Improve String::WriteAscii and add PRESERVE_ASCII_NULL option.

The PRESERVE_ASCII_NULL option prevents WriteAscii from converting
\0 to white space.

R=verwaest@chromium.org
BUG=v8:2252
TEST=test-api/StringWrite

Review URL: https://chromiumcodereview.appspot.com/10810056

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@12174 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
This commit is contained in:
yangguo@chromium.org 2012-07-23 14:22:46 +00:00
parent 3667f92cbb
commit 6b489d9753
3 changed files with 61 additions and 18 deletions

View File

@ -1064,7 +1064,8 @@ class String : public Primitive {
enum WriteOptions {
NO_OPTIONS = 0,
HINT_MANY_WRITES_EXPECTED = 1,
NO_NULL_TERMINATION = 2
NO_NULL_TERMINATION = 2,
PRESERVE_ASCII_NULL = 4
};
// 16-bit character codes.

View File

@ -3847,6 +3847,9 @@ int String::WriteUtf8(char* buffer,
LOG_API(isolate, "String::WriteUtf8");
ENTER_V8(isolate);
i::Handle<i::String> str = Utils::OpenHandle(this);
if (options & HINT_MANY_WRITES_EXPECTED) {
FlattenString(str); // Flatten the string for efficiency.
}
int string_length = str->length();
if (str->IsAsciiRepresentation()) {
int len;
@ -3903,11 +3906,7 @@ int String::WriteUtf8(char* buffer,
// Slow case.
i::StringInputBuffer& write_input_buffer = *isolate->write_input_buffer();
isolate->string_tracker()->RecordWrite(str);
if (options & HINT_MANY_WRITES_EXPECTED) {
// Flatten the string for efficiency. This applies whether we are
// using StringInputBuffer or Get(i) to access the characters.
FlattenString(str);
}
write_input_buffer.Reset(0, *str);
int len = str->length();
// Encode the first K - 3 bytes directly into the buffer since we
@ -3949,8 +3948,9 @@ int String::WriteUtf8(char* buffer,
c,
unibrow::Utf16::kNoPreviousCharacter);
if (pos + written <= capacity) {
for (int j = 0; j < written; j++)
for (int j = 0; j < written; j++) {
buffer[pos + j] = intermediate[j];
}
pos += written;
nchars++;
} else {
@ -3963,8 +3963,9 @@ 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 == -1 || pos < capacity))) {
buffer[pos++] = '\0';
}
return pos;
}
@ -3977,28 +3978,45 @@ int String::WriteAscii(char* buffer,
if (IsDeadCheck(isolate, "v8::String::WriteAscii()")) return 0;
LOG_API(isolate, "String::WriteAscii");
ENTER_V8(isolate);
i::StringInputBuffer& write_input_buffer = *isolate->write_input_buffer();
ASSERT(start >= 0 && length >= -1);
i::Handle<i::String> str = Utils::OpenHandle(this);
isolate->string_tracker()->RecordWrite(str);
if (options & HINT_MANY_WRITES_EXPECTED) {
// Flatten the string for efficiency. This applies whether we are
// using StringInputBuffer or Get(i) to access the characters.
str->TryFlatten();
FlattenString(str); // Flatten the string for efficiency.
}
if (str->IsAsciiRepresentation()) {
// WriteToFlat is faster than using the StringInputBuffer.
if (length == -1) 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)) {
for (int i = 0; i < len; i++) {
if (buffer[i] == '\0') buffer[i] = ' ';
}
}
if (!(options & NO_NULL_TERMINATION) && length > len) {
buffer[len] = '\0';
}
return len;
}
i::StringInputBuffer& write_input_buffer = *isolate->write_input_buffer();
int end = length;
if ( (length == -1) || (length > str->length() - start) )
if ((length == -1) || (length > str->length() - start)) {
end = str->length() - start;
}
if (end < 0) return 0;
write_input_buffer.Reset(start, *str);
int i;
for (i = 0; i < end; i++) {
char c = static_cast<char>(write_input_buffer.GetNext());
if (c == '\0') c = ' ';
if (c == '\0' && !(options & PRESERVE_ASCII_NULL)) c = ' ';
buffer[i] = c;
}
if (!(options & NO_NULL_TERMINATION) && (length == -1 || i < length))
if (!(options & NO_NULL_TERMINATION) && (length == -1 || i < length)) {
buffer[i] = '\0';
}
return i;
}
@ -4017,7 +4035,7 @@ int String::Write(uint16_t* buffer,
if (options & HINT_MANY_WRITES_EXPECTED) {
// Flatten the string for efficiency. This applies whether we are
// using StringInputBuffer or Get(i) to access the characters.
str->TryFlatten();
FlattenString(str);
}
int end = start + length;
if ((length == -1) || (length > str->length() - start) )
@ -4203,8 +4221,9 @@ void v8::Object::SetPointerInInternalField(int index, void* value) {
i::Handle<i::Foreign> foreign =
isolate->factory()->NewForeign(
reinterpret_cast<i::Address>(value), i::TENURED);
if (!foreign.is_null())
Utils::OpenHandle(this)->SetInternalField(index, *foreign);
if (!foreign.is_null()) {
Utils::OpenHandle(this)->SetInternalField(index, *foreign);
}
}
ASSERT_EQ(value, GetPointerFromInternalField(index));
}

View File

@ -5570,6 +5570,7 @@ THREADED_TEST(StringWrite) {
v8::Handle<String> str = v8_str("abcde");
// abc<Icelandic eth><Unicode snowman>.
v8::Handle<String> str2 = v8_str("abc\303\260\342\230\203");
v8::Handle<String> str3 = v8::String::New("abc\0def", 7);
const int kStride = 4; // Must match stride in for loops in JS below.
CompileRun(
"var left = '';"
@ -5780,6 +5781,28 @@ THREADED_TEST(StringWrite) {
CHECK_NE(0, strcmp(utf8buf, "abc\303\260\342\230\203"));
utf8buf[8] = '\0';
CHECK_EQ(0, strcmp(utf8buf, "abc\303\260\342\230\203"));
memset(utf8buf, 0x1, sizeof(utf8buf));
utf8buf[5] = 'X';
len = str->WriteUtf8(utf8buf, sizeof(utf8buf), &charlen,
String::NO_NULL_TERMINATION);
CHECK_EQ(5, len);
CHECK_EQ('X', utf8buf[5]); // Test that the sixth character is untouched.
CHECK_EQ(5, charlen);
utf8buf[5] = '\0';
CHECK_EQ(0, strcmp(utf8buf, "abcde"));
memset(buf, 0x1, sizeof(buf));
len = str3->WriteAscii(buf);
CHECK_EQ(7, len);
CHECK_EQ(0, strcmp("abc def", buf));
memset(buf, 0x1, sizeof(buf));
len = str3->WriteAscii(buf, 0, -1, String::PRESERVE_ASCII_NULL);
CHECK_EQ(7, len);
CHECK_EQ(0, strcmp("abc", buf));
CHECK_EQ(0, buf[3]);
CHECK_EQ(0, strcmp("def", buf + 4));
}