stream.rb: get the partially decoded data when an error occurs during…

… decoding

Rakefile.rb

@@ -7,7 +7,7 @@
 
 spec = Gem::Specification.new do |s|
   s.name       = "origami"
-  s.version    = "1.2.2"
+  s.version    = "1.2.4"
   s.author     = "Guillaume Delugré"
   s.email      = "guillaume at security-labs dot org"
   s.homepage   = "http://aslr.fr/pages/Origami"

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:

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 @@ def decode(string)
 
     end
 
-    class InvalidASCII85StringError < Exception #:nodoc:
+    class InvalidASCII85StringError < InvalidFilterDataError #:nodoc:
     end
 
     #
@@ -141,7 +141,7 @@ def decode(string)
             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 @@ def decode(string)
             # 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
 

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:
@@ -404,7 +404,10 @@ def decode(stream)
 
           # 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 @@ def decode(stream)
               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

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 @@ def decode(stream)
         begin
           uncompressed = zlib_stream.inflate(stream)
         rescue Zlib::DataError => zlib_except
-          raise InvalidFlateDataError.new(zlib_stream, zlib_except)
+          raise InvalidFlateDataError.new(zlibexcept.message, zlib_stream.flush_next_out)
         end
 
         if @params.Predictor.is_a?(Integer)

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 @@ def decode(string)
             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 @@ def decode(string)
               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

lib/origami/filters/runlength.rb

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

lib/origami/pdf.rb

@@ -61,7 +61,7 @@
 
 module Origami
 
-  VERSION   = "1.2.3"
+  VERSION   = "1.2.4"
   REVISION  = "$Revision$" #:nodoc:
 
   #

lib/origami/stream.rb

@@ -249,11 +249,19 @@ def decode!
               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 @@ def decode_data(data, filter, params) #:nodoc:
         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: