diff options
Diffstat (limited to 'deps/v8/src/runtime/runtime-i18n.cc')
-rw-r--r-- | deps/v8/src/runtime/runtime-i18n.cc | 412 |
1 files changed, 34 insertions, 378 deletions
diff --git a/deps/v8/src/runtime/runtime-i18n.cc b/deps/v8/src/runtime/runtime-i18n.cc index e89175a37d..80c8a9cd01 100644 --- a/deps/v8/src/runtime/runtime-i18n.cc +++ b/deps/v8/src/runtime/runtime-i18n.cc @@ -15,7 +15,6 @@ #include "src/i18n.h" #include "src/isolate-inl.h" #include "src/messages.h" -#include "src/string-case.h" #include "src/utils.h" #include "unicode/brkiter.h" @@ -30,7 +29,6 @@ #include "unicode/fieldpos.h" #include "unicode/fpositer.h" #include "unicode/locid.h" -#include "unicode/normalizer2.h" #include "unicode/numfmt.h" #include "unicode/numsys.h" #include "unicode/rbbi.h" @@ -49,24 +47,6 @@ namespace v8 { namespace internal { -namespace { - -const UChar* GetUCharBufferFromFlat(const String::FlatContent& flat, - std::unique_ptr<uc16[]>* dest, - int32_t length) { - DCHECK(flat.IsFlat()); - if (flat.IsOneByte()) { - if (!*dest) { - dest->reset(NewArray<uc16>(length)); - CopyChars(dest->get(), flat.ToOneByteVector().start(), length); - } - return reinterpret_cast<const UChar*>(dest->get()); - } else { - return reinterpret_cast<const UChar*>(flat.ToUC16Vector().start()); - } -} - -} // namespace // ECMA 402 6.2.3 RUNTIME_FUNCTION(Runtime_CanonicalizeLanguageTag) { @@ -322,13 +302,13 @@ RUNTIME_FUNCTION(Runtime_CreateDateTimeFormat) { ASSIGN_RETURN_FAILURE_ON_EXCEPTION(isolate, local_object, JSObject::New(constructor, constructor)); - // Set date time formatter as internal field of the resulting JS object. + // Set date time formatter as embedder field of the resulting JS object. icu::SimpleDateFormat* date_format = DateFormat::InitializeDateTimeFormat(isolate, locale, options, resolved); if (!date_format) return isolate->ThrowIllegalOperation(); - local_object->SetInternalField(0, reinterpret_cast<Smi*>(date_format)); + local_object->SetEmbedderField(0, reinterpret_cast<Smi*>(date_format)); // Make object handle weak so we can delete the data format once GC kicks in. Handle<Object> wrapper = isolate->global_handles()->Create(*local_object); @@ -512,13 +492,13 @@ RUNTIME_FUNCTION(Runtime_CreateNumberFormat) { ASSIGN_RETURN_FAILURE_ON_EXCEPTION(isolate, local_object, JSObject::New(constructor, constructor)); - // Set number formatter as internal field of the resulting JS object. + // Set number formatter as embedder field of the resulting JS object. icu::DecimalFormat* number_format = NumberFormat::InitializeNumberFormat(isolate, locale, options, resolved); if (!number_format) return isolate->ThrowIllegalOperation(); - local_object->SetInternalField(0, reinterpret_cast<Smi*>(number_format)); + local_object->SetEmbedderField(0, reinterpret_cast<Smi*>(number_format)); Handle<Object> wrapper = isolate->global_handles()->Create(*local_object); GlobalHandles::MakeWeak(wrapper.location(), wrapper.location(), @@ -552,6 +532,29 @@ RUNTIME_FUNCTION(Runtime_InternalNumberFormat) { result.length()))); } +RUNTIME_FUNCTION(Runtime_CurrencyDigits) { + DCHECK_EQ(1, args.length()); + + CONVERT_ARG_HANDLE_CHECKED(String, currency, 0); + + // TODO(littledan): Avoid transcoding the string twice + v8::String::Utf8Value currency_string(v8::Utils::ToLocal(currency)); + icu::UnicodeString currency_icu = + icu::UnicodeString::fromUTF8(*currency_string); + + DisallowHeapAllocation no_gc; + UErrorCode status = U_ZERO_ERROR; +#if U_ICU_VERSION_MAJOR_NUM >= 59 + uint32_t fraction_digits = ucurr_getDefaultFractionDigits( + icu::toUCharPtr(currency_icu.getTerminatedBuffer()), &status); +#else + uint32_t fraction_digits = ucurr_getDefaultFractionDigits( + currency_icu.getTerminatedBuffer(), &status); +#endif + // For missing currency codes, default to the most common, 2 + if (!U_SUCCESS(status)) fraction_digits = 2; + return Smi::FromInt(fraction_digits); +} RUNTIME_FUNCTION(Runtime_CreateCollator) { HandleScope scope(isolate); @@ -569,13 +572,13 @@ RUNTIME_FUNCTION(Runtime_CreateCollator) { ASSIGN_RETURN_FAILURE_ON_EXCEPTION(isolate, local_object, JSObject::New(constructor, constructor)); - // Set collator as internal field of the resulting JS object. + // Set collator as embedder field of the resulting JS object. icu::Collator* collator = Collator::InitializeCollator(isolate, locale, options, resolved); if (!collator) return isolate->ThrowIllegalOperation(); - local_object->SetInternalField(0, reinterpret_cast<Smi*>(collator)); + local_object->SetEmbedderField(0, reinterpret_cast<Smi*>(collator)); Handle<Object> wrapper = isolate->global_handles()->Create(*local_object); GlobalHandles::MakeWeak(wrapper.location(), wrapper.location(), @@ -622,64 +625,6 @@ RUNTIME_FUNCTION(Runtime_InternalCompare) { } -RUNTIME_FUNCTION(Runtime_StringNormalize) { - HandleScope scope(isolate); - static const struct { - const char* name; - UNormalization2Mode mode; - } normalizationForms[] = { - {"nfc", UNORM2_COMPOSE}, - {"nfc", UNORM2_DECOMPOSE}, - {"nfkc", UNORM2_COMPOSE}, - {"nfkc", UNORM2_DECOMPOSE}, - }; - - DCHECK_EQ(2, args.length()); - - CONVERT_ARG_HANDLE_CHECKED(String, s, 0); - CONVERT_NUMBER_CHECKED(int, form_id, Int32, args[1]); - CHECK(form_id >= 0 && - static_cast<size_t>(form_id) < arraysize(normalizationForms)); - - int length = s->length(); - s = String::Flatten(s); - icu::UnicodeString result; - std::unique_ptr<uc16[]> sap; - UErrorCode status = U_ZERO_ERROR; - { - DisallowHeapAllocation no_gc; - String::FlatContent flat = s->GetFlatContent(); - const UChar* src = GetUCharBufferFromFlat(flat, &sap, length); - icu::UnicodeString input(false, src, length); - // Getting a singleton. Should not free it. - const icu::Normalizer2* normalizer = - icu::Normalizer2::getInstance(nullptr, normalizationForms[form_id].name, - normalizationForms[form_id].mode, status); - DCHECK(U_SUCCESS(status)); - CHECK(normalizer != nullptr); - int32_t normalized_prefix_length = - normalizer->spanQuickCheckYes(input, status); - // Quick