Add some support for the code-point code-unit distinction.

runtime/vm/unicode.h

Index: runtime/vm/unicode.h
diff --git a/runtime/vm/unicode.h b/runtime/vm/unicode.h
index ce572dc633e32d5e00418ffa7252670899f9ccef..fdde8e16da3dbc83faf29a84aa6bfb652499dff1 100644
--- a/runtime/vm/unicode.h
+++ b/runtime/vm/unicode.h
@@ -12,6 +12,55 @@ namespace dart {
 
 class String;
 
+class Utf16 : AllStatic {
+ public:
+  static const uint32_t kMaxCodeUnit = 0xffff;
+  static const uint32_t kMaxCodePoint = 0x10ffff;
+
+  static bool IsLeadSurrogate(uint32_t c) {
+    return c >= kLeadBase && c < kLeadEnd;
+  }
+
+  static bool IsTrailSurrogate(uint32_t c) {
+    return c >= kTrailBase && c < kTrailEnd;
+  }
+
+  static bool IsSurrogate(uint32_t c) {
+    return (c & 0xfffff800u) == 0xd800u;
+  }
+
+  static int32_t CodePointFromCodeUnits(int32_t lead, int32_t trail) {
+    return kSurrogateEncodingBase +
+           ((lead & kSurrogateMask) << 10) + (trail & kSurrogateMask);
+  }
+
+  static int32_t LeadFromCodePoint(uint32_t code_point) {
+    ASSERT(code_point >= kSurrogateEncodingBase);
+    return kLeadBase +
+           (((code_point - kSurrogateEncodingBase) >> 10) & kSurrogateMask);
+  }
+
+  static int32_t TrailFromCodePoint(uint32_t code_point) {
+    ASSERT(code_point >= kSurrogateEncodingBase);
+    return kTrailBase + (code_point & kSurrogateMask);
+  }
+
+  // Gets the 21 bit Unicode code point at the given index in a string.  If the
+  // returned value is greater than kMaxCodePoint then the next position of the
+  // string encodes a trail surrogate and should be skipped on iteration.  May
+  // return individual surrogate values if they are not part of a pair.
+  static uint32_t CodePointAt(const String& str, int index);
+
+ private:
+  static const uint32_t kLeadBase = 0xd800;
+  static const uint32_t kLeadEnd = 0xdbff;
+  static const uint32_t kTrailBase = 0xdc00;
+  static const uint32_t kTrailEnd = 0xdfff;
+  static const uint32_t kSurrogateMask = 0x3ff;
+  static const uint32_t kSurrogateEncodingBase = 0x10000;
+};
+
+
 class Utf8 : AllStatic {
  public:
   enum Type {
@@ -24,14 +73,12 @@ class Utf8 : AllStatic {
   static const intptr_t kMaxTwoByteChar   = 0x7FF;
   static const intptr_t kMaxThreeByteChar = 0xFFFF;
   static const intptr_t kMaxFourByteChar  = 0x10FFFF;
-  static const intptr_t kMaxBmpCodepoint  = 0xffff;
-  static const int32_t kLeadOffset = (0xD800 - (0x10000 >> 10));
-  static const int32_t kSurrogateOffset = (0x10000 - (0xD800 << 10) - 0xDC00);
 
-  static void ConvertUTF32ToUTF16(int32_t codepoint, uint16_t* dst);
-  static intptr_t CodePointCount(const uint8_t* utf8_array,
-                                 intptr_t array_len,
-                                 Type* type);
+  static const uint32_t kInvalidCodePoint =  0xffffffffu;

[cshapiro, 2012/11/15 20:14:51]

can we keep the casing of these values consistent?  This file mostly uses all
upper...

[erikcorry, 2012/11/15 23:47:05]

Done.

+
+  static intptr_t CodeUnitCount(const uint8_t* utf8_array,
+                                intptr_t array_len,
+                                Type* type);
 
   // Returns true if 'utf8_array' is a valid UTF-8 string.
   static bool IsValid(const uint8_t* utf8_array, intptr_t array_len);
@@ -44,7 +91,7 @@ class Utf8 : AllStatic {
 
   static intptr_t Decode(const uint8_t* utf8_array,
                          intptr_t array_len,
-                         int32_t* ch);
+                         uint32_t* ch);
 
   static bool DecodeToAscii(const uint8_t* utf8_array,
                             intptr_t array_len,