origami 1.2.5 → 1.2.6

data/lib/origami/filters/ccitt.rb

@@ -300,13 +300,14 @@ module Origami
       # Encodes data using CCITT-facsimile compression method.
       #
       def encode(stream)
-      
-        if @params.has_key?(:K) and @params.K != 0
+        mode = @params.has_key?(:K) ? @params.K.value : 0 
+
+        unless mode.is_a?(::Integer) and mode <= 0
           raise NotImplementedError, "CCITT encoding scheme not supported"
         end
 
         columns = @params.has_key?(:Columns) ? @params.Columns.value : (stream.size << 3)
-        unless columns.is_a?(::Integer) and columns > 0 and columns % 8 == 0
+        unless columns.is_a?(::Integer) and columns > 0 #and columns % 8 == 0
           raise CCITTFaxFilterError, "Invalid value for parameter `Columns'"
         end
 
@@ -314,42 +315,24 @@ module Origami
           raise CCITTFaxFilterError, "Data size is not a multiple of image width"
         end
 
-        white, black = (@params.BlackIs1 == true) ? [0,1] : [1,0]
-
+        colors = (@params.BlackIs1 == true) ? [0,1] : [1,0]
+        white, _black = colors
         bitr = Utils::BitReader.new(stream)
         bitw = Utils::BitWriter.new
 
-        until bitr.eod?
-          
-          bitw.write(*EOL)
-          scan_len = 0
-          current_color = white
-
-          # Process each bit in line.
-          begin
-            if bitr.read(1) == current_color
-              scan_len += 1
-            else
-              if current_color == white
-                put_white_bits(bitw, scan_len)
-              else
-                put_black_bits(bitw, scan_len)
-              end
-
-              current_color ^= 1
-              scan_len = 1
-            end
-          end while bitr.pos % columns != 0
-
-          if current_color == white
-            put_white_bits(bitw, scan_len)
-          else
-            put_black_bits(bitw, scan_len)
-          end
+        # Group 4 requires an imaginary white line
+        if mode < 0
+          prev_line = Utils::BitWriter.new
+          write_bit_range(prev_line, white, columns)
+          prev_line = Utils::BitReader.new(prev_line.final.to_s)
+        end
 
-          # Align encoded lign on a 8-bit boundary.
-          if @params.EncodedByteAlign == true and bitw.pos % 8 != 0
-            bitw.write(0, 8 - (bitw.pos % 8))
+        until bitr.eod?
+          case
+            when mode == 0
+              encode_one_dimensional_line(bitr, bitw, columns, colors)  
+            when mode < 0
+              encode_two_dimensional_line(bitr, bitw, columns, colors, prev_line)
           end
         end
 
@@ -363,21 +346,27 @@ module Origami
       # Decodes data using CCITT-facsimile compression method.
       #
       def decode(stream)
-        if @params.has_key?(:K) and @params.K != 0
+        mode = @params.has_key?(:K) ? @params.K.value : 0 
+
+        unless mode.is_a?(::Integer) and mode <= 0
           raise NotImplementedError, "CCITT encoding scheme not supported"
         end
 
         columns = @params.has_key?(:Columns) ? @params.Columns.value : 1728
-        unless columns.is_a?(::Integer) and columns > 0 and columns % 8 == 0
+        unless columns.is_a?(::Integer) and columns > 0 #and columns % 8 == 0
           raise CCITTFaxFilterError, "Invalid value for parameter `Columns'"
         end
 
-        white, black = (@params.BlackIs1 == true) ? [0,1] : [1,0]
-        aligned = @params.EncodedByteAlign == true
-        has_eob = @params.EndOfBlock.nil? or @params.EndOfBlock == true
-        has_eol = @params.EndOfLine == true
+        colors = (@params.BlackIs1 == true) ? [0,1] : [1,0]
+        white, _black = colors
+        params = 
+        {
+          :is_aligned? => (@params.EncodedByteAlign == true),
+          :has_eob? => (@params.EndOfBlock.nil? or @params.EndOfBlock == true),
+          :has_eol? => (@params.EndOfLine == true)
+        }
 
