Fix input and output to handle UTF16 surrogate pairs.

src/api.cc

Index: src/api.cc
===================================================================
--- src/api.cc	(revision 10944)
+++ src/api.cc	(working copy)
@@ -1429,7 +1429,7 @@
 
 
 ScriptData* ScriptData::PreCompile(const char* input, int length) {
-  i::Utf8ToUC16CharacterStream stream(
+  i::Utf8ToUtf16CharacterStream stream(
       reinterpret_cast<const unsigned char*>(input), length);
   return i::ParserApi::PreParse(&stream, NULL, i::FLAG_harmony_scoping);
 }
@@ -1438,11 +1438,11 @@
 ScriptData* ScriptData::PreCompile(v8::Handle<String> source) {
   i::Handle<i::String> str = Utils::OpenHandle(*source);
   if (str->IsExternalTwoByteString()) {
-    i::ExternalTwoByteStringUC16CharacterStream stream(
+    i::ExternalTwoByteStringUtf16CharacterStream stream(
       i::Handle<i::ExternalTwoByteString>::cast(str), 0, str->length());
     return i::ParserApi::PreParse(&stream, NULL, i::FLAG_harmony_scoping);
   } else {
-    i::GenericStringUC16CharacterStream stream(str, 0, str->length());
+    i::GenericStringUtf16CharacterStream stream(str, 0, str->length());
     return i::ParserApi::PreParse(&stream, NULL, i::FLAG_harmony_scoping);
   }
 }
@@ -3689,7 +3689,13 @@
 int String::Utf8Length() const {
   i::Handle<i::String> str = Utils::OpenHandle(this);
   if (IsDeadCheck(str->GetIsolate(), "v8::String::Utf8Length()")) return 0;
-  return str->Utf8Length();
+  int length = str->Utf8Length();
+  if (length < 0) {
+    FlattenString(str);
+    length = str->Utf8Length();
+  }
+  ASSERT(length >= 0);
+  return length;
 }
 
 
@@ -3735,11 +3741,13 @@
   int i;
   int pos = 0;
   int nchars = 0;
+  int previous = unibrow::Utf8::kNoPreviousCharacter;
   for (i = 0; i < len && (capacity == -1 || pos < fast_end); i++) {
     i::uc32 c = write_input_buffer.GetNext();
-    int written = unibrow::Utf8::Encode(buffer + pos, c);
+    int written = unibrow::Utf8::Encode(buffer + pos, c, previous);
     pos += written;
     nchars++;
+    previous = c;
   }
   if (i < len) {
     // For the last characters we need to check the length for each one
@@ -3748,16 +3756,34 @@
     char intermediate[unibrow::Utf8::kMaxEncodedSize];
     for (; i < len && pos < capacity; i++) {
       i::uc32 c = write_input_buffer.GetNext();
-      int written = unibrow::Utf8::Encode(intermediate, c);
-      if (pos + written <= capacity) {
-        for (int j = 0; j < written; j++)
-          buffer[pos + j] = intermediate[j];
+      if (unibrow::Utf16::IsTrailSurrogate(c) &&
+          previous != unibrow::Utf8::kNoPreviousCharacter &&
+          unibrow::Utf16::IsLeadSurrogate(previous)) {
+        // We can't use the intermediate buffer here because the encoding
+        // of surrogate pairs is done under assumption that you can step
+        // back and fix the UTF8 stream.  Luckily we only need space for one
+        // more byte, so there is always space.
+        ASSERT(pos < capacity);
+        int written = unibrow::Utf8::Encode(buffer + pos, c, previous);
+        ASSERT(written == 1);
         pos += written;
         nchars++;
       } else {
-        // We've reached the end of the buffer
-        break;
+        int written =
+            unibrow::Utf8::Encode(intermediate,
+                                  c,
+                                  unibrow::Utf8::kNoPreviousCharacter);
+        if (pos + written <= capacity) {
+          for (int j = 0; j < written; j++)
+            buffer[pos + j] = intermediate[j];
+          pos += written;
+          nchars++;
+        } else {
+          // We've reached the end of the buffer
+          break;
+        }
       }
+      previous = c;
     }
   }
   if (nchars_ref != NULL) *nchars_ref = nchars;