Ascii85 2.0.0 → 2.0.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: e134217a95580db89dd446ac40511345e8c0a317be617b9b04e2ac8ca5db9670
-  data.tar.gz: bcc1f3f71f2f5748958825602ef58c7365ef52f7fe3265a9409fd1b84b5c8d4f
+  metadata.gz: 32c27ed5387adae778bb5f1b5ef82da6ca878d6cd3128dd9a6d65c182ad89296
+  data.tar.gz: 81ba054e37db894fcb7411c143c6cfade2437bc91fd00956eeaf16233e96c25e
 SHA512:
-  metadata.gz: 4c15434410e46485ada5dcc04929092f77da1533861d22ffe65ab0e2f1e8300a821b16524fbeb5ad137cc6a139232a5684dfe6fd34c5ad33e19d68a95d3f604e
-  data.tar.gz: a397048e6009d3adf0c2582cfe9dbd913f6b8888c19fbf48939b1a7ab48de704fa7fb76253d5d3a05d105215fb9e9f2da299d91c12cf5e6c2a3cb2636a2b6f8a
+  metadata.gz: da7757b98444ed6e745c0c9aa7930aa20346a273f94d316cb6eae473a6be293fd592320718f3d1eaef0ebc7d877deb4cb8ec0a43b13f4595835f56053abf3e5d
+  data.tar.gz: e881585d8b1f1cbdba501e7891daa2ed19b268c0b67d0cc8e424d944ce84d16c685b563bc18f34cd92db45b9a8e306e99e7b62b6d41908443eecdc15ee8bcf30

data/CHANGELOG.md

@@ -1,5 +1,11 @@
 # Ascii85 Changelog
 
+## [2.0.1] - 2024-09-15
+
+### Fixed
+
+- Decoding binary data could lead to Encoding errors (Issue #8)
+
 ## [2.0.0] - 2024-08-20
 
 ### BREAKING CHANGES

data/lib/Ascii85/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Ascii85
-  VERSION = '2.0.0'
+  VERSION = '2.0.1'
 end

data/lib/ascii85.rb

@@ -14,6 +14,11 @@ require 'stringio'
 #
 module Ascii85
   class << self
+    EMPTY_STRING  = ''.dup.force_encoding(Encoding::ASCII_8BIT)
+    START_MARKER  = '<~'.dup.force_encoding(Encoding::ASCII_8BIT)
+    ENDING_MARKER = '~>'.dup.force_encoding(Encoding::ASCII_8BIT)
+    LINE_BREAK    = "\n".dup.force_encoding(Encoding::ASCII_8BIT)
+
     #
     # Encodes the bytes of the given String or IO-like object as Ascii85.
     #
@@ -55,22 +60,24 @@ module Ascii85
                  StringIO.new(str_or_io.to_s, 'rb')
                end
 
-      return ''.dup if reader.eof?
+      return EMPTY_STRING.dup if reader.eof?
 
       # Setup buffered Reader and Writers
       bufreader = BufferedReader.new(reader, unencoded_chunk_size)
       bufwriter = BufferedWriter.new(out || StringIO.new(String.new, 'wb'), encoded_chunk_size)
       writer = wrap_lines ? Wrapper.new(bufwriter, wrap_lines) : DummyWrapper.new(bufwriter)
 
-      padding = "\0\0\0\0"
-      tuplebuf = '!!!!!'.dup
+      padding = unfrozen_binary_copy("\0\0\0\0")
+      tuplebuf = unfrozen_binary_copy('!!!!!')
+      exclamations = unfrozen_binary_copy('!!!!!')
+      z = unfrozen_binary_copy('z')
 
       bufreader.each_chunk do |chunk|
         chunk.unpack('N*').each do |word|
           # Encode each big-endian 32-bit word into a 5-character tuple (except
           # for 0, which encodes to 'z')
           if word.zero?
-            writer.write('z')
+            writer.write(z)
           else
             word, b0 = word.divmod(85)
             word, b1 = word.divmod(85)
@@ -98,7 +105,7 @@ module Ascii85
 