-        unless has_eob
+        unless params[:has_eob?]
           unless @params.has_key?(:Rows) and @params.Rows.is_a?(::Integer) and @params.Rows.value > 0
             raise CCITTFaxFilterError, "Invalid value for parameter `Rows'"
           end
@@ -388,16 +377,21 @@ module Origami
         bitr = Utils::BitReader.new(stream)
         bitw = Utils::BitWriter.new
 
+        # Group 4 requires an imaginary white line
+        if mode < 0
+          prev_line = Utils::BitWriter.new
+          write_bit_range(prev_line, white, columns)
+          prev_line = Utils::BitReader.new(prev_line.final.to_s)
+        end
+
         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
+          if params[:is_aligned?] and bitr.pos % 8 != 0
             bitr.pos += 8 - (bitr.pos % 8)          
           end
 
           # received return-to-control code 
-          if has_eob and bitr.peek(RTC[1]) == RTC[0]
+          if params[:has_eob?] and bitr.peek(RTC[1]) == RTC[0]
             bitr.pos += RTC[1]
             break
           end
@@ -407,41 +401,99 @@ module Origami
             raise InvalidCCITTFaxDataError.new(
               "No end-of-line pattern found (at bit pos #{bitr.pos}/#{bitr.size}})",
               bitw.final.to_s
-            )if has_eol
+            ) if params[:has_eol?]
           else
             bitr.pos += EOL[1]
           end
 
-          line_length = 0
-          while line_length < columns
+          case
+            when mode == 0
+              decode_one_dimensional_line(bitr, bitw, columns, colors)
+            when mode < 0
+              decode_two_dimensional_line(bitr, bitw, columns, colors, prev_line)
+          end 
+
+
+          rows -= 1 unless params[:has_eob?]
+        end
+
+        bitw.final.to_s
+      end
+
+      private
+
+      def encode_one_dimensional_line(input, output, columns, colors) #:nodoc:
+        output.write(*EOL)
+        scan_len = 0
+        white, black = colors
+        current_color = white
+
+        # Process each bit in line.
+        begin
+          if input.read(1) == current_color
+            scan_len += 1
+          else
             if current_color == white
-              bit_length = get_white_bits(bitr)
+              put_white_bits(output, scan_len)
             else
-              bit_length = get_black_bits(bitr)
+              put_black_bits(output, scan_len)
             end
 
-            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 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
+            scan_len = 1
           end
+        end while input.pos % columns != 0
 
-          rows -= 1 unless has_eob
+        if current_color == white
+          put_white_bits(output, scan_len)
+        else
+          put_black_bits(output, scan_len)
         end
 
-        bitw.final.to_s
+        # Align encoded lign on a 8-bit boundary.
+        if @params.EncodedByteAlign == true and output.pos % 8 != 0
+          output.write(0, 8 - (output.pos % 8))
+        end
       end
 
-      private
+      def encode_two_dimensional_line(input, output, columns, colors, prev_line) #:nodoc:
+        raise NotImplementedError, "CCITT two-dimensional encoding scheme not supported."
+
+        white, black = colors
+        current_color = white
+      end
+
+      def decode_one_dimensional_line(input, output, columns, colors) #:nodoc:
+        white, black = colors
+        current_color = white
+        
+        line_length = 0
+        while line_length < columns
+          if current_color == white
+            bit_length = get_white_bits(input)
+          else
+            bit_length = get_black_bits(input)
+          end
+
+          raise InvalidCCITTFaxDataError.new(
+            "Unfinished line (at bit pos #{input.pos}/#{input.size}})",
+            output.final.to_s
+          ) if bit_length.nil?
+          
+          line_length += bit_length
+          raise InvalidCCITTFaxDataError.new(
+            "Line is too long (at bit pos #{input.pos}/#{input.size}})",
+            output.final.to_s
+          ) if line_length > columns
+
+          write_bit_range(output, current_color, bit_length)
+          current_color ^= 1
+        end
+      end
+
+      def decode_two_dimensional_line(input, output, columns, colors, prev_line) #:nodoc:
+        raise NotImplementedError, "CCITT two-dimensional decoding scheme not supported."
+      end
 
