* enc/utf_16be.c (UTF16_IS_SURROGATE_FIRST): avoid branch.

  (UTF16_IS_SURROGATE_SECOND): ditto.
  (UTF16_IS_SURROGATE): defined.
  (utf16be_mbc_enc_len): validation implemented.

* enc/utf_16le.c (UTF16_IS_SURROGATE_FIRST): avoid branch.
  (UTF16_IS_SURROGATE_SECOND): ditto.
  (UTF16_IS_SURROGATE): defined.
  (utf16le_mbc_enc_len): validation implemented.


git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@15338 b2dd03c8-39d4-4d8f-98ff-823fe69b080e

1 files changed, 25 insertions, 3 deletions

diff --git a/enc/utf_16be.c b/enc/utf_16be.c
index ab5ee1d4a3..5311ba3641 100644
--- a/enc/utf_16be.c
+++ b/enc/utf_16be.c
@@ -29,8 +29,9 @@
 
 #include "regenc.h"
 
-#define UTF16_IS_SURROGATE_FIRST(c)    (c >= 0xd8 && c <= 0xdb)
-#define UTF16_IS_SURROGATE_SECOND(c)   (c >= 0xdc && c <= 0xdf)
+#define UTF16_IS_SURROGATE_FIRST(c)    (((c) & 0xfc) == 0xd8)
+#define UTF16_IS_SURROGATE_SECOND(c)   (((c) & 0xfc) == 0xdc)
+#define UTF16_IS_SURROGATE(c)          (((c) & 0xf8) == 0xd8)
 
 static const int EncLen_UTF16[] = {
   2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
@@ -55,7 +56,28 @@ static int
 utf16be_mbc_enc_len(const UChar* p, const OnigUChar* e ARG_UNUSED,
 		    OnigEncoding enc ARG_UNUSED)
 {
-  return EncLen_UTF16[*p];
+  int byte = p[0];
+  if (!UTF16_IS_SURROGATE(byte)) {
+    if (2 <= e-p)
+      return ONIGENC_CONSTRUCT_MBCLEN_CHARFOUND(2);
+    else
+      return ONIGENC_CONSTRUCT_MBCLEN_NEEDMORE(1);
+  }
+  if (UTF16_IS_SURROGATE_FIRST(byte)) {
+    switch (e-p) {
+      case 1: return ONIGENC_CONSTRUCT_MBCLEN_NEEDMORE(3);
+      case 2: return ONIGENC_CONSTRUCT_MBCLEN_NEEDMORE(2);
+      case 3:
+        if (UTF16_IS_SURROGATE_SECOND(p[2]))
+          return ONIGENC_CONSTRUCT_MBCLEN_NEEDMORE(1);
+        break;
+      default:
+        if (UTF16_IS_SURROGATE_SECOND(p[2]))
+          return ONIGENC_CONSTRUCT_MBCLEN_CHARFOUND(4);
+        break;
+    }
+  }
+  return ONIGENC_CONSTRUCT_MBCLEN_INVALID();
 }
 
 static int