         # Encode the last word and cut off any padding
         if word.zero?
-          writer.write('!!!!!'[0..(4 - padding_length)])
+          writer.write(exclamations[0..(4 - padding_length)])
         else
           word, b0 = word.divmod(85)
           word, b1 = word.divmod(85)
@@ -119,7 +126,7 @@ module Ascii85
       # If no output IO-object was provided, extract the encoded String from the
       # default StringIO writer. We force the encoding to 'ASCII-8BIT' to work
       # around a TruffleRuby bug.
-      return writer.finish.io.string.force_encoding('ASCII-8BIT') if out.nil?
+      return writer.finish.io.string.force_encoding(Encoding::ASCII_8BIT) if out.nil?
 
       # Otherwise we make sure to flush the output writer, and then return it.
       writer.finish.io
@@ -151,8 +158,8 @@ module Ascii85
 
       # Get the positions of the opening/closing delimiters. If there is no pair
       # of opening/closing delimiters, return an unfrozen empty String.
-      (start_pos = input.index(opening_delim))                or return ''.dup
-      (end_pos   = input.index(closing_delim, start_pos + 2)) or return ''.dup
+      (start_pos = input.index(opening_delim))                or return EMPTY_STRING.dup
+      (end_pos   = input.index(closing_delim, start_pos + 2)) or return EMPTY_STRING.dup
 
       # Get the String inside the delimiter-pair
       input[(start_pos + 2)...end_pos]
@@ -226,7 +233,7 @@ module Ascii85
                end
 
       # Return an unfrozen String on empty input
-      return ''.dup if reader.eof?
+      return EMPTY_STRING.dup if reader.eof?
 
       # Setup buffered Reader and Writers
       bufreader = BufferedReader.new(reader, encoded_chunk_size)
@@ -238,7 +245,8 @@ module Ascii85
       # Decode
       word   = 0
       count  = 0
-      wordbuf = "\0\0\0\0".dup
+      zeroes = unfrozen_binary_copy("\0\0\0\0")
+      wordbuf = zeroes.dup
 
       bufreader.each_chunk do |chunk|
         chunk.each_byte do |c|
@@ -251,7 +259,7 @@ module Ascii85
             raise(Ascii85::DecodingError, "Found 'z' inside Ascii85 5-tuple") unless count.zero?
 
             # Expand z to 0-word
-            bufwriter.write("\0\0\0\0")
+            bufwriter.write(zeroes)
 
           when '!'..'u'
             # Decode 5 characters into a 4-byte word
@@ -286,7 +294,7 @@ module Ascii85
       # We're done if all 5-tuples have been consumed
       if count.zero?
         bufwriter.flush
-        return out || bufwriter.io.string.force_encoding('ASCII-8BIT')
+        return out || bufwriter.io.string.force_encoding(Encoding::ASCII_8BIT)
       end
 
       raise(Ascii85::DecodingError, 'Last 5-tuple consists of single character') if count == 1
@@ -300,11 +308,17 @@ module Ascii85
       bufwriter.write(((word >> 8) & 0xff).chr) if count == 3
       bufwriter.flush
 
-      out || bufwriter.io.string.force_encoding('ASCII-8BIT')
+      out || bufwriter.io.string.force_encoding(Encoding::ASCII_8BIT)
     end
 
     private
 
+    # Copies the given String and forces the encoding of the returned copy to
+    # be Encoding::ASCII_8BIT.
+    def unfrozen_binary_copy(str)
+      str.dup.force_encoding(Encoding::ASCII_8BIT)
+    end
+
     # Buffers an underlying IO object to increase efficiency. You do not need
     # to use this directly.
     #
@@ -337,7 +351,7 @@ module Ascii85
       def initialize(io, buffer_size)
         @io = io
         @buffer_size = buffer_size
-        @buffer = String.new(capacity: buffer_size)
+        @buffer = String.new(capacity: buffer_size, encoding: Encoding::ASCII_8BIT)
       end
 
       def write(tuple)
@@ -360,7 +374,7 @@ module Ascii85
     class DummyWrapper
       def initialize(out)
         @out = out
-        @out.write('<~')
+        @out.write(START_MARKER)
       end
 
       def write(buffer)
@@ -368,7 +382,7 @@ module Ascii85
       end
 
       def finish
-        @out.write('~>')
+        @out.write(ENDING_MARKER)
         @out.flush
 