       def get_white_bits(bitr) #:nodoc:
         get_color_bits(bitr, WHITE_CONFIGURATION_DECODE_TABLE, WHITE_TERMINAL_DECODE_TABLE)

data/lib/origami/filters/flate.rb

@@ -83,7 +83,11 @@ module Origami
         begin
           uncompressed = zlib_stream.inflate(stream)
         rescue Zlib::DataError => zlib_except
-          raise InvalidFlateDataError.new(zlib_except.message, zlib_stream.flush_next_out)
+          uncompressed = zlib_stream.flush_next_out
+
+          unless Origami::OPTIONS[:ignore_zlib_errors]
+            raise InvalidFlateDataError.new(zlib_except.message, uncompressed)
+          end
         end
 
         if @params.Predictor.is_a?(Integer)

data/lib/origami/filters.rb

@@ -71,7 +71,7 @@ module Origami
           while length > 0
             if length >= 8 - @ptr_bit
               length -= 8 - @ptr_bit
-              @last_byte = 0 unless @last_byte
+              @last_byte ||= 0 
               @last_byte |= (data >> length) & ((1 << (8 - @ptr_bit)) - 1)
 
               data &= (1 << length) - 1
@@ -79,7 +79,7 @@ module Origami
               @last_byte = nil
               @ptr_bit = 0
             else
-              @last_byte = 0 unless @last_byte
+              @last_byte ||= 0 
               @last_byte |= (data & ((1 << length) - 1)) << (8 - @ptr_bit - length)
               @ptr_bit += length
               
@@ -130,6 +130,11 @@ module Origami
       # Internally used by some filters.
       #
       class BitReader
+        BRUIJIN_TABLE = ::Array.new(32)
+        BRUIJIN_TABLE.size.times { |i| 
+          BRUIJIN_TABLE[((0x77cb531 * (1 << i)) >> 27) & 31] = i 
+        }
+
         def initialize(data)
           @data = data
           reset
@@ -215,6 +220,83 @@ module Origami
           n
         end
 
+        #
+        # Used for bit scanning.
+        # Counts leading zeros. Does not advance read pointer.
+        #
+        def clz
+          count = 0
+          if @ptr_bit != 0
+            bits = peek(8 - @ptr_bit) 
+            count = clz32(bits << (32 - (8 - @ptr_bit)))
+
+            return count if count < (8 - @ptr_bit)
+          end
+
+          delta = 0
+          while @data.size > @ptr_byte + delta * 4
+            word = @data[@ptr_byte + delta * 4, 4] # next 32 bits
+            z = clz32((word << (4 - word.size)).unpack("N")[0])
+            
+            count += z
+            delta += 1
+
+            return count if z < 32 - ((4 - word.size) << 3)
+          end
+
+          count
+        end
+
+        #
+        # Used for bit scanning.
+        # Count leading ones. Does not advance read pointer.
+        #
+        def clo
+          count = 0
+          if @ptr_bit != 0
+            bits = peek(8 - @ptr_bit) 
+            count = clz32(~(bits << (32 - (8 - @ptr_bit))) & 0xff)
+
+            return count if count < (8 - @ptr_bit)
+          end
+
+          delta = 0
+          while @data.size > @ptr_byte + delta * 4
+            word = @data[@ptr_byte + delta * 4, 4] # next 32 bits
+            z = clz32(~((word << (4 - word.size)).unpack("N")[0]) & 0xffff_ffff)
+            
+            count += z
+            delta += 1
+
+            return count if z < 32 - ((4 - word.size) << 3)
+          end
+
+          count
+        end
+
+        private
+
+        def bitswap8(i) #:nodoc
+          ((i * 0x0202020202) & 0x010884422010) % 1023
+        end
+
+        def bitswap32(i) #:nodoc:
+          (bitswap8((i >> 0) & 0xff) << 24) |
+          (bitswap8((i >> 8) & 0xff) << 16) |
+          (bitswap8((i >> 16) & 0xff) << 8) |
+          (bitswap8((i >> 24) & 0xff) << 0) 
+        end
+
+        def ctz32(i) #:nodoc:
+          if i == 0 then 32
+          else
+            BRUIJIN_TABLE[(((i & -i) * 0x77cb531) >> 27) & 31]
+          end
+        end
+
+        def clz32(i) #:nodoc:
+          ctz32 bitswap32 i
+        end
       end
     end
 

