Move invalid string length check to the factory.

src/factory.cc

 // Copyright 2014 the V8 project authors. All rights reserved.
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file.
 
 #include "src/factory.h"
 
 #include "src/conversions.h"
 #include "src/isolate-inl.h"
 #include "src/macro-assembler.h"
 
@@ skipping 194 matching lines @@
 
 template Handle<String> Factory::InternalizeStringWithKey<
     SubStringKey<uint8_t> > (SubStringKey<uint8_t>* key);
 template Handle<String> Factory::InternalizeStringWithKey<
     SubStringKey<uint16_t> > (SubStringKey<uint16_t>* key);
 
 
 MaybeHandle<String> Factory::NewStringFromOneByte(Vector<const uint8_t> string,
                                                   PretenureFlag pretenure) {
   int length = string.length();
-  if (length == 1) {
-    return LookupSingleCharacterStringFromCode(string[0]);
-  }
+  if (length == 1) return LookupSingleCharacterStringFromCode(string[0]);
   Handle<SeqOneByteString> result;
   ASSIGN_RETURN_ON_EXCEPTION(
       isolate(),
       result,
       NewRawOneByteString(string.length(), pretenure),
       String);
 
   DisallowHeapAllocation no_gc;
   // Copy the characters into the new object.
   CopyChars(SeqOneByteString::cast(*result)->GetChars(),
             string.start(),
             length);
   return result;
 }
 
 MaybeHandle<String> Factory::NewStringFromUtf8(Vector<const char> string,
                                                PretenureFlag pretenure) {
   // Check for ASCII first since this is the common case.
   const char* start = string.start();
   int length = string.length();
   int non_ascii_start = String::NonAsciiStart(start, length);
   if (non_ascii_start >= length) {
     // If the string is ASCII, we do not need to convert the characters
     // since UTF8 is backwards compatible with ASCII.
     return NewStringFromOneByte(Vector<const uint8_t>::cast(string), pretenure);
   }
+
   // Non-ASCII and we need to decode.
-  CALL_HEAP_FUNCTION(
-      isolate(),
-      isolate()->heap()->AllocateStringFromUtf8Slow(string,
-                                                    non_ascii_start,
-                                                    pretenure),
+  Access<UnicodeCache::Utf8Decoder>
+      decoder(isolate()->unicode_cache()->utf8_decoder());
+  decoder->Reset(string.start() + non_ascii_start,
+                 length - non_ascii_start);
+  int utf16_length = decoder->Utf16Length();
+  ASSERT(utf16_length > 0);
+  // Allocate string.
+  Handle<SeqTwoByteString> result;
+  ASSIGN_RETURN_ON_EXCEPTION(
+      isolate(), result,
+      NewRawTwoByteString(non_ascii_start + utf16_length, pretenure),
       String);
+  // Copy ascii portion.
+  uint16_t* data = result->GetChars();
+  const char* ascii_data = string.start();
+  for (int i = 0; i < non_ascii_start; i++) {
+    *data++ = *ascii_data++;
+  }
+  // Now write the remainder.
+  decoder->WriteUtf16(data, utf16_length);
+  return result;
 }
 
 
 MaybeHandle<String> Factory::NewStringFromTwoByte(Vector<const uc16> string,
                                                   PretenureFlag pretenure) {
-  CALL_HEAP_FUNCTION(
-      isolate(),
-      isolate()->heap()->AllocateStringFromTwoByte(string, pretenure),
-      String);
+  int length = string.length();
+  const uc16* start = string.start();
+  if (String::IsOneByte(start, length)) {
+    Handle<SeqOneByteString> result;
+    ASSIGN_RETURN_ON_EXCEPTION(
+        isolate(),
+        result,
+        NewRawOneByteString(length, pretenure),
+        String);
+    CopyChars(result->GetChars(), start, length);
+    return result;
+  } else {
+    Handle<SeqTwoByteString> result;
+    ASSIGN_RETURN_ON_EXCEPTION(
+        isolate(),
+        result,
+        NewRawTwoByteString(length, pretenure),
+        String);
+    CopyChars(result->GetChars(), start, length);
+    return result;
+  }
 }
 