return if the input is already normalized. - if (length == normalized_prefix_length) return *s; - icu::UnicodeString unnormalized = - input.tempSubString(normalized_prefix_length); - // Read-only alias of the normalized prefix. - result.setTo(false, input.getBuffer(), normalized_prefix_length); - // copy-on-write; normalize the suffix and append to |result|. - normalizer->normalizeSecondAndAppend(result, unnormalized, status); - } - - if (U_FAILURE(status)) { - return isolate->heap()->undefined_value(); - } - - RETURN_RESULT_OR_FAILURE( - isolate, isolate->factory()->NewStringFromTwoByte(Vector<const uint16_t>( - reinterpret_cast<const uint16_t*>(result.getBuffer()), - result.length()))); -} - - RUNTIME_FUNCTION(Runtime_CreateBreakIterator) { HandleScope scope(isolate); @@ -696,15 +641,15 @@ RUNTIME_FUNCTION(Runtime_CreateBreakIterator) { ASSIGN_RETURN_FAILURE_ON_EXCEPTION(isolate, local_object, JSObject::New(constructor, constructor)); - // Set break iterator as internal field of the resulting JS object. + // Set break iterator as embedder field of the resulting JS object. icu::BreakIterator* break_iterator = V8BreakIterator::InitializeBreakIterator( isolate, locale, options, resolved); if (!break_iterator) return isolate->ThrowIllegalOperation(); - local_object->SetInternalField(0, reinterpret_cast<Smi*>(break_iterator)); + local_object->SetEmbedderField(0, reinterpret_cast<Smi*>(break_iterator)); // Make sure that the pointer to adopted text is NULL. - local_object->SetInternalField(1, static_cast<Smi*>(nullptr)); + local_object->SetEmbedderField(1, static_cast<Smi*>(nullptr)); // Make object handle weak so we can delete the break iterator once GC kicks // in. @@ -729,7 +674,7 @@ RUNTIME_FUNCTION(Runtime_BreakIteratorAdoptText) { CHECK_NOT_NULL(break_iterator); icu::UnicodeString* u_text = reinterpret_cast<icu::UnicodeString*>( - break_iterator_holder->GetInternalField(1)); + break_iterator_holder->GetEmbedderField(1)); delete u_text; int length = text->length(); @@ -739,7 +684,7 @@ RUNTIME_FUNCTION(Runtime_BreakIteratorAdoptText) { std::unique_ptr<uc16[]> sap; const UChar* text_value = GetUCharBufferFromFlat(flat, &sap, length); u_text = new icu::UnicodeString(text_value, length); - break_iterator_holder->SetInternalField(1, reinterpret_cast<Smi*>(u_text)); + break_iterator_holder->SetEmbedderField(1, reinterpret_cast<Smi*>(u_text)); break_iterator->setText(*u_text); @@ -823,295 +768,6 @@ RUNTIME_FUNCTION(Runtime_BreakIteratorBreakType) { } } -namespace { -MUST_USE_RESULT Object* LocaleConvertCase(Handle<String> s, Isolate* isolate, - bool is_to_upper, const char* lang) { - auto case_converter = is_to_upper ? u_strToUpper : u_strToLower; - int32_t src_length = s->length(); - int32_t dest_length = src_length; - UErrorCode status; - Handle<SeqTwoByteString> result; - std::unique_ptr<uc16[]> sap; - - if (dest_length == 0) return isolate->heap()->empty_string(); - - // This is not a real loop. It'll be executed only once (no