commit 3ef787dbeca8a5fb1086949cda830dccee07bfbd
author Ben Murdoch <benm@google.com> Thu Apr 12 10:51:47 2012 +0100
committer Ben Murdoch <benm@google.com> Mon Apr 16 16:41:38 2012 +0100
tree 0a22edd97aa148abffdd405c585b22213fccbc82
parent 85b71799222b55eb5dd74ea26efe0c64ab655c8c

Merge V8 at 3.9.24.13

Bug: 5688872
Change-Id: Id0aa8d23375030494d3189c31774059c0f5398fc

src/unicode-inl.h

diff --git a/src/unicode-inl.h b/src/unicode-inl.h
index c0649d7..9c0ebf9 100644
--- a/src/unicode-inl.h
+++ b/src/unicode-inl.h
@@ -78,7 +78,7 @@
 }
 
 
-unsigned Utf8::Encode(char* str, uchar c) {
+unsigned Utf8::Encode(char* str, uchar c, int previous) {
   static const int kMask = ~(1 << 6);
   if (c <= kMaxOneByteChar) {
     str[0] = c;
@@ -88,6 +88,13 @@
     str[1] = 0x80 | (c & kMask);
     return 2;
   } else if (c <= kMaxThreeByteChar) {
+    if (Utf16::IsTrailSurrogate(c) &&
+        Utf16::IsLeadSurrogate(previous)) {
+      const int kUnmatchedSize = kSizeOfUnmatchedSurrogate;
+      return Encode(str - kUnmatchedSize,
+                    Utf16::CombineSurrogatePair(previous, c),
+                    Utf16::kNoPreviousCharacter) - kUnmatchedSize;
+    }
     str[0] = 0xE0 | (c >> 12);
     str[1] = 0x80 | ((c >> 6) & kMask);
     str[2] = 0x80 | (c & kMask);
@@ -113,12 +120,16 @@
   return CalculateValue(bytes, length, cursor);
 }
 
-unsigned Utf8::Length(uchar c) {
+unsigned Utf8::Length(uchar c, int previous) {
   if (c <= kMaxOneByteChar) {
     return 1;
   } else if (c <= kMaxTwoByteChar) {
     return 2;
   } else if (c <= kMaxThreeByteChar) {
+    if (Utf16::IsTrailSurrogate(c) &&
+        Utf16::IsLeadSurrogate(previous)) {
+      return kSizeOfUnmatchedSurrogate - kBytesSavedByCombiningSurrogates;
+    }
     return 3;
   } else {
     return 4;