {:Color => "blue", :Weight => Pango::WEIGHT_BOLD, :Style => Pango::STYLE_NORMAL}
       @@appearance[:Body] = {:Color => "purple", :Weight => Pango::WEIGHT_BOLD, :Style => Pango::STYLE_NORMAL}
@@ -376,7 +376,7 @@ module PDFWalker
       @@appearance[:XRefSubSection] = {:Color => "brown", :Weight => Pango::WEIGHT_BOLD, :Style => Pango::STYLE_NORMAL}
       @@appearance[:XRef] = {:Color => "gray20", :Weight => Pango::WEIGHT_BOLD, :Style => Pango::STYLE_NORMAL}
       @@appearance[:Trailer] = {:Color => "purple", :Weight => Pango::WEIGHT_BOLD, :Style => Pango::STYLE_NORMAL}
-      @@appearance[:StartXref] = {:Color => "black", :Weight => Pango::WEIGHT_BOLD, :Style => Pango::STYLE_NORMAL}
+      @@appearance[:StartXref] = {:Weight => Pango::WEIGHT_BOLD, :Style => Pango::STYLE_NORMAL}
       @@appearance[:String] = {:Color => "red", :Weight => Pango::WEIGHT_NORMAL, :Style => Pango::STYLE_ITALIC}
       @@appearance[:Name] = {:Color => "gray", :Weight => Pango::WEIGHT_NORMAL, :Style => Pango::STYLE_ITALIC}
       @@appearance[:Number] = {:Color => "orange", :Weight => Pango::WEIGHT_NORMAL, :Style => Pango::STYLE_NORMAL}
@@ -384,7 +384,7 @@ module PDFWalker
       @@appearance[:Stream] = {:Color => "darkcyan", :Weight => Pango::WEIGHT_BOLD, :Style => Pango::STYLE_NORMAL}
       @@appearance[:StreamData] = {:Color => "darkcyan", :Weight => Pango::WEIGHT_NORMAL, :Style => Pango::STYLE_OBLIQUE}
       @@appearance[:Array] = {:Color => "darkgreen", :Weight => Pango::WEIGHT_BOLD, :Style => Pango::STYLE_NORMAL}
-      @@appearance[:Reference] = {:Color => "black", :Weight => Pango::WEIGHT_NORMAL, :Style => Pango::STYLE_OBLIQUE}
+      @@appearance[:Reference] = {:Weight => Pango::WEIGHT_NORMAL, :Style => Pango::STYLE_OBLIQUE}
       @@appearance[:Boolean] = {:Color => "deeppink", :Weight => Pango::WEIGHT_NORMAL, :Style => Pango::STYLE_NORMAL}
       
     end

data/bin/gui/walker.rb

@@ -35,7 +35,7 @@ include Gtk
 begin
   require 'origami'
 rescue LoadError
-  ORIGAMIDIR = "#{File.dirname(__FILE__)}/../.."
+  ORIGAMIDIR = "#{File.dirname(__FILE__)}/../../lib"
   $: << ORIGAMIDIR
   require 'origami'
 end

data/bin/pdf2graph

@@ -27,7 +27,7 @@
 begin
   require 'origami'
 rescue LoadError
-  ORIGAMIDIR = "#{File.dirname(__FILE__)}/.."
+  ORIGAMIDIR = "#{File.dirname(__FILE__)}/../lib"
   $: << ORIGAMIDIR
   require 'origami'
 end

data/bin/pdf2pdfa

@@ -28,7 +28,7 @@
 begin
   require 'origami'
 rescue LoadError
-  ORIGAMIDIR = "#{File.dirname(__FILE__)}/.."
+  ORIGAMIDIR = "#{File.dirname(__FILE__)}/../lib"
   $: << ORIGAMIDIR
   require 'origami'
 end

data/bin/pdf2ruby

@@ -30,7 +30,7 @@ require 'ftools'
 begin
   require 'origami'
 rescue LoadError
-  ORIGAMIDIR = "#{File.dirname(__FILE__)}/.."
+  ORIGAMIDIR = "#{File.dirname(__FILE__)}/../lib"
   $: << ORIGAMIDIR
   require 'origami'
 end

data/bin/pdfcocoon

@@ -27,7 +27,7 @@
 begin
   require 'origami'
 rescue LoadError
-  ORIGAMIDIR = "#{File.dirname(__FILE__)}/.."
+  ORIGAMIDIR = "#{File.dirname(__FILE__)}/../lib"
   $: << ORIGAMIDIR
   require 'origami'
 end

data/bin/pdfcop

@@ -30,7 +30,7 @@
 begin
   require 'origami'
 rescue LoadError
-  ORIGAMIDIR = "#{File.dirname(__FILE__)}/.."
+  ORIGAMIDIR = "#{File.dirname(__FILE__)}/../lib"
   $: << ORIGAMIDIR
   require 'origami'
 end

