ruby-protocol-buffers 1.2.1 → 1.2.2

data/VERSION

@@ -1 +1 @@
-1.2.1
+1.2.2

data/lib/protocol_buffers/runtime/encoder.rb

@@ -1,12 +1,16 @@
 module ProtocolBuffers
 
-  class EncodeError < StandardError; end
+  class EncodeError < StandardError
+    attr_reader :invalid_field
+
+    def initialize(invalid_field)
+      @invalid_field = invalid_field
+    end
+  end
 
   module Encoder # :nodoc: all
     def self.encode(io, message)
-      unless message.valid?
-        raise(EncodeError, "invalid message")
-      end
+      message.validate!
 
       message.fields.each do |tag, field|
         next unless message.value_for_tag?(tag)
@@ -40,7 +44,7 @@ module ProtocolBuffers
       when 1, 5 # FIXED64, FIXED32
         io.write(serialized)
       when 2 # LENGTH_DELIMITED
-        Varint.encode(io, serialized.length)
+        Varint.encode(io, serialized.bytesize)
         io.write(serialized)
       when 3, 4 # deprecated START_GROUP/END_GROUP types
         raise(EncodeError, "groups are deprecated and unsupported")

data/lib/protocol_buffers/runtime/field.rb

@@ -142,7 +142,7 @@ module ProtocolBuffers
           def #{name}=(value)
             field = fields[#{tag}]
             if value.nil?
-              @set_fields.delete_at(#{tag})
+              @set_fields[#{tag}] = false
               @#{name} = field.default_value
             else
               field.check_valid(value)
@@ -263,6 +263,18 @@ module ProtocolBuffers
       #
       # Maybe we just punt on this except in Ruby 1.9 where we can rely on the
       # language ensuring the string is always UTF-8?
+
+      def deserialize(value)
+        # To get bytes, the value was being read as ASCII.  Ruby 1.9 stores an encoding
+        # with its strings, and they were getting returned with Encoding ASCII-8BIT.
+        # Protobuffers are supposed to only return UTF-8 strings.  This attempts to
+        # force the encoding to UTF-8 if on Ruby 1.9 (force_encoding is defined on String).
+        read_value = value.read.to_s
+        if read_value.respond_to?("force_encoding")
+          read_value.force_encoding("UTF-8")
+        end
+        read_value
+      end
     end
 
     class NumericField < Field
@@ -301,8 +313,9 @@ module ProtocolBuffers
 
     class SignedVarintField < VarintField
       def deserialize(value)
+        # This is to handle negatives...they are always 64-bit
         if value > max
-          value - (1<<bits)
+          value - (1<<64)
         else
           value
         end

data/lib/protocol_buffers/runtime/message.rb

@@ -458,16 +458,25 @@ module ProtocolBuffers
       self.class.valid?(self)
     end
 
-    def self.valid?(message)
+    def self.valid?(message, raise_exception=false)
       return true unless @has_required_field
 
       fields.each do |tag, field|
-        if field.otype == :required
-          return false unless message.value_for_tag?(tag)
+        if field.otype == :required && !message.value_for_tag?(tag)
+          return false unless raise_exception
+          raise(ProtocolBuffers::EncodeError.new(field), "Required field '#{field.name}' is invalid")
         end
       end
     end
 
+    def validate!
+      self.class.validate!(self)
+    end
+
+    def self.validate!(message)
+      valid?(message, true)
+    end
+
     def remember_unknown_field(tag_int, value)
       @unknown_fields || @unknown_fields = []
       @unknown_fields << [tag_int, value]

data/spec/negative_int32_spec.rb

@@ -0,0 +1,33 @@
+# Tests for issues with decoding negative int32 fields
+
+# By Mark Herman <mherman@iseatz.com> based on code from
+# Tom Zetty <tzetty@iseatz.com>
+
+require File.expand_path(File.dirname(__FILE__) + '/spec_helper')
+
+$LOAD_PATH.unshift(File.join(File.dirname(__FILE__), "..", "lib"))
+require 'protocol_buffers'
+
+describe "Testing for decode errors for negative int32 fields" do
+  # These should fail without the int32 negative handling fix 
+  it "should return -1 given -1" do
+    (validate_pbr(SignedIntTest, -1, true)).should be_true
+  end
+  
+  it "should return -1111 given -1111" do
+    (validate_pbr(SignedIntTest, -1111, true)).should be_true
+  end
+  
+  # These should pass with or without the negative handling fix
+  it "should return 1 given 1" do
+    (validate_pbr(SignedIntTest, 1, true)).should be_true
+  end
+  
+  it "should return 0 given 0" do
+    (validate_pbr(SignedIntTest, 0, true)).should be_true
+  end
+  
+  it "should return 100000 given 100000" do
+    (validate_pbr(SignedIntTest, 100000, true)).should be_true    
+  end
+end

