origami 1.2.3 → 1.2.4

data/bin/pdfextract

@@ -135,7 +135,7 @@ begin
     
     pdf.root_objects.find_all{|obj| obj.is_a?(Stream)}.each do |stream|
       stream_file = "#{OUTPUT_DIR}/streams/stream_#{stream.reference.refno}.dmp"
-      File.open(stream_file, "w") do |fd|
+      File.open(stream_file, "wb") do |fd|
         fd.write(stream.data)
       end
       nstreams += 1
@@ -150,7 +150,7 @@ begin
 
     pdf.ls(/^JS$/).each do |script|
       script_file = "#{OUTPUT_DIR}/scripts/script_#{script.hash}.js"
-      File.open(script_file, "w") do |fd|
+      File.open(script_file, "wb") do |fd|
         fd.write(
           case script
             when Stream then
@@ -187,7 +187,7 @@ begin
       xfadoc = REXML::Document.new(xml)
       REXML::XPath.match(xfadoc, "//script").each do |script|
         script_file = "#{OUTPUT_DIR}/script_#{script.hash}.js"
-        File.open(script_file, 'w') do |fd|
+        File.open(script_file, "wb") do |fd|
           fd.write(script.text)
         end
         nscripts += 1
@@ -205,7 +205,7 @@ begin
       attached_file = "#{OUTPUT_DIR}/attachments/attached_#{File.basename(name)}"
       spec = attachment.solve
       if spec and spec.EF and f = spec.EF.F and f.is_a?(Stream)
-        File.open(attached_file, "w") do |fd|
+        File.open(attached_file, "wb") do |fd|
           fd.write(f.data)
         end
         nattach += 1
@@ -223,7 +223,7 @@ begin
       font = stream.xrefs.find{|obj| obj.is_a?(FontDescriptor)}
       if font
         font_file = "#{OUTPUT_DIR}/fonts/font_#{File.basename(font.FontName.value.to_s)}"
-        File.open(font_file, "w") do |fd|
+        File.open(font_file, "wb") do |fd|
           fd.write(stream.data)
         end
         nfonts += 1
@@ -239,7 +239,7 @@ begin
 
     pdf.root_objects.find_all{|obj| obj.is_a?(MetadataStream)}.each do |stream|
       metadata_file = "#{OUTPUT_DIR}/metadata/metadata_#{stream.reference.refno}.xml"
-      File.open(metadata_file, "w") do |fd|
+      File.open(metadata_file, "wb") do |fd|
         fd.write(stream.data)
       end
       nmeta += 1
@@ -261,7 +261,7 @@ begin
           STDERR.puts "Warning: file '#{image_file}' is intended to be viewed in CMYK color space."
         end
 
-        File.open(image_file, "w") do |fd|
+        File.open(image_file, "wb") do |fd|
           fd.write(image_data)
         end
         nimages += 1

data/bin/pdfwalker

@@ -1,6 +1,6 @@
 #!/usr/bin/env ruby
 
-$:.unshift "#{File.dirname(__FILE__)}" if RUBY_VERSION >= '1.9'
+$:.unshift "#{File.dirname(__FILE__)}"
 require 'gui/walker'
 
 PDFWalker::Walker.start(ARGV[0])

data/bin/shell/console.rb

@@ -41,7 +41,7 @@ module Origami
     end
   end
 
-  unless (RUBY_VERSION < '1.9' and RUBY_PLATFORM =~ /win32/)
+  unless RUBY_VERSION < '1.9'
     require 'tempfile'
 
     class Stream
@@ -50,7 +50,7 @@ module Origami
         tmpfile.write(self.data)
         tmpfile.close
 