data/bin/pdfdecompress

@@ -27,7 +27,7 @@
 begin
   require 'origami'
 rescue LoadError
-  ORIGAMIDIR = "#{File.dirname(__FILE__)}/.."
+  ORIGAMIDIR = "#{File.dirname(__FILE__)}/../lib"
   $: << ORIGAMIDIR
   require 'origami'
 end

data/bin/pdfdecrypt

@@ -27,7 +27,7 @@
 begin
   require 'origami'
 rescue LoadError
-  ORIGAMIDIR = "#{File.dirname(__FILE__)}/.."
+  ORIGAMIDIR = "#{File.dirname(__FILE__)}/../lib"
   $: << ORIGAMIDIR
   require 'origami'
 end

data/bin/pdfencrypt

@@ -27,7 +27,7 @@
 begin
   require 'origami'
 rescue LoadError
-  ORIGAMIDIR = "#{File.dirname(__FILE__)}/.."
+  ORIGAMIDIR = "#{File.dirname(__FILE__)}/../lib"
   $: << ORIGAMIDIR
   require 'origami'
 end

data/bin/pdfextract

@@ -30,7 +30,7 @@
 begin
   require 'origami'
 rescue LoadError
-  ORIGAMIDIR = "#{File.dirname(__FILE__)}/.."
+  ORIGAMIDIR = "#{File.dirname(__FILE__)}/../lib"
   $: << ORIGAMIDIR
   require 'origami'
 end
@@ -42,7 +42,7 @@ require 'rexml/document'
 class OptParser
   BANNER = <<USAGE
 Usage: #{$0} <PDF-file> [-afjms] [-d <output-directory>]
-Extracts various data out of a document (streams, scripts, fonts, metadata, attachments).
+Extracts various data out of a document (streams, scripts, images, fonts, metadata, attachments).
 Bug reports or feature requests at: http://origami-pdf.googlecode.com/
 
 Options:
@@ -76,6 +76,10 @@ USAGE
         options[:metadata] = true
       end
 
+      opts.on("-i", "--images", "Extracts embedded images") do
+        options[:images] = true
+      end
+
       opts.on_tail("-h", "--help", "Show this message") do
         puts opts
         exit
@@ -104,11 +108,12 @@ begin
     target = ARGV.shift
   end
 
-  unless [:streams,:javascript,:attachments,:fonts,:metadata].any? {|opt| @options[opt]} 
+  unless [:streams,:javascript,:attachments,:fonts,:metadata,:images].any? {|opt| @options[opt]} 
     @options[:streams] =
     @options[:javascript] =
     @options[:fonts] =
-    @options[:attachments] = true
+    @options[:attachments] = 
+    @options[:images] = true
   end
 
   if @options[:output_dir].nil?
@@ -125,17 +130,26 @@ begin
   pdf = PDF.read(target, params)
 
   if @options[:streams]
+    nstreams = 0
+    Dir::mkdir("#{OUTPUT_DIR}/streams") unless File.directory?("#{OUTPUT_DIR}/streams")
+    
     pdf.root_objects.find_all{|obj| obj.is_a?(Stream)}.each do |stream|
-      stream_file = "#{OUTPUT_DIR}/stream_#{stream.reference.refno}.dmp" 
+      stream_file = "#{OUTPUT_DIR}/streams/stream_#{stream.reference.refno}.dmp"
       File.open(stream_file, "w") do |fd|
         fd.write(stream.data)
       end
+      nstreams += 1
     end
+
+    puts "Extracted #{nstreams} PDF streams to '#{OUTPUT_DIR}/streams'."
   end
 
   if @options[:javascript]
+    nscripts = 0
+    Dir::mkdir("#{OUTPUT_DIR}/scripts") unless File.directory?("#{OUTPUT_DIR}/scripts")
+
     pdf.ls(/^JS$/).each do |script|
-      script_file = "#{OUTPUT_DIR}/script_#{script.hash}.js"
+      script_file = "#{OUTPUT_DIR}/scripts/script_#{script.hash}.js"
       File.open(script_file, "w") do |fd|
         fd.write(
           case script
@@ -146,6 +160,7 @@ begin
           end
         )
       end
+      nscripts += 1
     end
 
     # Also checking for presence of JavaScript in XML forms.
@@ -175,42 +190,88 @@ begin
         File.open(script_file, 'w') do |fd|
           fd.write(script.text)
         end
+        nscripts += 1
       end
     end
+
+    puts "Extracted #{nscripts} scripts to '#{OUTPUT_DIR}/scripts'."
   end
 
   if @options[:attachments]
+    nattach = 0
+    Dir::mkdir("#{OUTPUT_DIR}/attachments") unless File.directory?("#{OUTPUT_DIR}/attachments")
+    
     pdf.ls_names(Names::Root::EMBEDDEDFILES).each do |name, attachment|
-      attached_file = "#{OUTPUT_DIR}/attached_#{File.basename(name)}"
+      attached_file = "#{OUTPUT_DIR}/attachments/attached_#{File.basename(name)}"
       spec = attachment.solve
-      ef = spec[:EF].solve
-      f = ef[:F].solve
-
-      File.open(attached_file, "w") do |fd|
-        fd.write(f.data)
+      if spec and spec.EF and f = spec.EF.F and f.is_a?(Stream)
+        File.open(attached_file, "w") do |fd|
+          fd.write(f.data)
+        end
+        nattach += 1
       end
     end
+    
+    puts "Extracted #{nattach} attachments to '#{OUTPUT_DIR}/attachments'."
   end
 
