Use URI.get_encoding to look up encodings.

Use the (undocumented) URI.get_encoding method introduced in Ruby 2.1 to
look up encodings by the aliases specified in HTML5. This means that the
behavior will differ slightly between versions of Ruby, but the
encodings selected are largely compatible.

For example, `ISO-8859-1` is an alias for `Windows-1252` per the HTML5
specification, while in ruby versions < 2.1 it will be used as is. These
two encodings are largely compatible, and the alias exists due to
servers that return a `charset=ISO-8859-1` when they actually are using
`Windows-1252`.

Other aliases that differ include `shift_jis` (rendered as
`Windows-31J`) and `euc-jp` (rendered as `CP51932`).

lib/restclient/response.rb

@@ -49,21 +49,29 @@ module RestClient
 
     private
 
+    # Automatically set the encoding of the response object based on the
+    # presence of a Content-Type... charset header.
+    #
+    # If a charset is found and represents a valid encoding, call
+    # force_encoding on the response to alter it to the correct representation.
+    #
+    # @param [Response] response
+    #
+    # @return [Encoding,nil]
+    #
     def self.fix_encoding(response)
       charset = RestClient::Utils.get_encoding_from_headers(response.headers)
-      encoding = nil
+      return unless charset
 
-      begin
-        encoding = Encoding.find(charset) if charset
-      rescue ArgumentError
+      encoding = RestClient::Utils.find_encoding(charset)
+
+      if encoding
+        response.force_encoding(encoding)
+      elsif RestClient.log
         RestClient.log << "No such encoding: #{charset.inspect}"
       end
 
-      return unless encoding
-
-      response.force_encoding(encoding)
-
-      response
+      encoding
     end
 
     def body_truncated(length)

lib/restclient/utils.rb

@@ -29,6 +29,36 @@ module RestClient
       nil
     end
 
+    # Return the Encoding for a String encoding name.
+    #
+    # In ruby 2.1+ use URI.get_encoding() in order to support the encoding
+    # names and aliases specified by HTML5. Otherwise call Encoding.find().
+    #
+    # Note that the HTML5 specification indicates that certain valid encodings
+    # be treated as other similar encodings. For example, `ISO-8859-1` is
+    # rendered as `Windows-1252` even though it differs in certain control
+    # characters.
+    #
+    # @param [String] name A string encoding name, such as "utf-8"
+    # @return [Encoding, nil]
+    #
+    # @see Encoding.find
+    # @see URI.get_encoding
+    # @see https://encoding.spec.whatwg.org/#concept-encoding-get
+    #
+    def self.find_encoding(name)
+      if URI.respond_to?(:get_encoding)
+        return URI.get_encoding(name)
+      end
+
+      begin
+        Encoding.find(name)
+      rescue ArgumentError => e
+        raise unless e.message.include?('unknown encoding name')
+        nil
+      end
+    end
+
     # Parse semi-colon separated, potentially quoted header string iteratively.
     #
     # @private

spec/integration/integration_spec.rb

@@ -61,7 +61,7 @@ describe RestClient do
       body = "\xfe".force_encoding('ASCII-8BIT')
       stub_request(:get, "www.example.com").to_return(
         :body => body, :status => 200, :headers => {
-          'Content-Type' => 'application/octet-stream; charset=binary'
+          'Content-Type' => 'application/octet-stream'
       })
       response = RestClient.get "www.example.com"
       response.encoding.should eq Encoding::BINARY
@@ -71,7 +71,7 @@ describe RestClient do
       response.valid_encoding?.should eq true
     end
 
