Index: src/objects.cc |
diff --git a/src/objects.cc b/src/objects.cc |
index 20553eb64bc94889c24418fab89dca2eeadf3346..e1d6da6df8518eb80212d79d311c3cfc4a76ea48 100644 |
--- a/src/objects.cc |
+++ b/src/objects.cc |
@@ -1048,8 +1048,7 @@ void String::StringShortPrint(StringStream* accumulator) { |
return; |
} |
- ConsStringIteratorOp op; |
- StringCharacterStream stream(this, &op); |
+ StringInputBuffer buf(this); |
bool truncated = false; |
if (len > kMaxShortPrintLength) { |
@@ -1058,17 +1057,17 @@ void String::StringShortPrint(StringStream* accumulator) { |
} |
bool ascii = true; |
for (int i = 0; i < len; i++) { |
- uint16_t c = stream.GetNext(); |
+ int c = buf.GetNext(); |
if (c < 32 || c >= 127) { |
ascii = false; |
} |
} |
- stream.Reset(this); |
+ buf.Reset(this); |
if (ascii) { |
accumulator->Add("<String[%u]: ", length()); |
for (int i = 0; i < len; i++) { |
- accumulator->Put(static_cast<char>(stream.GetNext())); |
+ accumulator->Put(buf.GetNext()); |
} |
accumulator->Put('>'); |
} else { |
@@ -1076,7 +1075,7 @@ void String::StringShortPrint(StringStream* accumulator) { |
// characters and that backslashes are therefore escaped. |
accumulator->Add("<String[%u]\\: ", length()); |
for (int i = 0; i < len; i++) { |
- uint16_t c = stream.GetNext(); |
+ int c = buf.GetNext(); |
if (c == '\n') { |
accumulator->Add("\\n"); |
} else if (c == '\r') { |
@@ -1086,7 +1085,7 @@ void String::StringShortPrint(StringStream* accumulator) { |
} else if (c < 32 || c > 126) { |
accumulator->Add("\\x%02x", c); |
} else { |
- accumulator->Put(static_cast<char>(c)); |
+ accumulator->Put(c); |
} |
} |
if (truncated) { |
@@ -1538,16 +1537,15 @@ MaybeObject* JSObject::AddFastPropertyUsingMap(Map* new_map, |
} |
-static bool IsIdentifier(UnicodeCache* cache, String* string) { |
+static bool IsIdentifier(UnicodeCache* cache, |
+ unibrow::CharacterStream* buffer) { |
// Checks whether the buffer contains an identifier (no escape). |
- if (string->length() == 0) return false; |
- ConsStringIteratorOp op; |
- StringCharacterStream stream(string, &op); |
- if (!cache->IsIdentifierStart(stream.GetNext())) { |
+ if (!buffer->has_more()) return false; |
+ if (!cache->IsIdentifierStart(buffer->GetNext())) { |
return false; |
} |
- while (stream.HasMore()) { |
- if (!cache->IsIdentifierPart(stream.GetNext())) { |
+ while (buffer->has_more()) { |
+ if (!cache->IsIdentifierPart(buffer->GetNext())) { |
return false; |
} |
} |
@@ -1568,7 +1566,8 @@ MaybeObject* JSObject::AddFastProperty(String* name, |
// hidden symbols) and is not a real identifier. |
// Normalize the object if it will have too many fast properties. |
Isolate* isolate = GetHeap()->isolate(); |
- if ((!IsIdentifier(isolate->unicode_cache(), name) |
+ StringInputBuffer buffer(name); |
+ if ((!IsIdentifier(isolate->unicode_cache(), &buffer) |
&& name != isolate->heap()->hidden_symbol()) || |
(map()->unused_property_fields() == 0 && |
TooManyFastProperties(properties()->length(), store_mode))) { |
@@ -6595,14 +6594,14 @@ SmartArrayPointer<char> String::ToCString(AllowNullsFlag allow_nulls, |
if (length < 0) length = kMaxInt - offset; |
