Add basic support for Latin1 strings to the API.

src/api.cc

Index: src/api.cc
diff --git a/src/api.cc b/src/api.cc
index f3f5b16914ea2c5f4a146575f8dc3f06a587b58b..c9a31d369b555d4bce3514baeebc8019141d7387 100644
--- a/src/api.cc
+++ b/src/api.cc
@@ -3857,7 +3857,7 @@ int String::WriteUtf8(char* buffer,
   int string_length = str->length();
   if (str->IsAsciiRepresentation()) {
     int len;
-    if (capacity == -1) {
+    if (capacity == kUndefinedLength) {
       capacity = str->length() + 1;
       len = string_length;
     } else {
@@ -3872,7 +3872,7 @@ int String::WriteUtf8(char* buffer,
     return len;
   }
 
-  if (capacity == -1 || capacity / 3 >= string_length) {
+  if (capacity == kUndefinedLength || capacity / 3 >= string_length) {
     int32_t previous = unibrow::Utf16::kNoPreviousCharacter;
     const int kMaxRecursion = 100;
     int utf8_bytes =
@@ -3903,7 +3903,7 @@ int String::WriteUtf8(char* buffer,
     int utf8_bytes = i::Utf8Length(str);
     if ((options & NO_NULL_TERMINATION) == 0) utf8_bytes++;
     if (utf8_bytes <= capacity) {
-      return WriteUtf8(buffer, -1, nchars_ref, options);
+      return WriteUtf8(buffer, kUndefinedLength, nchars_ref, options);
     }
   }
 
@@ -3921,7 +3921,9 @@ int String::WriteUtf8(char* buffer,
   int pos = 0;
   int nchars = 0;
   int previous = unibrow::Utf16::kNoPreviousCharacter;
-  for (i = 0; i < len && (capacity == -1 || pos < fast_end); i++) {
+  for (i = 0;
+       i < len && (capacity == kUndefinedLength || pos < fast_end);
+       i++) {
     i::uc32 c = write_input_buffer.GetNext();
     int written = unibrow::Utf8::Encode(buffer + pos, c, previous);
     pos += written;
@@ -3967,7 +3969,7 @@ int String::WriteUtf8(char* buffer,
   }
   if (nchars_ref != NULL) *nchars_ref = nchars;
   if (!(options & NO_NULL_TERMINATION) &&
-      (i == len && (capacity == -1 || pos < capacity))) {
+      (i == len && (capacity == kUndefinedLength || pos < capacity))) {
     buffer[pos++] = '\0';
   }
   return pos;
@@ -3982,7 +3984,7 @@ int String::WriteAscii(char* buffer,
   if (IsDeadCheck(isolate, "v8::String::WriteAscii()")) return 0;
   LOG_API(isolate, "String::WriteAscii");
   ENTER_V8(isolate);
-  ASSERT(start >= 0 && length >= -1);
+  ASSERT(start >= 0 && length >= kUndefinedLength);
   i::Handle<i::String> str = Utils::OpenHandle(this);
   isolate->string_tracker()->RecordWrite(str);
   if (options & HINT_MANY_WRITES_EXPECTED) {
@@ -3991,7 +3993,7 @@ int String::WriteAscii(char* buffer,
 
   if (str->IsAsciiRepresentation()) {
     // WriteToFlat is faster than using the StringInputBuffer.
-    if (length == -1) length = str->length() + 1;
+    if (length == kUndefinedLength) length = str->length() + 1;
     int len = i::Min(length, str->length() - start);
     i::String::WriteToFlat(*str, buffer, start, start + len);
     if (!(options & PRESERVE_ASCII_NULL)) {
@@ -4007,7 +4009,7 @@ int String::WriteAscii(char* buffer,
 
   i::StringInputBuffer& write_input_buffer = *isolate->write_input_buffer();
   int end = length;
-  if ((length == -1) || (length > str->length() - start)) {
+  if ((length == kUndefinedLength) || (length > str->length() - start)) {
     end = str->length() - start;
   }
   if (end < 0) return 0;
@@ -4025,6 +4027,31 @@ int String::WriteAscii(char* buffer,
 }
 