-    it 'handles euc-jp' do
+    it 'handles CP51932 / EUC-JP' do
       body = "\xA4\xA2\xA4\xA4\xA4\xA6\xA4\xA8\xA4\xAA".
         force_encoding(Encoding::BINARY)
       body_utf8 = 'あいうえお'
@@ -82,12 +82,48 @@ describe RestClient do
           'Content-Type' => 'text/plain; charset=EUC-JP'
       })
       response = RestClient.get 'www.example.com'
-      response.encoding.should eq Encoding::EUC_JP
+
+      # URI.get_encoding turns EUC-JP into CP51932, though I'm not sure why.
+      [Encoding::EUC_JP, Encoding::CP51932].should include(response.encoding)
+
       response.valid_encoding?.should eq true
       response.length.should eq 5
       response.encode('utf-8').should eq body_utf8
     end
 
+    it 'handles big5 traditional chinese' do
+      body = "\xB5e\xAE\xD1"
+      body_utf8 = '畫書'
+      body_utf8.encoding.should eq Encoding::UTF_8
+
+      stub_request(:get, 'www.example.com').to_return(
+        :body => body, :status => 200, :headers => {
+          'Content-Type' => 'text/plain; charset=Big5'
+      })
+      response = RestClient.get 'www.example.com'
+      response.encoding.should eq Encoding::Big5
+      response.valid_encoding?.should eq true
+      response.length.should eq 2
+      response.encode('utf-8').should eq body_utf8
+    end
+
+    it 'handles gbk simplified chinese' do
+      body = "\xBB\xAD\xCA\xE9"
+      body_utf8 = '画书'
+      body_utf8.encoding.should eq Encoding::UTF_8
+
+      stub_request(:get, 'www.example.com').to_return(
+        :body => body, :status => 200, :headers => {
+          'Content-Type' => 'text/plain; charset=GBK'
+      })
+      response = RestClient.get 'www.example.com'
+      response.encoding.should eq Encoding::GBK
+      response.valid_encoding?.should eq true
+      response.length.should eq 2
+      response.encode('utf-8').should eq body_utf8
+    end
+
+
     it 'defaults to Encoding.default_external' do
       stub_request(:get, 'www.example.com').to_return(
         body: 'abc', status: 200, headers: {

spec/unit/utils_spec.rb

@@ -33,6 +33,52 @@ describe RestClient::Utils do
     end
   end
 
+  describe '.find_encoding' do
+    it 'finds various normal encoding names' do
+      {
+        'utf-8' => Encoding::UTF_8,
+        'big5' => Encoding::Big5,
+        'euc-kr' => Encoding::EUC_KR,
+        'WINDOWS-1252' => Encoding::Windows_1252,
+        'windows-31j' => Encoding::Windows_31J,
+      }.each_pair do |name, enc|
+        RestClient::Utils.find_encoding(name).should eq enc
+      end
+    end
+
+    it 'returns nil on failures' do
+      %w{nonexistent utf-99}.each do |name|
+        RestClient::Utils.find_encoding(name).should be_nil
+      end
+    end
+
+    it 'uses URI.get_encoding if available', if: RUBY_VERSION >= '2.1' do
+      {
+        'utf8' => Encoding::UTF_8,
+        'utf-16' => Encoding::UTF_16LE,
+        'latin1' => Encoding::Windows_1252,
+        'iso-8859-1' => Encoding::Windows_1252,
+        'shift_jis' => Encoding::Windows_31J,
+        'euc-jp' => Encoding::CP51932,
+      }.each_pair do |name, enc|
+        RestClient::Utils.find_encoding(name).should eq enc
+      end
+    end
+
+    it 'uses Encoding.find if URI.get_encoding unavailable', if: RUBY_VERSION < '2.1' do
+      {
+        'utf8' => nil,
+        'utf-16' => Encoding::UTF_16,
+        'latin1' => nil,
+        'iso-8859-1' => Encoding::ISO_8859_1,
+        'shift_jis' => Encoding::Shift_JIS,
+        'euc-jp' => Encoding::EUC_JP,
+      }.each_pair do |name, enc|
+        RestClient::Utils.find_encoding(name).should eq enc
+      end
+    end
+  end
+
   describe '.cgi_parse_header' do
     it 'parses headers' do
       RestClient::Utils.cgi_parse_header('text/plain').