coder 0.2.0 → 0.2.1

data/lib/coder.rb

@@ -1,14 +1,25 @@
 require 'coder/version'
 require 'coder/cleaner'
+require 'coder/error'
 
 module Coder
   extend self
+  DEFAULT_ENCODING = 'UTF-8'
 
-  def clean(str, encoding = nil)
-    Cleaner.new(encoding || 'UTF-8').clean(str)
+  def clean(str, encoding = DEFAULT_ENCODING)
+    Cleaner.new(encoding).clean(str)
   end
 
-  def clean!(str, encoding = nil)
+  def clean!(str, encoding = DEFAULT_ENCODING)
     str.replace clean(str, encoding)
   end
+
+  def force_encoding!(str, encoding = DEFAULT_ENCODING)
+    return str unless str.respond_to? :force_encoding
+    str.force_encoding(encoding.to_s)
+  end
+
+  def force_encoding(str, encoding = DEFAULT_ENCODING)
+    force_encoding! str.dup, encoding
+  end
 end

data/lib/coder/cleaner.rb

@@ -2,6 +2,7 @@ require 'coder/cleaner/builtin'
 require 'coder/cleaner/iconv'
 require 'coder/cleaner/java'
 require 'coder/cleaner/simple'
+require 'coder'
 
 module Coder
   module Cleaner

data/lib/coder/cleaner/builtin.rb

@@ -1,4 +1,4 @@
-require 'coder/error'
+require 'coder'
 
 module Coder
   module Cleaner

data/lib/coder/cleaner/iconv.rb

@@ -1,3 +1,4 @@
+require 'coder'
 require 'coder/error'
 require 'stringio'
 
@@ -31,15 +32,17 @@ module Coder
       end
 
       def initialize(encoding)
+        @encoding = encoding.to_s
         @nullbyte = "\0"
-        @iconv    = ::Iconv.new("#{encoding}//ignore", encoding.to_s)
-        @nullbyte.encode! encoding if @nullbyte.respond_to? :encode!
+        @iconv    = ::Iconv.new("#{encoding}//ignore", @encoding)
+        @nullbyte.encode! @encoding if @nullbyte.respond_to? :encode!
       rescue ::Iconv::InvalidEncoding => e
         raise Coder::InvalidEncoding, e.message
       end
 
       def clean(str)
-        @iconv.iconv(str).gsub(@nullbyte, "")
+        string = @iconv.iconv(str).gsub(@nullbyte, "")
+        Coder.force_encoding! string, @encoding
       rescue ::Iconv::Failure => e
         raise Coder::Error, e.message
       end

data/lib/coder/cleaner/java.rb

@@ -1,3 +1,4 @@
+require 'coder'
 require 'coder/error'
 
 module Coder
@@ -15,7 +16,7 @@ module Coder
       end
 
       def initialize(encoding)
-        encoding  = encoding.to_s.upcase
+        @encoding = encoding.to_s.upcase
         @nullbyte = "\0"
         @charset  = ::Java::JavaNioCharset::Charset.for_name(encoding)
         @decoder  = @charset.new_decoder
@@ -30,7 +31,8 @@ module Coder
 
       def clean(str)
         buffer = ::Java::JavaNio::ByteBuffer.wrap(str.to_java_bytes)
-        @decoder.decode(buffer).to_s.gsub(@nullbyte, '')
+        string = @decoder.decode(buffer).to_s
+        Coder.force_encoding!(string, @encoding).gsub(@nullbyte, '')
       rescue Java::JavaLang::RuntimeException => e
         raise Coder::Error, e.message, e.backtrace
       end

data/lib/coder/cleaner/simple.rb

@@ -1,4 +1,4 @@
-require 'coder/error'
+require 'coder'
 require 'coder/cleaner/simple/byte_buffer'
 require 'coder/cleaner/simple/encodings'
 
@@ -25,14 +25,7 @@ module Coder
       def clean(str)
         bytes = ByteBuffer.new(@encoding)
         str.each_byte { |b| bytes << b }
-        force_encoding bytes.to_s
-      end
-
-      private
-
-      def force_encoding(str)
-        return str unless str.respond_to? :force_encoding
-        str.force_encoding(@name)
+        Coder.force_encoding!(bytes.to_s, @name)
       end
     end
   end

data/lib/coder/version.rb

@@ -1,3 +1,3 @@
 module Coder
-  VERSION = "0.2.0"
+  VERSION = "0.2.1"
 end

data/spec/coder/cleaner_spec.rb

@@ -1,7 +1,5 @@
 # encoding: UTF-8
-require 'coder/cleaner'
-require 'coder/error'
-require 'support/clean_helpers'
+require 'spec_helper'
 
 shared_examples Coder::Cleaner do
   encoding "UTF-8" do
@@ -11,12 +9,14 @@ shared_examples Coder::Cleaner do
     cleans "\0", ""
     cleans "{foo \xC3 'bar'}", "{foo  'bar'}"
     cleans "yummy\xE2 \xF0\x9F\x8D\x94 \x9F\x8D\x94", "yummy 🍔 "
+    sets_encoding
   end
 