   if @options[:fonts]
+    nfonts = 0
+    Dir::mkdir("#{OUTPUT_DIR}/fonts") unless File.directory?("#{OUTPUT_DIR}/fonts")
+    
     pdf.root_objects.find_all{|obj| obj.is_a?(Stream)}.each do |stream|
       font = stream.xrefs.find{|obj| obj.is_a?(FontDescriptor)}
       if font
-        font_file = "#{OUTPUT_DIR}/font_#{File.basename(font.FontName.value.to_s)}"
+        font_file = "#{OUTPUT_DIR}/fonts/font_#{File.basename(font.FontName.value.to_s)}"
         File.open(font_file, "w") do |fd|
           fd.write(stream.data)
         end
+        nfonts += 1
       end
     end
+
+    puts "Extracted #{nfonts} fonts to '#{OUTPUT_DIR}/fonts'."
   end
 
   if @options[:metadata]
+    nmeta = 0
+    Dir::mkdir("#{OUTPUT_DIR}/metadata") unless File.directory?("#{OUTPUT_DIR}/metadata")
+
     pdf.root_objects.find_all{|obj| obj.is_a?(MetadataStream)}.each do |stream|
-      metadata_file = "#{OUTPUT_DIR}/metadata_#{stream.reference.refno}.xml"
+      metadata_file = "#{OUTPUT_DIR}/metadata/metadata_#{stream.reference.refno}.xml"
       File.open(metadata_file, "w") do |fd|
         fd.write(stream.data)
       end
+      nmeta += 1
+    end
+
+    puts "Extracted #{nmeta} metadata streams to '#{OUTPUT_DIR}/metadata'."
+  end
+
+  if @options[:images]
+    nimages = 0
+    Dir::mkdir("#{OUTPUT_DIR}/images") unless File.directory?("#{OUTPUT_DIR}/images")
+    
+    pdf.root_objects.find_all{|obj| obj.is_a?(Graphics::ImageXObject)}.each do |stream|
+      begin
+        ext, image_data = stream.to_image_file
+        image_file = "#{OUTPUT_DIR}/images/image_#{stream.reference.refno}.#{ext}"
+
+        if ext != 'png' and stream.ColorSpace == Graphics::Color::Space::DEVICE_CMYK
+          STDERR.puts "Warning: file '#{image_file}' is intended to be viewed in CMYK color space."
+        end
+
+        File.open(image_file, "w") do |fd|
+          fd.write(image_data)
+        end
+        nimages += 1
+
+      rescue Exception => e
+        STDERR.puts "Unable to decode image (stream #{stream.reference.refno}). #{e.message}"
+      end
     end
+    
+    puts "Extracted #{nimages} images to '#{OUTPUT_DIR}/images'."
   end
 
 rescue SystemExit

data/bin/pdfmetadata

@@ -27,7 +27,7 @@
 begin
   require 'origami'
 rescue LoadError
-  ORIGAMIDIR = "#{File.dirname(__FILE__)}/.."
+  ORIGAMIDIR = "#{File.dirname(__FILE__)}/../lib"
   $: << ORIGAMIDIR
   require 'origami'
 end

data/bin/shell/.irbrc

@@ -1,7 +1,7 @@
 begin
   require 'origami'
 rescue LoadError
-  ORIGAMIDIR = "#{File.dirname(__FILE__)}/../.."
+  ORIGAMIDIR = "#{File.dirname(__FILE__)}/../../lib"
   $: << ORIGAMIDIR
   require 'origami'
 end

data/{origami.rb → lib/origami.rb}

@@ -30,7 +30,7 @@ if RUBY_VERSION < '1.9'
   end
 end
 
-require 'origami/parsers/pdf'
-require 'origami/parsers/fdf'
-require 'origami/parsers/ppklite'
+require 'origami/pdf'
+require 'origami/extensions/fdf'
+require 'origami/extensions/ppklite'
 

data/{origami → lib/origami}/acroform.rb

@@ -145,7 +145,7 @@ module Origami
       RIGHT = 2
     end
       
-    def self.included(receiver)
+    def self.included(receiver) #:nodoc:
 
       receiver.field   :FT,     :Type => Name, :Default => Type::TEXT, :Required => true
       receiver.field   :Parent, :Type => Dictionary
@@ -166,7 +166,7 @@ module Origami
         
     end
    
-    def pre_build
+    def pre_build #:nodoc:
       
       if not self.T
         self.T = "undef#{::Array.new(5) {(0x30 + rand(10)).chr}.join}"