// Compute the size of the UTF-8 string. Start at the specified offset. |
- Access<ConsStringIteratorOp> op( |
- heap->isolate()->objects_string_iterator()); |
- StringCharacterStream stream(this, op.value(), offset); |
+ Access<StringInputBuffer> buffer( |
+ heap->isolate()->objects_string_input_buffer()); |
+ buffer->Reset(offset, this); |
int character_position = offset; |
int utf8_bytes = 0; |
int last = unibrow::Utf16::kNoPreviousCharacter; |
- while (stream.HasMore() && character_position++ < offset + length) { |
- uint16_t character = stream.GetNext(); |
+ while (buffer->has_more() && character_position++ < offset + length) { |
+ uint16_t character = buffer->GetNext(); |
utf8_bytes += unibrow::Utf8::Length(character, last); |
last = character; |
} |
@@ -6614,12 +6613,13 @@ SmartArrayPointer<char> String::ToCString(AllowNullsFlag allow_nulls, |
char* result = NewArray<char>(utf8_bytes + 1); |
// Convert the UTF-16 string to a UTF-8 buffer. Start at the specified offset. |
- stream.Reset(this, offset); |
+ buffer->Rewind(); |
+ buffer->Seek(offset); |
character_position = offset; |
int utf8_byte_position = 0; |
last = unibrow::Utf16::kNoPreviousCharacter; |
- while (stream.HasMore() && character_position++ < offset + length) { |
- uint16_t character = stream.GetNext(); |
+ while (buffer->has_more() && character_position++ < offset + length) { |
+ uint16_t character = buffer->GetNext(); |
if (allow_nulls == DISALLOW_NULLS && character == 0) { |
character = ' '; |
} |
@@ -6671,15 +6671,15 @@ SmartArrayPointer<uc16> String::ToWideCString(RobustnessFlag robust_flag) { |
} |
Heap* heap = GetHeap(); |
- Access<ConsStringIteratorOp> op( |
- heap->isolate()->objects_string_iterator()); |
- StringCharacterStream stream(this, op.value()); |
+ Access<StringInputBuffer> buffer( |
+ heap->isolate()->objects_string_input_buffer()); |
+ buffer->Reset(this); |
uc16* result = NewArray<uc16>(length() + 1); |
int i = 0; |
- while (stream.HasMore()) { |
- uint16_t character = stream.GetNext(); |
+ while (buffer->has_more()) { |
+ uint16_t character = buffer->GetNext(); |
result[i++] = character; |
} |
result[i] = 0; |
@@ -7457,28 +7457,46 @@ void String::WriteToFlat(String* src, |
} |
+template <typename IteratorA, typename IteratorB> |
+static inline bool CompareStringContents(IteratorA* ia, IteratorB* ib) { |
+ // General slow case check. We know that the ia and ib iterators |
+ // have the same length. |
+ while (ia->has_more()) { |
+ uint32_t ca = ia->GetNext(); |
+ uint32_t cb = ib->GetNext(); |
+ ASSERT(ca <= unibrow::Utf16::kMaxNonSurrogateCharCode); |
+ ASSERT(cb <= unibrow::Utf16::kMaxNonSurrogateCharCode); |
+ if (ca != cb) |
+ return false; |
+ } |
+ return true; |
+} |
+ |
+ |
// Compares the contents of two strings by reading and comparing |
// int-sized blocks of characters. |
template <typename Char> |
-static inline bool CompareRawStringContents(const Char* const a, |
- const Char* const b, |
- int length) { |
+static inline bool CompareRawStringContents(Vector<Char> a, Vector<Char> b) { |
+ int length = a.length(); |
+ ASSERT_EQ(length, b.length()); |
+ const Char* pa = a.start(); |
+ const Char* pb = b.start(); |
