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

sdk/lib/_internal/compiler/implementation/lib/interceptors.dart

Index: sdk/lib/_internal/compiler/implementation/lib/interceptors.dart
diff --git a/sdk/lib/_internal/compiler/implementation/lib/interceptors.dart b/sdk/lib/_internal/compiler/implementation/lib/interceptors.dart
index 2fc25a6fb80f706916b405d3a7fa2ea9a99b107f..b488693453b12b583c7aae1a4a63b054099d448d 100644
--- a/sdk/lib/_internal/compiler/implementation/lib/interceptors.dart
+++ b/sdk/lib/_internal/compiler/implementation/lib/interceptors.dart
@@ -90,13 +90,28 @@ iterator(receiver) {
   return UNINTERCEPTED(receiver.iterator());
 }
 
-charCodeAt(var receiver, int index) {
+codeUnitAt(var receiver, int index) {
   if (receiver is String) {
     if (index is !num) throw new ArgumentError(index);
     if (index < 0) throw new RangeError.value(index);
     if (index >= receiver.length) throw new RangeError.value(index);
     return JS('int', r'#.charCodeAt(#)', receiver, index);
   } else {
+    return UNINTERCEPTED(receiver.codeUnitAt(index));
+  }
+}
+
+charCodeAt(var receiver, int index) {
+  if (receiver is String) {
+    if (index is !num) throw new ArgumentError(index);
+    if (index < 0) throw new RangeError.value(index);
+    if (index >= receiver.length) throw new RangeError.value(index);
+    int codeUnit = JS('int', r'#.charCodeAt(#)', receiver, index);
+    if (codeUnit < 0xd800 || codeUnit >= 0xe000) return codeUnit;
+    if (index + 1 == receiver.length) return codeUnit;
+    int codeUnit2 = JS('int', r'#.charCodeAt(#)', receiver, index + 1);
+    return 0x10000 + ((codeUnit & 0x3ff) << 10) + (codeUnit2 & 0x3ff);
+  } else {
     return UNINTERCEPTED(receiver.charCodeAt(index));
   }
 }
@@ -646,12 +661,24 @@ get$hashCode(receiver) {
   return 0x1fffffff & (hash + JS('int', r'# << #', 0x00003fff & hash, 15));
 }
 
+get$codeUnits(receiver) {
+  if (receiver is !String) return UNINTERCEPTED(receiver.codeUnits);
+  int len = receiver.length;
+  List<int> result = new List<int>(len);
+  for (int i = 0; i < len; i++) {
+    result[i] = JS('int', r'#.charCodeAt(#)', receiver, i);

[floitsch, 2012/11/08 15:28:21]

why not
result[i] = receiver.codeUnitAt(i);
?

[erikcorry, 2012/11/15 13:28:25]

Unfortunately the optimizer can't optimize that away again:

-    t1 = receiver.charCodeAt(i);
+    t1 = $.getInterceptor(receiver).codeUnitAt$1(receiver, i);

+  }
+  return result;
+}
+
 get$charCodes(receiver) {
   if (receiver is !String) return UNINTERCEPTED(receiver.charCodes);
   int len = receiver.length;
   List<int> result = new List<int>(len);
-  for (int i = 0; i < len; i++) {
-    result[i] = receiver.charCodeAt(i);
+  for (int i = 0, j = 0; i < len; i++, j++) {
+    int code = receiver.charCodeAt(i);
+    if (code >= 0x10000) i++;
+    result[j] = code;
   }
   return result;
 }