   encoding "UCS-2BE" do
     cleans "\x00f\x00o\x00o"
     cleans "\x00f\x00ox", "\x00f\x00o"
     cleans "\x00f\x00o\x00\x00", "\x00f\x00o"
+    sets_encoding
   end
 
   encoding "UCS-4BE" do
@@ -24,6 +24,7 @@ shared_examples Coder::Cleaner do
     cleans "\x00\x00\x00f\x00\x00\x00o\x00\x00x", "\x00\x00\x00f\x00\x00\x00o"
     cleans "\x00\x00\x00f\x00\x00\x00o\x00\x00\x00\x00", "\x00\x00\x00f\x00\x00\x00o"
     cleans "\xFF\xFF\x10\x10", ""
+    sets_encoding
   end
 
   context "unknown encoding" do

data/spec/coder_spec.rb

@@ -0,0 +1,64 @@
+# encoding: UTF-8
+require 'spec_helper'
+
+describe Coder do
+  describe :clean do
+    it 'cleans up UTF-8 strings without having to specify the encoding' do
+      Coder.clean("yummy\xE2 \xF0\x9F\x8D\x94 \x9F\x8D\x94").should binary_equal("yummy 🍔 ")
+    end
+
+    it 'allows specifying the encoding' do
+      Coder.clean("\x00f\x00ox", "UCS-2BE").should binary_equal("\x00f\x00o")
+    end
+  end
+
+  describe :force_encoding do
+    it 'returns a different string' do
+      str = ''
+      Coder.force_encoding(str).should_not be_equal(str)
+    end
+
+    it 'leaves the content untouched' do
+      Coder.force_encoding('foo').should be == 'foo'
+    end
+
+    if ''.respond_to? :force_encoding
+      it 'sets the encoding to UTF-8' do
+        Coder.force_encoding(''.encode('binary')).encoding.name.should be == 'UTF-8'
+      end
+
+      it 'allows specifying a different encoding' do
+        Coder.force_encoding(''.encode('binary'), 'UTF-16BE').encoding.name.should be == 'UTF-16BE'
+      end
+
+      it 'does not modify the encoding of the passed in string' do
+        str = ''.encode('binary')
+        Coder.force_encoding str
+        str.encoding.name.should be == 'ASCII-8BIT'
+      end
+    end
+  end
+
+  describe :force_encoding! do
+    it 'returns the string' do
+      str = ''
+      Coder.force_encoding!(str).should be_equal(str)
+    end
+
+    if ''.respond_to? :force_encoding
+      it 'sets the encoding to UTF-8' do
+        Coder.force_encoding!(''.encode('binary')).encoding.name.should be == 'UTF-8'
+      end
+
+      it 'allows specifying a different encoding' do
+        Coder.force_encoding!(''.encode('binary'), 'UTF-16BE').encoding.name.should be == 'UTF-16BE'
+      end
+
+      it 'modifies the encoding of the passed in string' do
+        str = ''.encode('binary')
+        Coder.force_encoding! str
+        str.encoding.name.should be == 'UTF-8'
+      end
+    end
+  end
+end

data/spec/spec_helper.rb

@@ -0,0 +1,3 @@
+require 'coder'
+require 'support/clean_helpers'
+require 'support/binary_matcher'

data/spec/support/binary_matcher.rb

@@ -0,0 +1,4 @@
+RSpec::Matchers.define :binary_equal do |expected|
+  match { |actual| actual.bytes.to_a == expected.bytes.to_a }
+  diffable
+end

data/spec/support/clean_helpers.rb

@@ -3,22 +3,24 @@ module CleanHelpers
     def encoding(encoding, &block)
       return unless described_class.supports? encoding
       context(encoding) do
-        let(:encoding) { encoding}
+        let(:encoding) { encoding }
+        subject { described_class.new(encoding) }
         instance_eval(&block)
       end
     end
 
     def cleans(from, to = from)
       it "cleans #{from.inspect} to #{to.inspect}" do
-        result = described_class.new(encoding).clean(binary(from))
-        binary(result).should be == binary(to)
+        subject.clean(from).should binary_equal(to)
       end
     end
-  end
 
-  def binary(str)
-    return str unless str.respond_to? :force_encoding
-    str.force_encoding('binary')
+    def sets_encoding
+      return unless ''.respond_to? :encoding
+      it 'sets encoding properly' do
+        subject.clean(''.encode('US-ASCII')).encoding.names.should include(encoding)
+      end
+    end
   end
 
   def support(encoding)

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: coder
 version: !ruby/object:Gem::Version
-  version: 0.2.0
+  version: 0.2.1
   prerelease: 
 platform: ruby
 authors:
@@ -69,6 +69,9 @@ files:
 - lib/coder/error.rb
 - lib/coder/version.rb
 - spec/coder/cleaner_spec.rb
+- spec/coder_spec.rb
+- spec/spec_helper.rb
+- spec/support/binary_matcher.rb
 - spec/support/clean_helpers.rb
 homepage: http://github.com/rkh/coder
 licenses: []
@@ -96,5 +99,8 @@ specification_version: 3
 summary: library to handle encodings
 test_files:
 - spec/coder/cleaner_spec.rb
+- spec/coder_spec.rb
+- spec/spec_helper.rb
+- spec/support/binary_matcher.rb
 - spec/support/clean_helpers.rb
 has_rdoc: