string.c: respect BOM

* string.c (get_encoding): respect BOM on pseudo encodings.
  [ruby-dev:47895] [Bug #9415]

git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@44606 b2dd03c8-39d4-4d8f-98ff-823fe69b080e

ChangeLog

@@ -1,3 +1,8 @@
+Wed Jan 15 14:04:33 2014  Nobuyoshi Nakada  <nobu@ruby-lang.org>
+
+	* string.c (get_encoding): respect BOM on pseudo encodings.
+	  [ruby-dev:47895] [Bug #9415]
+
 Wed Jan 15 14:03:47 2014  Nobuyoshi Nakada  <nobu@ruby-lang.org>
 
 	* string.c (get_actual_encoding): get actual encoding according to

string.c

@@ -121,7 +121,7 @@ VALUE rb_cSymbol;
 #define STR_HEAP_PTR(str)  (RSTRING(str)->as.heap.ptr)
 #define STR_HEAP_SIZE(str) (RSTRING(str)->as.heap.aux.capa + TERM_LEN(str))
 
-#define STR_ENC_GET(str) rb_enc_from_index(ENCODING_GET(str))
+#define STR_ENC_GET(str) get_encoding(str)
 
 rb_encoding *rb_enc_get_from_index(int index);
 
@@ -155,6 +155,12 @@ get_actual_encoding(const int encidx, VALUE str)
     return rb_enc_from_index(encidx);
 }
 
+static rb_encoding *
+get_encoding(VALUE str)
+{
+    return get_actual_encoding(ENCODING_GET(str), str);
+}
+
 static int fstring_cmp(VALUE a, VALUE b);
 
 static st_table* frozen_strings;

test/ruby/test_m17n.rb

@@ -229,6 +229,7 @@ class TestM17N < Test::Unit::TestCase
   STR_WITHOUT_BOM = "\u3042".freeze
   STR_WITH_BOM = "\uFEFF\u3042".freeze
   bug8940 = '[ruby-core:59757] [Bug #8940]'
+  bug9415 = '[ruby-dev:47895] [Bug #9415]'
   %w/UTF-16 UTF-32/.each do |enc|
     %w/BE LE/.each do |endian|
       bom = "\uFEFF".encode("#{enc}#{endian}").force_encoding(enc)
@@ -242,6 +243,14 @@ class TestM17N < Test::Unit::TestCase
         assert_normal_exit("#{bom.b.dump}.force_encoding('#{enc}').inspect", bug8940)
       end
 
+      define_method("test_utf_16_32_codepoints(#{enc}#{endian})") do
+        assert_equal([0xFEFF], bom.codepoints, bug9415)
+      end
+
+      define_method("test_utf_16_32_ord(#{enc}#{endian})") do
+        assert_equal(0xFEFF, bom.ord, bug9415)
+      end
+
       define_method("test_utf_16_32_inspect(#{enc}#{endian}-BOM)") do
         s = STR_WITH_BOM.encode(enc + endian)
         # When a UTF-16/32 string has a BOM,