* lib/cgi/util.rb (CGI.escape): support a string with invalid byte

sequence.

* test/cgi/test_cgi_util.rb
  (test_cgi_escape_with_invalid_byte_sequence): test for the above
  change.

git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@34346 b2dd03c8-39d4-4d8f-98ff-823fe69b080e

ChangeLog

@@ -1,3 +1,12 @@
+Fri Jan 20 20:47:37 2012  Kenta Murata  <mrkn@cookpad.com>
+
+	* lib/cgi/util.rb (CGI.escape): support a string with invalid byte
+	  sequence.
+
+	* test/cgi/test_cgi_util.rb
+	  (test_cgi_escape_with_invalid_byte_sequence): test for the above
+	  change.
+
 Fri Jan 20 17:37:37 2012  NARUSE, Yui  <naruse@ruby-lang.org>
 
 	* vm.c (vm_exec): remove workaround for LLVM because r34278 fixes it.

lib/cgi/util.rb

@@ -4,9 +4,10 @@ class CGI
   #   url_encoded_string = CGI::escape("'Stop!' said Fred")
   #      # => "%27Stop%21%27+said+Fred"
   def CGI::escape(string)
-    string.gsub(/([^ a-zA-Z0-9_.-]+)/) do
+    encoding = string.encoding
+    string.dup.force_encoding('ASCII-8BIT').gsub(/([^ a-zA-Z0-9_.-]+)/) do
       '%' + $1.unpack('H2' * $1.bytesize).join('%').upcase
-    end.tr(' ', '+')
+    end.tr(' ', '+').force_encoding(encoding)
   end
 
   # URL-decode a string with encoding(optional).

test/cgi/test_cgi_util.rb

@@ -24,6 +24,12 @@ class CGIUtilTest < Test::Unit::TestCase
     assert_equal('%26%3C%3E%22+%E3%82%86%E3%82%93%E3%82%86%E3%82%93'.ascii_only?, CGI::escape(@str1).ascii_only?) if defined?(::Encoding)
   end
 
+  def test_cgi_escape_with_invalid_byte_sequence
+    assert_nothing_raised(ArgumentError) do
+      assert_equal('%C0%3C%3C', CGI::escape("\xC0<<".force_encoding("UTF-8")))
+    end
+  end
+
   def test_cgi_unescape
     assert_equal(@str1, CGI::unescape('%26%3C%3E%22+%E3%82%86%E3%82%93%E3%82%86%E3%82%93'))
     assert_equal(@str1.encoding, CGI::unescape('%26%3C%3E%22+%E3%82%86%E3%82%93%E3%82%86%E3%82%93').encoding) if defined?(::Encoding)