 
 Handle<String> Factory::NewInternalizedStringFromUtf8(Vector<const char> str,
                                                       int chars,
                                                       uint32_t hash_field) {
   CALL_HEAP_FUNCTION(
       isolate(),
       isolate()->heap()->AllocateInternalizedStringFromUtf8(
           str, chars, hash_field),
@@ skipping 51 matching lines @@
       return short_external_ascii_internalized_string_map();
     case SHORT_EXTERNAL_STRING_WITH_ONE_BYTE_DATA_TYPE:
       return short_external_internalized_string_with_one_byte_data_map();
     default: return MaybeHandle<Map>();  // No match found.
   }
 }
 
 
 MaybeHandle<SeqOneByteString> Factory::NewRawOneByteString(
     int length, PretenureFlag pretenure) {
+  if (length > String::kMaxLength || length < 0) {
+    return isolate()->Throw<SeqOneByteString>(NewInvalidStringLengthError());
+  }
   CALL_HEAP_FUNCTION(
       isolate(),
       isolate()->heap()->AllocateRawOneByteString(length, pretenure),
       SeqOneByteString);
 }
 
 
 MaybeHandle<SeqTwoByteString> Factory::NewRawTwoByteString(
     int length, PretenureFlag pretenure) {
+  if (length > String::kMaxLength || length < 0) {
+    return isolate()->Throw<SeqTwoByteString>(NewInvalidStringLengthError());
+  }
   CALL_HEAP_FUNCTION(
       isolate(),
       isolate()->heap()->AllocateRawTwoByteString(length, pretenure),
       SeqTwoByteString);
 }
 
 
 Handle<String> Factory::LookupSingleCharacterStringFromCode(uint32_t code) {
   if (code <= String::kMaxOneByteCharCodeU) {
     {
@@ skipping 229 matching lines @@
   slice->set_parent(*str);
   slice->set_offset(offset);
   return slice;
 }
 
 
 MaybeHandle<String> Factory::NewExternalStringFromAscii(
     const ExternalAsciiString::Resource* resource) {
   size_t length = resource->length();
   if (length > static_cast<size_t>(String::kMaxLength)) {
-    isolate()->ThrowInvalidStringLength();
-    return MaybeHandle<String>();
+    return isolate()->Throw<String>(NewInvalidStringLengthError());
   }
 
   Handle<Map> map = external_ascii_string_map();
   Handle<ExternalAsciiString> external_string =
       New<ExternalAsciiString>(map, NEW_SPACE);
   external_string->set_length(static_cast<int>(length));
   external_string->set_hash_field(String::kEmptyHashField);
   external_string->set_resource(resource);
 
   return external_string;
 }
 
 
 MaybeHandle<String> Factory::NewExternalStringFromTwoByte(
     const ExternalTwoByteString::Resource* resource) {
   size_t length = resource->length();
   if (length > static_cast<size_t>(String::kMaxLength)) {
-    isolate()->ThrowInvalidStringLength();
-    return MaybeHandle<String>();
+    return isolate()->Throw<String>(NewInvalidStringLengthError());
   }
 
   // For small strings we check whether the resource contains only
   // one byte characters.  If yes, we use a different string map.
   static const size_t kOneByteCheckLengthLimit = 32;
   bool is_one_byte = length <= kOneByteCheckLengthLimit &&
       String::IsOneByte(resource->data(), static_cast<int>(length));
   Handle<Map> map = is_one_byte ?
       external_string_with_one_byte_data_map() : external_string_map();
   Handle<ExternalTwoByteString> external_string =
@@ skipping 1717 matching lines @@
   return Handle<Object>::null();
 }
 
 
 Handle<Object> Factory::ToBoolean(bool value) {
   return value ? true_value() : false_value();
 }
 
 
 } }  // namespace v8::internal