data/spec/nil_bugs_spec.rb

@@ -0,0 +1,24 @@
+# Tests for issues with handling of nil fields in protobuffers
+
+# By Mark Herman <mherman@iseatz.com>
+
+require File.expand_path(File.dirname(__FILE__) + '/spec_helper')
+
+$LOAD_PATH.unshift(File.join(File.dirname(__FILE__), "..", "lib"))
+require 'protocol_buffers'
+
+# Without the nil fix, assigning nil to a field would cause the encoder to
+# start failing to encode fields.
+describe "testing handling of nil assignments to protobuffers" do
+  it "should accept nil assignments into a single field without losing other data" do
+    # I will assign nil a couple of times and then test whether the protobuffer encodes at all
+    test_pb = NilTest.new
+    test_pb.test_field_1 = "test_value_1"
+    test_pb.test_field_2 = "test_value_2"
+    
+    test_pb.test_field_1 = nil
+    test_pb.test_field_1 = nil
+    
+    (test_pb.to_s.length > 0).should be_true
+  end
+end

data/spec/runtime_spec.rb

@@ -333,6 +333,16 @@ describe ProtocolBuffers, "runtime" do
     res1 = TehUnknown::MyResult.new(:field_2 => 'b')
 
     proc { res1.to_s }.should raise_error(ProtocolBuffers::EncodeError)
+
+    begin
+      res1.to_s
+    rescue Exception => e
+      e.invalid_field.name.should == :field_1
+      e.invalid_field.tag.should == 1
+      e.invalid_field.otype.should == :required
+      e.invalid_field.default_value.should == ''
+    end
+
   end
 
   it "enforces required fields on deserialization" do

data/spec/spec_helper.rb

@@ -2,5 +2,97 @@ require 'protocol_buffers'
 
 require 'rspec'
 
+# These are a couple of classes used by tests
+class SignedIntTest < ::ProtocolBuffers::Message
+  required :int32, :test_member, 1
+  gen_methods! # new fields ignored after this point
+end
+
+class StringTest < ::ProtocolBuffers::Message
+  optional :string, :test_member, 1
+  gen_methods! # new fields ignored after this point
+end
+
+class NilTest < ::ProtocolBuffers::Message
+  optional :string, :test_field_1, 1
+  optional :string, :test_field_2, 2
+  gen_methods! # new fields ignored after this point
+end
+
+# klass should be derived from ::ProtocolBuffers::Message
+# and have a member named test_member
+# value is assigned to test member, then an encoding/parse cycle occurs
+# rescues are reported
+
+# suppress_output is meant for use in RSpec tests where output is not desired
+
+# This function returns true if value == decoded proto value
+def validate_pbr(klass, value, suppress_output = false)
+  unless suppress_output
+    puts
+    puts "Validate Pbr: class=#{klass.name}, value=#{value}"
+    puts "  creating and encoding"
+  end
+
+  encode_pbr = klass.new
+  encode_pbr.test_member = value
+  encoded_string = encode_pbr.to_s
+
+  unless suppress_output
+    puts "  encoded_string: #{to_hex_string encoded_string}"
+    puts "  encoded length: #{encoded_string.length}"
+    puts "  parsing encoded_string"
+  end
+  
+  decode_pbr = nil;
+  begin
+    decode_pbr = klass.parse encoded_string
+  rescue Exception => e
+    # Exceptions always return false
+    unless suppress_output
+      puts e.message
+      puts "  FAIL: RESCUE occured in #{klass.name}.parse"
+    end
+    return false
+  end
+  
+  if decode_pbr
+    unless suppress_output
+      puts "  decoded value: #{decode_pbr.test_member}"
+      puts "   passed value: #{value}"
+      puts "  decoded value <=> passed value = #{decode_pbr.test_member <=> value}"
+    end
+
+    if value.respond_to?("bytesize")
+      unless suppress_output
+        puts "  decoded value bytesize: #{decode_pbr.test_member.bytesize}"
+        puts "  passed value bytesize : #{value.bytesize}"
+        puts "  decoded value inspect : #{decode_pbr.test_member.inspect}"
+        puts "  passed value inspect  : #{value.inspect}"
+      end
+      
+      # Ruby 1.8 Strings don't have encodings
+      if decode_pbr.test_member.respond_to?("encoding")
+        unless suppress_output
+          puts "  decoded value encoding: #{decode_pbr.test_member.encoding.name}"
+          puts "  passed value encoding : #{value.encoding.name}"
+        end
+      end
+    end
+    
+    unless suppress_output
+      puts "  GOOD COMPARE" if decode_pbr.test_member == value
+    end
+    
+    decode_pbr.test_member == value
+  end
+end
+
+def to_hex_string ss
+  yy = []
+  ss.each_byte { |b| yy << b.to_s(16) }
+  yy.join(' ')
+end
+
 RSpec.configure do |config|
 end