int i = 0; |
#ifndef V8_HOST_CAN_READ_UNALIGNED |
// If this architecture isn't comfortable reading unaligned ints |
// then we have to check that the strings are aligned before |
// comparing them blockwise. |
const int kAlignmentMask = sizeof(uint32_t) - 1; // NOLINT |
- uint32_t pa_addr = reinterpret_cast<uint32_t>(a); |
- uint32_t pb_addr = reinterpret_cast<uint32_t>(b); |
+ uint32_t pa_addr = reinterpret_cast<uint32_t>(pa); |
+ uint32_t pb_addr = reinterpret_cast<uint32_t>(pb); |
if (((pa_addr & kAlignmentMask) | (pb_addr & kAlignmentMask)) == 0) { |
#endif |
const int kStepSize = sizeof(int) / sizeof(Char); // NOLINT |
int endpoint = length - kStepSize; |
// Compare blocks until we reach near the end of the string. |
for (; i <= endpoint; i += kStepSize) { |
- uint32_t wa = *reinterpret_cast<const uint32_t*>(a + i); |
- uint32_t wb = *reinterpret_cast<const uint32_t*>(b + i); |
+ uint32_t wa = *reinterpret_cast<const uint32_t*>(pa + i); |
+ uint32_t wb = *reinterpret_cast<const uint32_t*>(pb + i); |
if (wa != wb) { |
return false; |
} |
@@ -7496,145 +7514,25 @@ static inline bool CompareRawStringContents(const Char* const a, |
} |
-template<typename Chars1, typename Chars2> |
-class RawStringComparator : public AllStatic { |
- public: |
- static inline bool compare(const Chars1* a, const Chars2* b, int len) { |
- ASSERT(sizeof(Chars1) != sizeof(Chars2)); |
- for (int i = 0; i < len; i++) { |
- if (a[i] != b[i]) { |
- return false; |
- } |
- } |
- return true; |
- } |
-}; |
- |
- |
-template<> |
-class RawStringComparator<uint16_t, uint16_t> { |
- public: |
- static inline bool compare(const uint16_t* a, const uint16_t* b, int len) { |
- return CompareRawStringContents(a, b, len); |
- } |
-}; |
- |
- |
-template<> |
-class RawStringComparator<uint8_t, uint8_t> { |
- public: |
- static inline bool compare(const uint8_t* a, const uint8_t* b, int len) { |
- return CompareRawStringContents(a, b, len); |
- } |
-}; |
- |
- |
-class StringComparator { |
- class State { |
- public: |
- explicit inline State(ConsStringIteratorOp* op) |
- : op_(op), is_one_byte_(true), length_(0), buffer8_(NULL) {} |
- |
- inline void Init(String* string, unsigned len) { |
- op_->Reset(); |
- int32_t type = string->map()->instance_type(); |
- String::Visit(string, 0, *this, *op_, type, len); |
- } |
- |
- inline void VisitOneByteString(const uint8_t* chars, unsigned length) { |
- is_one_byte_ = true; |
- buffer8_ = chars; |
- length_ = length; |
- } |
- |
- inline void VisitTwoByteString(const uint16_t* chars, unsigned length) { |
- is_one_byte_ = false; |
- buffer16_ = chars; |
- length_ = length; |
- } |
- |
- void Advance(unsigned consumed) { |
- ASSERT(consumed <= length_); |
- // Still in buffer. |
- if (length_ != consumed) { |
- if (is_one_byte_) { |
- buffer8_ += consumed; |
- } else { |
- buffer16_ += consumed; |
- } |
- length_ -= consumed; |
- return; |
- } |
- // Advance state. |
- ASSERT(op_->HasMore()); |
- int32_t type = 0; |
- unsigned length = 0; |
- String* next = op_->ContinueOperation(&type, &length); |
- ASSERT(next != NULL); |
- ConsStringNullOp null_op; |