-        Process.wait Kernel.spawn "#{editor or 'vim'} #{tmpfile.path}"
+        Process.wait Kernel.spawn "#{editor} #{tmpfile.path}"
         
         self.data = File.read(tmpfile.path)
         tmpfile.unlink

data/lib/origami/encryption.rb

@@ -25,7 +25,6 @@
 
 require 'digest/md5'
 require 'digest/sha2'
-require 'iconv'
 
 module Origami
 
@@ -115,16 +114,15 @@ module Origami
         doc_id = doc_id.first
       end
 
-      if handler.is_owner_password?(passwd, doc_id)
+      if handler.is_user_password?(passwd, doc_id)
+        encryption_key = handler.compute_user_encryption_key(passwd, doc_id)
+      elsif handler.is_owner_password?(passwd, doc_id)
         if handler.V.to_i < 5
           user_passwd = handler.retrieve_user_password(passwd)
           encryption_key = handler.compute_user_encryption_key(user_passwd, doc_id)
         else
           encryption_key = handler.compute_owner_encryption_key(passwd)
         end
-      
-      elsif handler.is_user_password?(passwd, doc_id)
-        encryption_key = handler.compute_user_encryption_key(passwd, doc_id)
       else
         raise EncryptionInvalidPasswordError
       end
@@ -1325,10 +1323,10 @@ module Origami
             end
 
             64.times do |j|
-              x = aes.update(block)
-              unless vector.empty?
-                x += aes.update(vector)
-              end
+              x = ''
+              x += aes.update(password) unless password.empty?
+              x += aes.update(block)
+              x += aes.update(vector) unless vector.empty?
 
               if j == 0
                 block_size = 32 + (x.unpack("C16").inject(0) {|a,b| a+b} % 3) * 16

data/lib/origami/filters.rb

@@ -30,6 +30,16 @@ module Origami
   #
   module Filter
 
+    class InvalidFilterDataError < Exception # :nodoc:
+      attr_reader :decoded_data
+
+      def initialize(message, decoded_data = nil)
+        super(message)
+
+        @decoded_data = decoded_data
+      end
+    end
+
     module Utils
       
       class BitWriterError < Exception #:nodoc:

data/lib/origami/filters/ascii.rb

@@ -27,7 +27,7 @@ module Origami
 
   module Filter
     
-    class InvalidASCIIHexStringError < Exception #:nodoc:
+    class InvalidASCIIHexStringError < InvalidFilterDataError #:nodoc:
     end
     
     #
@@ -64,7 +64,7 @@ module Origami
       
     end
     
-    class InvalidASCII85StringError < Exception #:nodoc:
+    class InvalidASCII85StringError < InvalidFilterDataError #:nodoc:
     end
     
     #
@@ -141,7 +141,7 @@ module Origami
             codelen = 5
             
             if input.length - i < 5
-              raise InvalidASCII85StringError, "Invalid length" if input.length - i == 1
+              raise InvalidASCII85StringError.new("Invalid length", result) if input.length - i == 1
               
               addend = 5 - (input.length - i)
               input << "u" * addend
@@ -152,15 +152,20 @@ module Origami
             # Checking if this string is in base85
             5.times do |j|
               if input[i+j].ord > "u"[0].ord or input[i+j].ord < "!"[0].ord
-                raise InvalidASCII85StringError, "Invalid character sequence: #{input[i,5].inspect}"
+                raise InvalidASCII85StringError.new(
+                  "Invalid character sequence: #{input[i,5].inspect}",
+                  result
+                )
               else
                 inblock += (input[i+j].ord - "!"[0].ord) * 85 ** (4 - j)
               end
             end
           
-            if inblock >= 2**32
-              raise InvalidASCII85StringError, "Invalid value (#{inblock}) for block #{input[i,5].inspect}"
-            end
+            
+            raise InvalidASCII85StringError.new(
+              "Invalid value (#{inblock}) for block #{input[i,5].inspect}",
+              result
+            ) if inblock >= 2**32
           
           end
         

data/lib/origami/filters/ccitt.rb

@@ -27,7 +27,7 @@ module Origami
 
   module Filter
 
-    class InvalidCCITTFaxDataError < Exception #:nodoc:
+    class InvalidCCITTFaxDataError < InvalidFilterDataError #:nodoc:
     end
 
     class CCITTFaxFilterError < Exception #:nodoc:
@@ -388,8 +388,8 @@ module Origami
         bitr = Utils::BitReader.new(stream)
         bitw = Utils::BitWriter.new
 
-        current_color = white
         until bitr.eod? or rows == 0
+          current_color = white
           
           # realign the read line on a 8-bit boundary if required
           if aligned and bitr.pos % 8 != 0
@@ -404,7 +404,10 @@ module Origami
 