overflow) or - // twice (overflow). - for (int i = 0; i < 2; ++i) { - // Case conversion can increase the string length (e.g. sharp-S => SS) so - // that we have to handle RangeError exceptions here. - ASSIGN_RETURN_FAILURE_ON_EXCEPTION( - isolate, result, isolate->factory()->NewRawTwoByteString(dest_length)); - DisallowHeapAllocation no_gc; - DCHECK(s->IsFlat()); - String::FlatContent flat = s->GetFlatContent(); - const UChar* src = GetUCharBufferFromFlat(flat, &sap, src_length); - status = U_ZERO_ERROR; - dest_length = case_converter(reinterpret_cast<UChar*>(result->GetChars()), - dest_length, src, src_length, lang, &status); - if (status != U_BUFFER_OVERFLOW_ERROR) break; - } - - // In most cases, the output will fill the destination buffer completely - // leading to an unterminated string (U_STRING_NOT_TERMINATED_WARNING). - // Only in rare cases, it'll be shorter than the destination buffer and - // |result| has to be truncated. - DCHECK(U_SUCCESS(status)); - if (V8_LIKELY(status == U_STRING_NOT_TERMINATED_WARNING)) { - DCHECK(dest_length == result->length()); - return *result; - } - if (U_SUCCESS(status)) { - DCHECK(dest_length < result->length()); - return *Handle<SeqTwoByteString>::cast( - SeqString::Truncate(result, dest_length)); - } - return *s; -} - -inline bool IsASCIIUpper(uint16_t ch) { return ch >= 'A' && ch <= 'Z'; } - -const uint8_t kToLower[256] = { - 0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, 0x08, 0x09, 0x0A, 0x0B, - 0x0C, 0x0D, 0x0E, 0x0F, 0x10, 0x11, 0x12, 0x13, 0x14, 0x15, 0x16, 0x17, - 0x18, 0x19, 0x1A, 0x1B, 0x1C, 0x1D, 0x1E, 0x1F, 0x20, 0x21, 0x22, 0x23, - 0x24, 0x25, 0x26, 0x27, 0x28, 0x29, 0x2A, 0x2B, 0x2C, 0x2D, 0x2E, 0x2F, - 0x30, 0x31, 0x32, 0x33, 0x34, 0x35, 0x36, 0x37, 0x38, 0x39, 0x3A, 0x3B, - 0x3C, 0x3D, 0x3E, 0x3F, 0x40, 0x61, 0x62, 0x63, 0x64, 0x65, 0x66, 0x67, - 0x68, 0x69, 0x6A, 0x6B, 0x6C, 0x6D, 0x6E, 0x6F, 0x70, 0x71, 0x72, 0x73, - 0x74, 0x75, 0x76, 0x77, 0x78, 0x79, 0x7A, 0x5B, 0x5C, 0x5D, 0x5E, 0x5F, - 0x60, 0x61, 0x62, 0x63, 0x64, 0x65, 0x66, 0x67, 0x68, 0x69, 0x6A, 0x6B, - 0x6C, 0x6D, 0x6E, 0x6F, 0x70, 0x71, 0x72, 0x73, 0x74, 0x75, 0x76, 0x77, - 0x78, 0x79, 0x7A, 0x7B, 0x7C, 0x7D, 0x7E, 0x7F, 0x80, 0x81, 0x82, 0x83, - 0x84, 0x85, 0x86, 0x87, 0x88, 0x89, 0x8A, 0x8B, 0x8C, 0x8D, 0x8E, 0x8F, - 0x90, 0x91, 0x92, 0x93, 0x94, 0x95, 0x96, 0x97, 0x98, 0x99, 0x9A, 0x9B, - 0x9C, 0x9D, 0x9E, 0x9F, 0xA0, 0xA1, 0xA2, 0xA3, 0xA4, 0xA5, 0xA6, 0xA7, - 0xA8, 0xA9, 0xAA, 0xAB, 0xAC, 0xAD, 0xAE, 0xAF, 0xB0, 0xB1, 0xB2, 0xB3, - 0xB4, 0xB5, 0xB6, 0xB7, 0xB8, 0xB9, 0xBA, 0xBB, 0xBC, 0xBD, 0xBE, 0xBF, - 0xE0, 0xE1, 0xE2, 0xE3, 0xE4, 0xE5, 0xE6, 0xE7, 0xE8, 0xE9, 0xEA, 0xEB, - 0xEC, 0xED, 0xEE, 0xEF, 0xF0, 0xF1, 0xF2, 0xF3, 0xF4, 0xF5, 0xF6, 0xD7, - 0xF8, 0xF9, 0xFA, 0xFB, 0xFC, 0xFD, 0xFE, 0xDF, 0xE0, 0xE1, 0xE2, 0xE3, - 0xE4, 0xE5, 0xE6, 0xE7, 0xE8, 0xE9, 0xEA, 0xEB, 0xEC, 0xED, 0xEE, 0xEF, - 0xF0, 0xF1, 0xF2, 0xF3, 0xF4, 0xF5, 0xF6, 0xF7, 0xF8, 0xF9, 0xFA, 0xFB, - 0xFC, 0xFD, 0xFE, 0xFF, -}; - -inline