data/lib/origami/font.rb

@@ -25,6 +25,73 @@
 
 module Origami
 
+  #
+  # Embedded font stream.
+  #
+  class FontStream < Stream
+    
+    field   :Subtype,                 :Type => Name
+    field   :Length1,                 :Type => Integer
+    field   :Length2,                 :Type => Integer
+    field   :Length3,                 :Type => Integer
+
+  end
+  
+  #
+  # Class representing a font details in a document.
+  #
+  class FontDescriptor < Dictionary
+    
+    include StandardObject
+    
+    FIXEDPITCH  = 1 << 1
+    SERIF       = 1 << 2
+    SYMBOLIC    = 1 << 3
+    SCRIPT      = 1 << 4
+    NONSYMBOLIC = 1 << 6
+    ITALIC      = 1 << 7
+    ALLCAP      = 1 << 17
+    SMALLCAP    = 1 << 18
+    FORCEBOLD   = 1 << 19
+
+    field   :Type,                    :Type => Name, :Default => :FontDescriptor, :Required => true
+    field   :FontName,                :Type => Name, :Required => true
+    field   :FontFamily,              :Type => ByteString, :Version => "1.5"
+    field   :FontStretch,             :Type => Name, :Default => :Normal, :Version => "1.5"
+    field   :FontWeight,              :Type => Integer, :Default => 400, :Version => "1.5"
+    field   :Flags,                   :Type => Integer, :Required => true
+    field   :FontBBox,                :Type => Array
+    field   :ItalicAngle,             :Type => Number, :Required => true
+    field   :Ascent,                  :Type => Number
+    field   :Descent,                 :Type => Number
+    field   :Leading,                 :Type => Number, :Default => 0
+    field   :CapHeight,               :Type => Number
+    field   :XHeight,                 :Type => Number, :Default => 0
+    field   :StemV,                   :Type => Number
+    field   :StemH,                   :Type => Number, :Default => 0
+    field   :AvgWidth,                :Type => Number, :Default => 0
+    field   :MaxWidth,                :Type => Number, :Default => 0
+    field   :MissingWidth,            :Type => Number, :Default => 0
+    field   :FontFile,                :Type => FontStream
+    field   :FontFile2,               :Type => FontStream, :Version => "1.1"
+    field   :FontFile3,               :Type => FontStream, :Version => "1.2"
+    field   :CharSet,                 :Type => ByteString, :Version => "1.1"
+    
+  end
+  
+  #
+  # Class representing a character encoding in a document.
+  #
+  class Encoding < Dictionary
+    
+    include StandardObject
+
+    field   :Type,                    :Type => Name, :Default => :Encoding
+    field   :BaseEncoding,            :Type => Name
+    field   :Differences,             :Type => Array
+   
+  end
+ 
   #
   # Class representing a rendering font in a document.
   #
@@ -33,13 +100,13 @@ module Origami
     include StandardObject
    
     field   :Type,                    :Type => Name, :Default => :Font, :Required => true
-    field   :Subtype,                 :Type => Name, :Default => :Type1, :Required => true
+    field   :Subtype,                 :Type => Name, :Required => true
     field   :Name,                    :Type => Name
     field   :FirstChar,               :Type => Integer
     field   :LastChar,                :Type => Integer
     field   :Widths,                  :Type => Array
-    field   :FontDescriptor,          :Type => Dictionary
-    field   :Encoding,                :Type => [ Name, Dictionary ], :Default => :MacRomanEncoding
+    field   :FontDescriptor,          :Type => FontDescriptor
+    field   :Encoding,                :Type => [ Name, Encoding ], :Default => :MacRomanEncoding
     field   :ToUnicode,               :Type => Stream, :Version => "1.2"
    