 
+int String::WriteLatin1(char* buffer,
+                        int start,
+                        int length,
+                        int options) const {
+  i::Isolate* isolate = Utils::OpenHandle(this)->GetIsolate();
+  if (IsDeadCheck(isolate, "v8::String::WriteLatin1()")) return 0;
+  LOG_API(isolate, "String::WriteLatin1");
+  ENTER_V8(isolate);
+  ASSERT(start >= 0 && length >= kUndefinedLength);
+  i::Handle<i::String> str = Utils::OpenHandle(this);
+  isolate->string_tracker()->RecordWrite(str);
+  if (options & HINT_MANY_WRITES_EXPECTED) {
+    FlattenString(str);  // Flatten the string for efficiency.
+  }
+
+  if (length == kUndefinedLength) length = str->length() + 1;
+  int len = i::Min(length, str->length() - start);
+  i::String::WriteToFlat(*str, buffer, start, start + len);
+  if (!(options & NO_NULL_TERMINATION) && length > len) {
+    buffer[len] = '\0';
+  }
+  return len;
+}
+
+
 int String::Write(uint16_t* buffer,
                   int start,
                   int length,
@@ -4033,7 +4060,7 @@ int String::Write(uint16_t* buffer,
   if (IsDeadCheck(isolate, "v8::String::Write()")) return 0;
   LOG_API(isolate, "String::Write");
   ENTER_V8(isolate);
-  ASSERT(start >= 0 && length >= -1);
+  ASSERT(start >= 0 && length >= kUndefinedLength);
   i::Handle<i::String> str = Utils::OpenHandle(this);
   isolate->string_tracker()->RecordWrite(str);
   if (options & HINT_MANY_WRITES_EXPECTED) {
@@ -4042,7 +4069,7 @@ int String::Write(uint16_t* buffer,
     FlattenString(str);
   }
   int end = start + length;
-  if ((length == -1) || (length > str->length() - start) )
+  if ((length == kUndefinedLength) || (length > str->length() - start) )
     end = str->length();
   if (end < 0) return 0;
   i::String::WriteToFlat(*str, buffer, start, end);
@@ -4073,18 +4100,43 @@ bool v8::String::IsExternalAscii() const {
 }
 
 
-void v8::String::VerifyExternalStringResource(
-    v8::String::ExternalStringResource* value) const {
+void v8::String::VerifyExternalStringEncoding(int encoding) const {
+  typedef internal::Internals I;
+  i::Handle<i::String> str = Utils::OpenHandle(this);
+  switch (encoding) {
+    case UTF16_ENCODING | STRICT_ASCII_HINT:
+      CHECK(str->HasOnlyAsciiChars());
+      // Fall through
+    case UTF16_ENCODING | NOT_ASCII_HINT :
+      CHECK(str->IsExternalTwoByteString());
+      break;
+    case LATIN1_ENCODING | STRICT_ASCII_HINT:
+    CHECK(str->IsExternalAsciiString());
+      break;
+    default:
+      CHECK_EQ(INVALID_ENCODING, encoding);
+      CHECK(!str->IsExternalString());
+      break;
+  }
+}
+
+
+void v8::String::VerifyExternalStringResourceBase(
+    v8::String::ExternalStringResourceBase* value) const {
   i::Handle<i::String> str = Utils::OpenHandle(this);
-  const v8::String::ExternalStringResource* expected;
+  i::StringShape shape(*str);
+  const void* expected;
+  // We expect an external string at this point since GetExternalStringEncoding
+  // should have already been called to rule out non-external strings.
   if (i::StringShape(*str).IsExternalTwoByte()) {
-    const void* resource =
-        i::Handle<i::ExternalTwoByteString>::cast(str)->resource();
-    expected = reinterpret_cast<const ExternalStringResource*>(resource);
+    expected = i::ExternalTwoByteString::cast(*str)->resource();
   } else {
-    expected = NULL;
+    ASSERT(i::StringShape(*str).IsExternalAscii());
+    expected = i::ExternalAsciiString::cast(*str)->resource();
   }
-  CHECK_EQ(expected, value);
+
+  CHECK_EQ(expected,
+           reinterpret_cast<const ExternalStringResourceBase*>(value));
 }
 
