naruse	2013-11-22 17:50:14 +0900 (Fri, 22 Nov 2013)

  New Revision: 43802

  http://svn.ruby-lang.org/cgi-bin/viewvc.cgi?view=rev&revision=43802

  Log:
    * transcode.c (str_transcode0): don't scrub invalid chars if
      str.encode doesn't have explicit invalid: :replace.
      workaround fix for see #8995

  Modified files:
    trunk/ChangeLog
    trunk/test/ruby/test_transcode.rb
    trunk/transcode.c
Index: ChangeLog
===================================================================
--- ChangeLog	(revision 43801)
+++ ChangeLog	(revision 43802)
@@ -1,3 +1,9 @@ https://github.com/ruby/ruby/blob/trunk/ChangeLog#L1
+Fri Nov 22 17:20:50 2013  NARUSE, Yui  <naruse@r...>
+
+	* transcode.c (str_transcode0): don't scrub invalid chars if
+	  str.encode doesn't have explicit invalid: :replace.
+	  workaround fix for see #8995
+
 Fri Nov 22 17:11:26 2013  Narihiro Nakamura  <authornari@g...>
 
 	* include/ruby/intern.h, internal.h: Expose rb_gc_count().
Index: test/ruby/test_transcode.rb
===================================================================
--- test/ruby/test_transcode.rb	(revision 43801)
+++ test/ruby/test_transcode.rb	(revision 43802)
@@ -2071,4 +2071,13 @@ class TestTranscode < Test::Unit::TestCa https://github.com/ruby/ruby/blob/trunk/test/ruby/test_transcode.rb#L2071
       assert_equal(4, 'aaa'.encode(enc).length, "should count in #{enc} with BOM")
     end
   end
+
+  def test_encode_with_invalid_chars
+    bug8995 = '[ruby-dev:47747]'
+    EnvUtil.with_default_internal(Encoding::UTF_8) do
+      str = "\xff".force_encoding('utf-8')
+      assert_equal str, str.encode, bug8995
+      assert_equal "\ufffd", str.encode(invalid: :replace), bug8995
+    end
+  end
 end
Index: transcode.c
===================================================================
--- transcode.c	(revision 43801)
+++ transcode.c	(revision 43802)
@@ -2672,6 +2672,7 @@ str_transcode0(int argc, VALUE *argv, VA https://github.com/ruby/ruby/blob/trunk/transcode.c#L2672
     rb_encoding *senc, *denc;
     const char *sname, *dname;
     int dencidx;
+    int explicitly_invalid_replace = TRUE;
 
     rb_check_arity(argc, 0, 2);
 
@@ -2681,6 +2682,9 @@ str_transcode0(int argc, VALUE *argv, VA https://github.com/ruby/ruby/blob/trunk/transcode.c#L2682
 	    if (!ecflags) return -1;
 	    arg1 = rb_obj_encoding(str);
 	}
+	if (!(ecflags & ECONV_INVALID_MASK)) {
+	    explicitly_invalid_replace = FALSE;
+	}
 	ecflags |= ECONV_INVALID_REPLACE | ECONV_UNDEF_REPLACE;
     }
     else {
@@ -2694,7 +2698,7 @@ str_transcode0(int argc, VALUE *argv, VA https://github.com/ruby/ruby/blob/trunk/transcode.c#L2698
                     ECONV_XML_ATTR_CONTENT_DECORATOR|
                     ECONV_XML_ATTR_QUOTE_DECORATOR)) == 0) {
         if (senc && senc == denc) {
-	    if (ecflags & ECONV_INVALID_MASK) {
+	    if ((ecflags & ECONV_INVALID_MASK) && explicitly_invalid_replace) {
 		VALUE rep = Qnil;
 		if (!NIL_P(ecopts)) {
 		    rep = rb_hash_aref(ecopts, sym_replace);

--
ML: ruby-changes@q...
Info: http://www.atdot.net/~ko1/quickml/