     # TODO: Type0 and CID Fonts
@@ -49,7 +116,7 @@ module Origami
     #
     class Type1 < Font
 
-      field   :BaseFont,              :Type => Name, :Default => :Helvetica, :Required => true
+      field   :BaseFont,              :Type => Name, :Required => true
       field   :Subtype,               :Type => Name, :Default => :Type1, :Required => true
       
       #
@@ -139,71 +206,4 @@ module Origami
 
   end
 
-  #
-  # Embedded font stream.
-  #
-  class FontStream < Stream
-    
-    field   :Subtype,                 :Type => Name
-    field   :Length1,                 :Type => Integer
-    field   :Length2,                 :Type => Integer
-    field   :Length3,                 :Type => Integer
-
-  end
-  
-  #
-  # Class representing a font details in a document.
-  #
-  class FontDescriptor < Dictionary
-    
-    include StandardObject
-    
-    FIXEDPITCH  = 1 << 1
-    SERIF       = 1 << 2
-    SYMBOLIC    = 1 << 3
-    SCRIPT      = 1 << 4
-    NONSYMBOLIC = 1 << 6
-    ITALIC      = 1 << 7
-    ALLCAP      = 1 << 17
-    SMALLCAP    = 1 << 18
-    FORCEBOLD   = 1 << 19
-
-    field   :Type,                    :Type => Name, :Default => :FontDescriptor, :Required => true
-    field   :FontName,                :Type => Name, :Required => true
-    field   :FontFamily,              :Type => ByteString, :Version => "1.5"
-    field   :FontStretch,             :Type => Name, :Default => :Normal, :Version => "1.5"
-    field   :FontWeight,              :Type => Integer, :Default => 400, :Version => "1.5"
-    field   :Flags,                   :Type => Integer, :Required => true
-    field   :FontBBox,                :Type => Array
-    field   :ItalicAngle,             :Type => Number, :Required => true
-    field   :Ascent,                  :Type => Number
-    field   :Descent,                 :Type => Number
-    field   :Leading,                 :Type => Number, :Default => 0
-    field   :CapHeight,               :Type => Number
-    field   :XHeight,                 :Type => Number, :Default => 0
-    field   :StemV,                   :Type => Number
-    field   :StemH,                   :Type => Number, :Default => 0
-    field   :AvgWidth,                :Type => Number, :Default => 0
-    field   :MaxWidth,                :Type => Number, :Default => 0
-    field   :MissingWidth,            :Type => Number, :Default => 0
-    field   :FontFile,                :Type => Stream
-    field   :FontFile2,               :Type => Stream, :Version => "1.1"
-    field   :FontFile3,               :Type => Stream, :Version => "1.2"
-    field   :CharSet,                 :Type => ByteString, :Version => "1.1"
-    
-  end
-  
-  #
-  # Class representing a character encoding in a document.
-  #
-  class Encoding < Dictionary
-    
-    include StandardObject
-
-    field   :Type,                    :Type => Name, :Default => :Encoding
-    field   :BaseEncoding,            :Type => Name
-    field   :Differences,             :Type => Array
-   
-  end
- 
 end

data/lib/origami/graphics/patterns.rb

@@ -41,6 +41,7 @@ module Origami
       
       class Tiling < ContentStream
         include Pattern
+        include ResourcesHolder
 
         module PaintType
           COLOURED    = 1
@@ -59,7 +60,7 @@ module Origami
         field   :BBox,                  :Type => Array, :Required => true
         field   :XStep,                 :Type => Number, :Required => true
         field   :YStep,                 :Type => Number, :Required => true
-        field   :Resources,             :Type => Dictionary, :Required => true
+        field   :Resources,             :Type => Resources, :Required => true
         field   :Matrix,                :Type => Array, :Default => [ 1, 0, 0, 1, 0, 0 ]
       
       end