data/spec/unicode_string_spec.rb

@@ -0,0 +1,33 @@
+# Tests for Unicode issues in protobuffer library.
+
+# By Mark Herman <mherman@iseatz.com> based on code from
+# Tom Zetty <tzetty@iseatz.com>
+
+require File.expand_path(File.dirname(__FILE__) + '/spec_helper')
+
+$LOAD_PATH.unshift(File.join(File.dirname(__FILE__), "..", "lib"))
+require 'protocol_buffers'
+
+describe "Testing for bugs in Unicode encoding and decoding" do
+  # This should always pass
+  it "should return the input string given a regular string" do
+    validate_pbr(StringTest, "test_string", true).should be_true
+  end
+
+  # Encoding objects aren't defined in Ruby 1.8.  This will only run on 1.9 or above.
+  # It should fail on Ruby 1.9 without the Unicode fix.
+  if ''.respond_to? "encoding"
+    it "should return the given UTF8 string" do
+      validate_pbr(StringTest, ''.encode(Encoding::UTF_8) + "utf8_rulz", true)
+    end
+  end
+
+  # This should pass in Ruby 1.8 with the Unicode fix, but it will know nothing of the encoding
+  # This should fail without the Unicode fix. The previous implementation was encoding
+  # with the character length rather than the byte length.  It was also returning strings
+  # with their encoding set to ASCII.  The new code forces it to UTF-8.
+  it "should return the given Unicode string" do
+    string_with_r = "(R) Char: \u00AE"
+    validate_pbr(StringTest, string_with_r, true).should be_true
+  end
+end

metadata

@@ -1,13 +1,13 @@
 --- !ruby/object:Gem::Specification 
 name: ruby-protocol-buffers
 version: !ruby/object:Gem::Version 
-  hash: 29
+  hash: 27
   prerelease: 
   segments: 
   - 1
   - 2
-  - 1
-  version: 1.2.1
+  - 2
+  version: 1.2.2
 platform: ruby
 authors: 
 - Brian Palmer
@@ -17,8 +17,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2011-10-01 00:00:00 -06:00
-default_executable: 
+date: 2012-01-16 00:00:00 Z
 dependencies: 
 - !ruby/object:Gem::Dependency 
   name: rspec
@@ -137,6 +136,8 @@ files:
 - ruby-protocol-buffers.gemspec
 - spec/compiler_spec.rb
 - spec/fields_spec.rb
+- spec/negative_int32_spec.rb
+- spec/nil_bugs_spec.rb
 - spec/proto_files/depends.proto
 - spec/proto_files/dotted_package.proto
 - spec/proto_files/featureful.proto
@@ -147,7 +148,7 @@ files:
 - spec/runtime_spec.rb
 - spec/spec.opts
 - spec/spec_helper.rb
-has_rdoc: true
+- spec/unicode_string_spec.rb
 homepage: https://github.com/mozy/ruby-protocol-buffers
 licenses: []
 
@@ -177,13 +178,15 @@ required_rubygems_version: !ruby/object:Gem::Requirement
 requirements: []
 
 rubyforge_project: 
-rubygems_version: 1.6.2
+rubygems_version: 1.8.11
 signing_key: 
 specification_version: 3
 summary: Ruby compiler and runtime for the google protocol buffers library.
 test_files: 
 - spec/compiler_spec.rb
 - spec/fields_spec.rb
+- spec/negative_int32_spec.rb
+- spec/nil_bugs_spec.rb
 - spec/proto_files/depends.proto
 - spec/proto_files/dotted_package.proto
 - spec/proto_files/featureful.proto
@@ -194,3 +197,4 @@ test_files:
 - spec/runtime_spec.rb
 - spec/spec.opts
 - spec/spec_helper.rb
+- spec/unicode_string_spec.rb