           # checking for the presence of EOL
           if bitr.peek(EOL[1]) != EOL[0]
-            raise CCITTFaxFilterError, "No end-of-line pattern found (at bit pos #{bitr.pos}/#{bitr.size}})" if has_eol
+            raise InvalidCCITTFaxDataError.new(
+              "No end-of-line pattern found (at bit pos #{bitr.pos}/#{bitr.size}})",
+              bitw.final.to_s
+            )if has_eol
           else
             bitr.pos += EOL[1]
           end
@@ -417,10 +420,16 @@ module Origami
               bit_length = get_black_bits(bitr)
             end
 
-            raise CCITTFaxFilterError, "Unfinished line (at bit pos #{bitr.pos}/#{bitr.size}})" if bit_length.nil?
+            raise InvalidCCITTFaxDataError.new(
+              "Unfinished line (at bit pos #{bitr.pos}/#{bitr.size}})",
+              bitw.final.to_s
+            ) if bit_length.nil?
             
             line_length += bit_length
-            raise CCITTFaxFilterError, "Line is too long (at bit pos #{bitr.pos}/#{bitr.size}})" if line_length > columns
+            raise InvalidCCITTFaxDataError.new(
+              "Line is too long (at bit pos #{bitr.pos}/#{bitr.size}})",
+              bitw.final.to_s
+            ) if line_length > columns
 
             write_bit_range(bitw, current_color, bit_length)
             current_color ^= 1

data/lib/origami/filters/flate.rb

@@ -30,16 +30,7 @@ module Origami
   
   module Filter
 
-    class InvalidFlateDataError < Exception #:nodoc:
-      attr_reader :zlib_stream, :zlib_except
-
-      def initialize(zlib_stream, zlib_except)
-        super(zlib_except.message)
-
-        @zlib_stream = zlib_stream
-        @zlib_except = zlib_except
-      end
-    end
+    class InvalidFlateDataError < InvalidFilterDataError; end #:nodoc:
     
     #
     # Class representing a Filter used to encode and decode data with zlib/Flate compression algorithm.
@@ -92,7 +83,7 @@ module Origami
         begin
           uncompressed = zlib_stream.inflate(stream)
         rescue Zlib::DataError => zlib_except
-          raise InvalidFlateDataError.new(zlib_stream, zlib_except)
+          raise InvalidFlateDataError.new(zlib_except.message, zlib_stream.flush_next_out)
         end
 
         if @params.Predictor.is_a?(Integer)

data/lib/origami/filters/lzw.rb

@@ -29,7 +29,7 @@ module Origami
 
   module Filter
 
-    class InvalidLZWDataError < Exception #:nodoc:
+    class InvalidLZWDataError < InvalidFilterDataError #:nodoc:
     end
     
     #
@@ -131,8 +131,10 @@ module Origami
             when 4095
               if byte != CLEARTABLE
               then
-                raise InvalidLZWDataError, 
-                  "LZW table is full and no clear flag was set (codeword #{byte.to_s(2).rjust(codesize,'0')} at bit #{bstring.pos - codesize}/#{bstring.size})"
+                raise InvalidLZWDataError.new(
+                  "LZW table is full and no clear flag was set (codeword #{byte.to_s(2).rjust(codesize,'0')} at bit #{bstring.pos - codesize}/#{bstring.size})",
+                  result
+                )
               end
           end
 
@@ -150,6 +152,11 @@ module Origami
               result << table.key(byte)
               redo
             else
+              raise InvalidLZWDataError.new(
+                "No entry for codeword #{prevbyte.to_s(2).rjust(codesize,'0')}.",
+                result
+              ) unless table.key(prevbyte)
+
               if table.has_value?(byte)
                 entry = table.key(byte)
               else

data/lib/origami/filters/runlength.rb

@@ -27,7 +27,7 @@ module Origami
   
   module Filter
     
-    class InvalidRunLengthDataError < Exception #:nodoc:
+    class InvalidRunLengthDataError < InvalidFilterDataError #:nodoc:
     end
     
     #

data/lib/origami/object.rb

@@ -490,13 +490,15 @@ module Origami
       end
 
       def skip_until_next_obj(stream) #:nodoc:
-        stream.pos += 1 until stream.match?(@@regexp_obj) or 
-          stream.match?(/xref/) or stream.match?(/trailer/) or stream.match?(/startxref/) or
-          stream.eos?
+        [ @@regexp_obj, /xref/, /trailer/, /startxref/ ].each do |re|
+          if stream.scan_until(re)
+            stream.pos -= stream.matched_size
+            return true
+          end
+        end
         
