string.c: fix false coderange

* string.c (rb_enc_str_scrub): enc can differ from the actual
  encoding of the string, the cached coderange is useless then.
  [ruby-core:82674] [Bug #13874]

git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@59763 b2dd03c8-39d4-4d8f-98ff-823fe69b080e

2 files changed, 36 insertions, 9 deletions

diff --git a/string.c b/string.c
index 734184085c..0ed9489ae3 100644
--- a/string.c
+++ b/string.c
@@ -9553,6 +9553,8 @@ str_compat_and_valid(VALUE str, rb_encoding *enc)
     return str;
 }
 
+static VALUE enc_str_scrub(rb_encoding *enc, VALUE str, VALUE repl, int cr);
+
 /**
  * @param str the string to be scrubbed
  * @param repl the replacement character
@@ -9561,13 +9563,25 @@ str_compat_and_valid(VALUE str, rb_encoding *enc)
 VALUE
 rb_str_scrub(VALUE str, VALUE repl)
 {
-    return rb_enc_str_scrub(STR_ENC_GET(str), str, repl);
+    rb_encoding *enc = STR_ENC_GET(str);
+    return enc_str_scrub(enc, str, repl, ENC_CODERANGE(str));
 }
 
 VALUE
 rb_enc_str_scrub(rb_encoding *enc, VALUE str, VALUE repl)
 {
-    int cr = ENC_CODERANGE(str);
+    int cr = ENC_CODERANGE_UNKNOWN;
+    if (enc == STR_ENC_GET(str)) {
+	/* cached coderange makes sense only when enc equals the
+	 * actual encoding of str */
+	cr = ENC_CODERANGE(str);
+    }
+    return enc_str_scrub(enc, str, repl, cr);
+}
+
+static VALUE
+enc_str_scrub(rb_encoding *enc, VALUE str, VALUE repl, int cr)
+{
     int encidx;
     VALUE buf = Qnil;
     const char *rep;
diff --git a/test/ruby/test_transcode.rb b/test/ruby/test_transcode.rb
index bdd1788be8..085d84ae21 100644
--- a/test/ruby/test_transcode.rb
+++ b/test/ruby/test_transcode.rb
@@ -2180,17 +2180,19 @@ class TestTranscode < Test::Unit::TestCase
 
   def test_valid_dummy_encoding
     bug9314 = '[ruby-core:59354] [Bug #9314]'
-    assert_separately(%W[- -- #{bug9314}], <<-'end;')
-    bug = ARGV.shift
-    result = assert_nothing_raised(TypeError, bug) {break "test".encode(Encoding::UTF_16)}
-    assert_equal("\xFE\xFF\x00t\x00e\x00s\x00t", result.b, bug)
-    result = assert_nothing_raised(TypeError, bug) {break "test".encode(Encoding::UTF_32)}
-    assert_equal("\x00\x00\xFE\xFF\x00\x00\x00t\x00\x00\x00e\x00\x00\x00s\x00\x00\x00t", result.b, bug)
+    assert_separately(%W[- -- #{bug9314}], "#{<<~"begin;"}\n#{<<~'end;'}")
+    begin;
+      bug = ARGV.shift
+      result = assert_nothing_raised(TypeError, bug) {break "test".encode(Encoding::UTF_16)}
+      assert_equal("\xFE\xFF\x00t\x00e\x00s\x00t", result.b, bug)
+      result = assert_nothing_raised(TypeError, bug) {break "test".encode(Encoding::UTF_32)}
+      assert_equal("\x00\x00\xFE\xFF\x00\x00\x00t\x00\x00\x00e\x00\x00\x00s\x00\x00\x00t", result.b, bug)
     end;
   end
 
   def test_loading_race
-    assert_separately([], <<-'end;') #do
+    assert_separately([], "#{<<~"begin;"}\n#{<<~'end;'}")
+    begin;
       bug11277 = '[ruby-dev:49106] [Bug #11277]'
       num = 2
       th = (0...num).map do |i|
@@ -2207,6 +2209,17 @@ class TestTranscode < Test::Unit::TestCase
     end;
   end
 
+  def test_scrub_encode_with_coderange
+    bug = '[ruby-core:82674] [Bug #13874]'
+    s = "\xe5".b
+    u = Encoding::UTF_8
+    assert_equal("?", s.encode(u, u, invalid: :replace, replace: "?"),
+                 "should replace invalid byte")
+    assert_predicate(s, :valid_encoding?, "any char is valid in binary")
+    assert_equal("?", s.encode(u, u, invalid: :replace, replace: "?"),
+                 "#{bug} coderange should not have side effects")
+  end
+
   def test_universal_newline
     bug11324 = '[ruby-core:69841] [Bug #11324]'
     usascii = Encoding::US_ASCII