- String::Visit(next, 0, *this, null_op, type, length); |
- } |
- |
- ConsStringIteratorOp* const op_; |
- bool is_one_byte_; |
- unsigned length_; |
- union { |
- const uint8_t* buffer8_; |
- const uint16_t* buffer16_; |
- }; |
- DISALLOW_IMPLICIT_CONSTRUCTORS(State); |
- }; |
- |
- public: |
- inline StringComparator(ConsStringIteratorOp* op_1, |
- ConsStringIteratorOp* op_2) |
- : state_1_(op_1), |
- state_2_(op_2) { |
- } |
- |
- template<typename Chars1, typename Chars2> |
- static inline bool Equals(State* state_1, State* state_2, unsigned to_check) { |
- const Chars1* a = reinterpret_cast<const Chars1*>(state_1->buffer8_); |
- const Chars2* b = reinterpret_cast<const Chars2*>(state_2->buffer8_); |
- return RawStringComparator<Chars1, Chars2>::compare(a, b, to_check); |
- } |
- |
- bool Equals(unsigned length, String* string_1, String* string_2) { |
- ASSERT(length != 0); |
- state_1_.Init(string_1, length); |
- state_2_.Init(string_2, length); |
- while (true) { |
- unsigned to_check = Min(state_1_.length_, state_2_.length_); |
- ASSERT(to_check > 0 && to_check <= length); |
- bool is_equal; |
- if (state_1_.is_one_byte_) { |
- if (state_2_.is_one_byte_) { |
- is_equal = Equals<uint8_t, uint8_t>(&state_1_, &state_2_, to_check); |
- } else { |
- is_equal = Equals<uint8_t, uint16_t>(&state_1_, &state_2_, to_check); |
- } |
- } else { |
- if (state_2_.is_one_byte_) { |
- is_equal = Equals<uint16_t, uint8_t>(&state_1_, &state_2_, to_check); |
- } else { |
- is_equal = Equals<uint16_t, uint16_t>(&state_1_, &state_2_, to_check); |
- } |
- } |
- // Looping done. |
- if (!is_equal) return false; |
- length -= to_check; |
- // Exit condition. Strings are equal. |
- if (length == 0) return true; |
- state_1_.Advance(to_check); |
- state_2_.Advance(to_check); |
+template <typename IteratorA> |
+static inline bool CompareStringContentsPartial(Isolate* isolate, |
+ IteratorA* ia, |
+ String* b) { |
+ String::FlatContent content = b->GetFlatContent(); |
+ if (content.IsFlat()) { |
+ if (content.IsAscii()) { |
+ VectorIterator<char> ib(content.ToAsciiVector()); |
+ return CompareStringContents(ia, &ib); |
+ } else { |
+ VectorIterator<uc16> ib(content.ToUC16Vector()); |
+ return CompareStringContents(ia, &ib); |
} |
+ } else { |
+ isolate->objects_string_compare_buffer_b()->Reset(0, b); |
+ return CompareStringContents(ia, |
+ isolate->objects_string_compare_buffer_b()); |
} |
- |
- private: |
- State state_1_; |
- State state_2_; |
- DISALLOW_IMPLICIT_CONSTRUCTORS(StringComparator); |
-}; |
+} |
bool String::SlowEquals(String* other) { |
@@ -7670,19 +7568,58 @@ bool String::SlowEquals(String* other) { |
String* lhs = this->TryFlattenGetString(); |
String* rhs = other->TryFlattenGetString(); |
- // TODO(dcarney): Compare all types of flat strings with a Visitor. |
if (StringShape(lhs).IsSequentialAscii() && |
StringShape(rhs).IsSequentialAscii()) { |
const char* str1 = SeqOneByteString::cast(lhs)->GetChars(); |
const char* str2 = SeqOneByteString::cast(rhs)->GetChars(); |
- return CompareRawStringContents(str1, str2, len); |
+ return CompareRawStringContents(Vector<const char>(str1, len), |
+ Vector<const char>(str2, len)); |