 
@@ -4692,17 +4744,44 @@ Local<String> v8::String::Empty() {
 }
 
 
-Local<String> v8::String::New(const char* data, int length) {
+static i::Handle<i::String> NewOneByteEncodedString(
+    i::Factory* factory, const char* data, int length, int encoding) {
+  if (length == String::kUndefinedLength) length = i::StrLength(data);
+  typedef v8::String S;
+
+  static const int kAsciiHintShift = 16;
+  ASSERT(IS_POWER_OF_TWO(encoding & S::kAsciiHintMask));
+  i::String::AsciiHint ascii_hint =
+      static_cast<i::String::AsciiHint>(encoding >> kAsciiHintShift);
+  STATIC_ASSERT(i::String::MAYBE_ASCII == 0);
+  STATIC_ASSERT(i::String::NOT_ASCII ==
+                (v8::String::NOT_ASCII_HINT >> kAsciiHintShift));
+  STATIC_ASSERT(i::String::STRICT_ASCII ==
+                (v8::String::STRICT_ASCII_HINT >> kAsciiHintShift));
+
+  int masked_encoding = encoding & S::kStringEncodingMask;
+
+  if (masked_encoding == S::UTF8_ENCODING) {
+    return factory->NewStringFromUtf8(
+               i::Vector<const char>(data, length), i::NOT_TENURED, ascii_hint);
+  } else if (masked_encoding == S::LATIN1_ENCODING) {
+    return factory->NewStringFromLatin1(
+               i::Vector<const char>(data, length), i::NOT_TENURED, ascii_hint);
+  } else {  // Wrong encoding.
+    return i::Handle<i::String>();
+  }
+}
+
+
+Local<String> v8::String::New(
+    const char* data, int length, int encoding) {
   i::Isolate* isolate = i::Isolate::Current();
   EnsureInitializedForIsolate(isolate, "v8::String::New()");
   LOG_API(isolate, "String::New(char)");
   if (length == 0) return Empty();
   ENTER_V8(isolate);
-  if (length == -1) length = i::StrLength(data);
-  i::Handle<i::String> result =
-      isolate->factory()->NewStringFromUtf8(
-          i::Vector<const char>(data, length));
-  return Utils::ToLocal(result);
+  return Utils::ToLocal(
+      NewOneByteEncodedString(isolate->factory(), data, length, encoding));
 }
 
 
@@ -4719,15 +4798,14 @@ Local<String> v8::String::Concat(Handle<String> left, Handle<String> right) {
 }
 