         @out
@@ -385,7 +399,7 @@ module Ascii85
         @line_length = [2, wrap_lines.to_i].max
 
         @out = out
-        @out.write('<~')
+        @out.write(START_MARKER)
 
         @cur_len = 2
       end
@@ -402,7 +416,7 @@ module Ascii85
 
           remaining = @line_length - @cur_len
           @out.write(buffer[0...remaining])
-          @out.write("\n")
+          @out.write(LINE_BREAK)
           @cur_len = 0
           buffer = buffer[remaining..]
           return if buffer.empty?
@@ -411,8 +425,8 @@ module Ascii85
 
       def finish
         # Add the closing delimiter (may need to be pushed to the next line)
-        @out.write("\n") if @cur_len + 2 > @line_length
-        @out.write('~>')
+        @out.write(LINE_BREAK) if @cur_len + 2 > @line_length
+        @out.write(ENDING_MARKER)
 
         @out.flush
         @out

data/spec/lib/ascii85_spec.rb

@@ -35,7 +35,11 @@ TEST_CASES = {
       '<~j+42iJVN3:K&_E6j+<0KJW/W?W8iG`j+EuaK"9on^Z0sZj+FJoK:LtSKB%T?~>',
 
   [Math::PI].pack('G') => '<~5RAV2<(&;T~>',
-  [Math::E].pack('G') => '<~5R"n0M\\K6,~>'
+  [Math::E].pack('G') => '<~5R"n0M\\K6,~>',
+
+  # Minified example from Github issue 8.
+  # Note that OT and OU as the trailing characters are equivalent.
+  "\x9B\xB6\xB9+\x91" => '<~S$ojXOT~>'
 }.freeze
 
 describe Ascii85 do
@@ -59,6 +63,12 @@ describe Ascii85 do
       end
     end
 
+    it 'should always return unfrozen Strings' do
+      TEST_CASES.each_pair do |input, encoded|
+        assert_equal false, Ascii85.encode(input).frozen?
+      end
+    end
+
     it 'should encode Strings in different encodings correctly' do
       input_euc_jp = 'どうもありがとうミスターロボット'.encode('EUC-JP')
       input_binary = input_euc_jp.force_encoding('ASCII-8BIT')
@@ -139,6 +149,12 @@ describe Ascii85 do
       end
     end
 
+    it 'should always return unfrozen Strings' do
+      TEST_CASES.each_pair do |input, encoded|
+        assert_equal false, Ascii85.decode(encoded).frozen?
+      end
+    end
+
     it 'should accept valid input in encodings other than the default' do
       input = 'Ragnarök  τέχνη  русский язык  I ♥ Ruby'
       input_ascii85 = Ascii85.encode(input)
@@ -217,6 +233,13 @@ describe Ascii85 do
       end
     end
 
+    it 'should always return unfrozen Strings' do
+      TEST_CASES.each_pair do |decoded, input|
+        raw_input = input[2...-2] # Remove '<~' and '~>'
+        assert_equal false, Ascii85.decode_raw(raw_input).frozen?
+      end
+    end
+
     it 'should decode from an IO object' do
       input = StringIO.new(';KZGo')
       result = Ascii85.decode_raw(input)

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: Ascii85
 version: !ruby/object:Gem::Version
-  version: 2.0.0
+  version: 2.0.1
 platform: ruby
 authors:
 - Johannes Holzfuß
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2024-08-20 00:00:00.000000000 Z
+date: 2024-09-15 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: minitest