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

runtime/lib/string.cc

Index: runtime/lib/string.cc
diff --git a/runtime/lib/string.cc b/runtime/lib/string.cc
index 7707ed63abcb199f12128f13646516d188b0a0dc..f3bb5af9e946f504dec8a6c1dae6c2f44b05d08f 100644
--- a/runtime/lib/string.cc
+++ b/runtime/lib/string.cc
@@ -11,7 +11,7 @@
 
 namespace dart {
 
-DEFINE_NATIVE_ENTRY(StringBase_createFromCodePoints, 1) {
+DEFINE_NATIVE_ENTRY(StringBase_createFromCodeUnits, 1) {

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

I might have missed something, but it seemed clear to me that we would not
introduce the name "code units" in the API as it is inherently ambiguous.

This should be called "createFromUTF16" or some such thing.

Moreover, I would have expected this function (createFromCodePoints) to have
been renamed to createFromUTF32.

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

Done.
No, this one creates from UTF16 code units.

   GET_NATIVE_ARGUMENT(Array, a, arguments->At(0));
   // TODO(srdjan): Check that parameterized type is an int.
   Zone* zone = isolate->current_zone();
@@ -19,7 +19,6 @@ DEFINE_NATIVE_ENTRY(StringBase_createFromCodePoints, 1) {
 
   // Unbox the array and determine the maximum element width.
   bool is_one_byte_string = true;
-  intptr_t utf16_len = array_len;
   uint32_t* utf32_array = zone->Alloc<uint32_t>(array_len);
   Object& index_object = Object::Handle(isolate);
   for (intptr_t i = 0; i < array_len; i++) {
@@ -29,24 +28,21 @@ DEFINE_NATIVE_ENTRY(StringBase_createFromCodePoints, 1) {
       args.Add(&index_object);
       Exceptions::ThrowByType(Exceptions::kArgument, args);
     }
-    intptr_t value = Smi::Cast(index_object).Value();
-    if (value < 0) {
+    uintptr_t value = Smi::Cast(index_object).Value();
+    if (value > Utf16::kMaxCodeUnit) {
       GrowableArray<const Object*> args;
       Exceptions::ThrowByType(Exceptions::kArgument, args);
     } else {
       if (value > 0x7F) {
         is_one_byte_string = false;
       }
-      if (value > 0xFFFF) {
-        utf16_len += 1;
-      }
     }
     utf32_array[i] = value;
   }
   if (is_one_byte_string) {
     return OneByteString::New(utf32_array, array_len, Heap::kNew);
   }
-  return TwoByteString::New(utf16_len, utf32_array, array_len, Heap::kNew);
+  return TwoByteString::New(array_len, utf32_array, array_len, Heap::kNew);
 }
 
 
@@ -76,7 +72,7 @@ DEFINE_NATIVE_ENTRY(String_getLength, 1) {
 }
 
 
-static int32_t StringValueAt(const String& str, const Integer& index) {
+static int32_t StringCodeUnitAt(const String& str, const Integer& index) {
   if (index.IsSmi()) {
     Smi& smi = Smi::Handle();
     smi ^= index.raw();
@@ -86,7 +82,7 @@ static int32_t StringValueAt(const String& str, const Integer& index) {
       arguments.Add(&smi);
       Exceptions::ThrowByType(Exceptions::kRange, arguments);
     }
-    return str.CharAt(index);
+    return str.CodeUnitAt(index);
   } else {
     // An index larger than Smi is always illegal.
     GrowableArray<const Object*> arguments;
@@ -100,17 +96,21 @@ static int32_t StringValueAt(const String& str, const Integer& index) {
 DEFINE_NATIVE_ENTRY(String_charAt, 2) {
   const String& receiver = String::CheckedHandle(arguments->At(0));
   GET_NATIVE_ARGUMENT(Integer, index, arguments->At(1));
-  uint32_t value = StringValueAt(receiver, index);
-  ASSERT(value <= 0x10FFFF);

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

I am not sure why this ASSERT was removed.  It is still relevant in the new
version.

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

In the new version the value is always a UTF-16 code unit.  I have inserted an
assert that it is always <= 0xffff.

-  return Symbols::New(&value, 1);
+  uint32_t value = StringCodeUnitAt(receiver, index);
+  // The VM does not GC interned strings, so we only create max 128 of them.
+  if (value <= 0x7f) {

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

Can we name this constant?

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

kFrequentCharacterLimit.

+    return Symbols::New(&value, 1);
+  } else {
+    return String::New(&value, 1);
+  }
 }
 
-DEFINE_NATIVE_ENTRY(String_charCodeAt, 2) {
+
+DEFINE_NATIVE_ENTRY(String_codeUnitAt, 2) {
   const String& receiver = String::CheckedHandle(arguments->At(0));
   GET_NATIVE_ARGUMENT(Integer, index, arguments->At(1));
-  int32_t value = StringValueAt(receiver, index);
-  ASSERT(value >= 0);
-  ASSERT(value <= 0x10FFFF);
+  uint32_t value = StringCodeUnitAt(receiver, index);
+  ASSERT(value <= Utf16::kMaxCodeUnit);
   return Smi::New(value);
 }