 
-Local<String> v8::String::NewUndetectable(const char* data, int length) {
+Local<String> v8::String::NewUndetectable(
+    const char* data, int length, int encoding) {
   i::Isolate* isolate = i::Isolate::Current();
   EnsureInitializedForIsolate(isolate, "v8::String::NewUndetectable()");
   LOG_API(isolate, "String::NewUndetectable(char)");
   ENTER_V8(isolate);
-  if (length == -1) length = i::StrLength(data);
   i::Handle<i::String> result =
-      isolate->factory()->NewStringFromUtf8(
-          i::Vector<const char>(data, length));
+      NewOneByteEncodedString(isolate->factory(), data, length, encoding);
   result->MarkAsUndetectable();
   return Utils::ToLocal(result);
 }
@@ -4746,7 +4824,7 @@ Local<String> v8::String::New(const uint16_t* data, int length) {
   LOG_API(isolate, "String::New(uint16_)");
   if (length == 0) return Empty();
   ENTER_V8(isolate);
-  if (length == -1) length = TwoByteStringLength(data);
+  if (length == kUndefinedLength) length = TwoByteStringLength(data);
   i::Handle<i::String> result =
       isolate->factory()->NewStringFromTwoByte(
           i::Vector<const uint16_t>(data, length));
@@ -4759,7 +4837,7 @@ Local<String> v8::String::NewUndetectable(const uint16_t* data, int length) {
   EnsureInitializedForIsolate(isolate, "v8::String::NewUndetectable()");
   LOG_API(isolate, "String::NewUndetectable(uint16_)");
   ENTER_V8(isolate);
-  if (length == -1) length = TwoByteStringLength(data);
+  if (length == kUndefinedLength) length = TwoByteStringLength(data);
   i::Handle<i::String> result =
       isolate->factory()->NewStringFromTwoByte(
           i::Vector<const uint16_t>(data, length));
@@ -4796,28 +4874,54 @@ Local<String> v8::String::NewExternal(
 }
 
 
-bool v8::String::MakeExternal(v8::String::ExternalStringResource* resource) {
-  i::Handle<i::String> obj = Utils::OpenHandle(this);
-  i::Isolate* isolate = obj->GetIsolate();
+template<class StringResourceType>
+static bool MakeStringExternal(
+    i::Handle<i::String> string, StringResourceType* resource) {
+  i::Isolate* isolate = string->GetIsolate();
   if (IsDeadCheck(isolate, "v8::String::MakeExternal()")) return false;
-  if (i::StringShape(*obj).IsExternalTwoByte()) {
+  if (i::StringShape(*string).IsExternal()) {
     return false;  // Already an external string.
   }
   ENTER_V8(isolate);
-  if (isolate->string_tracker()->IsFreshUnusedString(obj)) {
+  if (isolate->string_tracker()->IsFreshUnusedString(string)) {
     return false;
   }
   if (isolate->heap()->IsInGCPostProcessing()) {
     return false;
   }
-  bool result = obj->MakeExternal(resource);
-  if (result && !obj->IsSymbol()) {
-    isolate->heap()->external_string_table()->AddString(*obj);
+  bool result = string->MakeExternal(resource);
+  if (result && !string->IsSymbol()) {
+    isolate->heap()->external_string_table()->AddString(*string);
   }
   return result;
 }
 
 
+bool v8::String::MakeExternal(ExternalStringResource* resource) {
+  i::Handle<i::String> obj = Utils::OpenHandle(this);
+  return MakeStringExternal(obj, resource);
+}
+
+
+bool v8::String::MakeExternal(ExternalAsciiStringResource* resource) {
+  i::Handle<i::String> obj = Utils::OpenHandle(this);
+  return MakeStringExternal(obj, resource);
+}
+
+
+bool v8::String::MakeExternal(ExternalLatin1StringResource* resource) {

[Erik Corry, 2012/08/10 11:28:51]

I don't see why we need this.

[Yang, 2012/08/10 13:14:45]

Explained in v8.h

+  i::Handle<i::String> obj = Utils::OpenHandle(this);
+  if (!obj->HasOnlyAsciiChars()) {
+    // External Latin1 strings are internally represented as external ASCII
+    // strings.  We need to make sure that the string only contains ASCII
+    // characters.
+    return false;
+  }
+  return MakeStringExternal(
+             obj, static_cast<ExternalAsciiStringResource*>(resource));
+}
+
+
 Local<String> v8::String::NewExternal(
       v8::String::ExternalAsciiStringResource* resource) {
   i::Isolate* isolate = i::Isolate::Current();
@@ -4830,26 +4934,36 @@ Local<String> v8::String::NewExternal(
 }
 