} |
Isolate* isolate = GetIsolate(); |
- StringComparator comparator(isolate->objects_string_compare_iterator_a(), |
- isolate->objects_string_compare_iterator_b()); |
- |
- return comparator.Equals(static_cast<unsigned>(len), lhs, rhs); |
+ String::FlatContent lhs_content = lhs->GetFlatContent(); |
+ String::FlatContent rhs_content = rhs->GetFlatContent(); |
+ if (lhs_content.IsFlat()) { |
+ if (lhs_content.IsAscii()) { |
+ Vector<const char> vec1 = lhs_content.ToAsciiVector(); |
+ if (rhs_content.IsFlat()) { |
+ if (rhs_content.IsAscii()) { |
+ Vector<const char> vec2 = rhs_content.ToAsciiVector(); |
+ return CompareRawStringContents(vec1, vec2); |
+ } else { |
+ VectorIterator<char> buf1(vec1); |
+ VectorIterator<uc16> ib(rhs_content.ToUC16Vector()); |
+ return CompareStringContents(&buf1, &ib); |
+ } |
+ } else { |
+ VectorIterator<char> buf1(vec1); |
+ isolate->objects_string_compare_buffer_b()->Reset(0, rhs); |
+ return CompareStringContents(&buf1, |
+ isolate->objects_string_compare_buffer_b()); |
+ } |
+ } else { |
+ Vector<const uc16> vec1 = lhs_content.ToUC16Vector(); |
+ if (rhs_content.IsFlat()) { |
+ if (rhs_content.IsAscii()) { |
+ VectorIterator<uc16> buf1(vec1); |
+ VectorIterator<char> ib(rhs_content.ToAsciiVector()); |
+ return CompareStringContents(&buf1, &ib); |
+ } else { |
+ Vector<const uc16> vec2(rhs_content.ToUC16Vector()); |
+ return CompareRawStringContents(vec1, vec2); |
+ } |
+ } else { |
+ VectorIterator<uc16> buf1(vec1); |
+ isolate->objects_string_compare_buffer_b()->Reset(0, rhs); |
+ return CompareStringContents(&buf1, |
+ isolate->objects_string_compare_buffer_b()); |
+ } |
+ } |
+ } else { |
+ isolate->objects_string_compare_buffer_a()->Reset(0, lhs); |
+ return CompareStringContentsPartial(isolate, |
+ isolate->objects_string_compare_buffer_a(), rhs); |
+ } |
} |
@@ -7827,12 +7764,11 @@ uint32_t String::ComputeAndSetHash() { |
} |
-bool String::ComputeArrayIndex(uint32_t* index) { |
- int length = this->length(); |
+bool String::ComputeArrayIndex(unibrow::CharacterStream* buffer, |
+ uint32_t* index, |
+ int length) { |
if (length == 0 || length > kMaxArrayIndexSize) return false; |
- ConsStringIteratorOp op; |
- StringCharacterStream stream(this, &op); |
- uint16_t ch = stream.GetNext(); |
+ uc32 ch = buffer->GetNext(); |
// If the string begins with a '0' character, it must only consist |
// of it to be a legal array index. |
@@ -7845,8 +7781,8 @@ bool String::ComputeArrayIndex(uint32_t* index) { |
int d = ch - '0'; |
if (d < 0 || d > 9) return false; |
uint32_t result = d; |
- while (stream.HasMore()) { |
- d = stream.GetNext() - '0'; |
+ while (buffer->has_more()) { |
+ d = buffer->GetNext() - '0'; |
if (d < 0 || d > 9) return false; |
// Check that the new result is below the 32 bit limit. |
if (result > 429496729U - ((d > 5) ? 1 : 0)) return false; |
@@ -7867,7 +7803,8 @@ bool String::SlowAsArrayIndex(uint32_t* index) { |
*index = (kArrayIndexHashMask & field) >> kHashShift; |
return true; |
} else { |
- return ComputeArrayIndex(index); |
+ StringInputBuffer buffer(this); |
+ return ComputeArrayIndex(&buffer, index, length()); |
} |
} |