-        not stream.eos?
+        false
       end
-      
     end
     
     def pdf_version_required #:nodoc:

data/lib/origami/parsers/pdf/linear.rb

@@ -45,15 +45,15 @@ module Origami
           begin
             pdf.add_new_revision unless revision.zero?
             revision = revision + 1
-            
+
             info "...Parsing revision #{pdf.revisions.size}..."
             loop do
               break if (object = parse_object).nil?
               pdf.insert(object)
             end
-            
+
             pdf.revisions.last.xreftable = parse_xreftable
-           
+
             trailer = parse_trailer
             pdf.revisions.last.trailer = trailer
 
@@ -61,7 +61,7 @@ module Origami
               (pdf.get_object_by_offset(trailer.XRefStm) if trailer.has_field? :XRefStm)
 
             if not xrefstm.nil?
-              debug "Found a XRefStream for this revision at #{xrefstm.reference}"
+              warn "Found a XRefStream for this revision at #{xrefstm.reference}"
               pdf.revisions.last.xrefstm = xrefstm
             end
 
@@ -70,10 +70,10 @@ module Origami
           rescue Exception => e
             error "Cannot read : " + (@data.peek(10) + "...").inspect
             error "Stopped on exception : " + e.message
-            
+
             break
           end
-          
+
         end
 
         parse_finalize(pdf)

data/lib/origami/pdf.rb

@@ -61,8 +61,8 @@ require 'origami/parsers/pdf'
 
 module Origami
 
-  VERSION   = "1.2.3"
-  REVISION  = "$Revision: rev 143/, 2011/10/20 16:22:40 $" #:nodoc:
+  VERSION   = "1.2.4"
+  REVISION  = "$Revision$" #:nodoc:
   
   #
   # Global options for Origami.

data/lib/origami/stream.rb

@@ -249,11 +249,19 @@ module Origami
               filters = [ filters ] unless filters.is_a?(::Array)
           
               @data = @rawdata.dup
+              @data.freeze
+
               filters.length.times do |layer|
                 params = dparams[layer].is_a?(Dictionary) ? dparams[layer] : {}
                 filter = filters[layer]
 
-                @data = decode_data(@data, filter, params)
+                begin
+                  @data = decode_data(@data, filter, params)
+                rescue Filter::InvalidFilterDataError => e
+                  @data = e.decoded_data if e.decoded_data
+                  raise InvalidStreamObjectError, 
+                    "Error while decoding stream #{self.reference}\n\t-> [#{e.class}] #{e.message}"
+                end
               end
             else
               raise InvalidStreamObjectError, "Invalid Filter type parameter"
@@ -357,15 +365,8 @@ module Origami
         raise InvalidStreamObjectError, "Unknown filter : #{filter}"
       end
 
-      begin
-        Origami::Filter.const_get(filter.value.to_s.sub(/Decode$/,"")).decode(data, params)
-      
-      rescue Filter::InvalidFlateDataError => flate_e
-        return flate_e.zlib_stream.flush_next_out   
+      Origami::Filter.const_get(filter.value.to_s.sub(/Decode$/,"")).decode(data, params)
       
-      rescue Exception => e
-        raise InvalidStreamObjectError, "Error while decoding stream #{self.reference}\n\t-> [#{e.class}] #{e.message}"
-      end
     end
     
     def encode_data(data, filter, params) #:nodoc:

data/lib/origami/string.rb

@@ -125,12 +125,16 @@ module Origami
     # Convert String object to an UTF8 encoded Ruby string.
     #
     def to_utf8
-      require 'iconv'
-
       infer_encoding
-      i = Iconv.new("UTF-8", "UTF-16")
-        utf8str = i.iconv(self.encoding.to_utf16be(self.value))
-      i.close
+
+      if RUBY_VERSION < '1.9'
+        require 'iconv'
+        i = Iconv.new("UTF-8", "UTF-16")
+          utf8str = i.iconv(self.encoding.to_utf16be(self.value))
+        i.close
+      else
+        utf8str = self.encoding.to_utf16be(self.value).encode("utf-8", "utf-16")
+      end
 
       utf8str
     end