 
-bool v8::String::MakeExternal(
-    v8::String::ExternalAsciiStringResource* resource) {
-  i::Handle<i::String> obj = Utils::OpenHandle(this);
-  i::Isolate* isolate = obj->GetIsolate();
-  if (IsDeadCheck(isolate, "v8::String::MakeExternal()")) return false;
-  if (i::StringShape(*obj).IsExternalTwoByte()) {
-    return false;  // Already an external string.
-  }
+Local<String> v8::String::NewExternal(ExternalLatin1StringResource* resource,
+                                      int encoding) {
+  typedef v8::internal::Internals I;
+  i::Isolate* isolate = i::Isolate::Current();
+  EnsureInitializedForIsolate(isolate, "v8::String::NewExternal()");
+  LOG_API(isolate, "String::NewExternal");
   ENTER_V8(isolate);
-  if (isolate->string_tracker()->IsFreshUnusedString(obj)) {
-    return false;
-  }
-  if (isolate->heap()->IsInGCPostProcessing()) {
-    return false;
-  }
-  bool result = obj->MakeExternal(resource);
-  if (result && !obj->IsSymbol()) {
-    isolate->heap()->external_string_table()->AddString(*obj);
+  ASSERT((encoding & kStringEncodingMask) == LATIN1_ENCODING);
+  bool ascii_hint = (encoding & kAsciiHintMask);
+  i::Handle<i::String> result;
+
+  if (ascii_hint == STRICT_ASCII_HINT ||
+      (ascii_hint != NOT_ASCII_HINT &&
+      i::String::IsAscii(resource->data(), resource->length()))) {
+    // Assert that the ascii hint is correct.
+    ASSERT(ascii_hint != STRICT_ASCII_HINT ||
+           i::String::IsAscii(resource->data(), resource->length()));
+    result = NewExternalAsciiStringHandle(isolate, resource);
+    isolate->heap()->external_string_table()->AddString(*result);
+  } else {
+    // We cannot simply take the backing store and use it as an ASCII string,
+    // since it's not.  Instead, we convert it to an internal string and dispose
+    // the external resource.
+    result = isolate->factory()->NewStringFromLatin1(
+        i::Vector<const char>(resource->data(), resource->length()),
+        i::NOT_TENURED,
+        i::String::NOT_ASCII);
+    resource->Dispose();
   }
-  return result;
+  return Utils::ToLocal(result);
 }
 
 
@@ -5112,14 +5226,28 @@ Local<Object> Array::CloneElementAt(uint32_t index) {
 }
 
 
-Local<String> v8::String::NewSymbol(const char* data, int length) {
+Local<String> v8::String::NewSymbol(
+    const char* data, int length, int encoding) {
   i::Isolate* isolate = i::Isolate::Current();
   EnsureInitializedForIsolate(isolate, "v8::String::NewSymbol()");
   LOG_API(isolate, "String::NewSymbol(char)");
   ENTER_V8(isolate);
-  if (length == -1) length = i::StrLength(data);
-  i::Handle<i::String> result =
-      isolate->factory()->LookupSymbol(i::Vector<const char>(data, length));
+  if (length == kUndefinedLength) length = i::StrLength(data);
+  i::Handle<i::String> result;
+
+  ASSERT(IS_POWER_OF_TWO(encoding & kAsciiHintMask));
+  if (((encoding & kStringEncodingMask) == LATIN1_ENCODING) &&
+      ((encoding & kAsciiHintMask) == NOT_ASCII_HINT ||
+       !i::String::IsAscii(data, length))) {
+    result = isolate->factory()->NewStringFromLatin1(
+                 i::Vector<const char>(data, length),
+                 i::NOT_TENURED,
+                 i::String::NOT_ASCII);
+    result = isolate->factory()->LookupSymbol(result);
+  } else {  // We can handle UTF8 and ASCII strings here.
+    result =
+        isolate->factory()->LookupSymbol(i::Vector<const char>(data, length));
+  }
   return Utils::ToLocal(result);
 }