uint16_t ToLatin1Lower(uint16_t ch) { - return static_cast<uint16_t>(kToLower[ch]); -} - -inline uint16_t ToASCIIUpper(uint16_t ch) { - return ch & ~((ch >= 'a' && ch <= 'z') << 5); -} - -// Does not work for U+00DF (sharp-s), U+00B5 (micron), U+00FF. -inline uint16_t ToLatin1Upper(uint16_t ch) { - DCHECK(ch != 0xDF && ch != 0xB5 && ch != 0xFF); - return ch & - ~(((ch >= 'a' && ch <= 'z') || (((ch & 0xE0) == 0xE0) && ch != 0xF7)) - << 5); -} - -template <typename Char> -bool ToUpperFastASCII(const Vector<const Char>& src, - Handle<SeqOneByteString> result) { - // Do a faster loop for the case where all the characters are ASCII. - uint16_t ored = 0; - int32_t index = 0; - for (auto it = src.begin(); it != src.end(); ++it) { - uint16_t ch = static_cast<uint16_t>(*it); - ored |= ch; - result->SeqOneByteStringSet(index++, ToASCIIUpper(ch)); - } - return !(ored & ~0x7F); -} - -const uint16_t sharp_s = 0xDF; - -template <typename Char> -bool ToUpperOneByte(const Vector<const Char>& src, uint8_t* dest, - int* sharp_s_count) { - // Still pretty-fast path for the input with non-ASCII Latin-1 characters. - - // There are two special cases. - // 1. U+00B5 and U+00FF are mapped to a character beyond U+00FF. - // 2. Lower case sharp-S converts to "SS" (two characters) - *sharp_s_count = 0; - for (auto it = src.begin(); it != src.end(); ++it) { - uint16_t ch = static_cast<uint16_t>(*it); - if (V8_UNLIKELY(ch == sharp_s)) { - ++(*sharp_s_count); - continue; - } - if (V8_UNLIKELY(ch == 0xB5 || ch == 0xFF)) { - // Since this upper-cased character does not fit in an 8-bit string, we - // need to take the 16-bit path. - return false; - } - *dest++ = ToLatin1Upper(ch); - } - - return true; -} - -template <typename Char> -void ToUpperWithSharpS(const Vector<const Char>& src, - Handle<SeqOneByteString> result) { - int32_t dest_index = 0; - for (auto it = src.begin(); it != src.end(); ++it) { - uint16_t ch = static_cast<uint16_t>(*it); - if (ch == sharp_s) { - result->SeqOneByteStringSet(dest_index++, 'S'); - result->SeqOneByteStringSet(dest_index++, 'S'); - } else { - result->SeqOneByteStringSet(dest_index++, ToLatin1Upper(ch)); - } - } -} - -inline int FindFirstUpperOrNonAscii(Handle<String> s, int length) { - for (int index = 0; index < length; ++index) { - uint16_t ch = s->Get(index); - if (V8_UNLIKELY(IsASCIIUpper(ch) || ch & ~0x7F)) { - return index; - } - } - return length; -} - -MUST_USE_RESULT Object* ConvertToLower(Handle<String> s, Isolate* isolate) { - if (!s->HasOnlyOneByteChars()) { - // Use a slower implementation for strings with characters beyond U+00FF. - return LocaleConvertCase(s, isolate, false, ""); - } - - int length = s->length(); - - // We depend here on the invariant that the length of a Latin1 - // string is invariant under ToLowerCase, and the result always - // fits in the Latin1 range in the *root locale*. It does not hold - // for ToUpperCase even in the root locale. - - // Scan the string for uppercase and non-ASCII characters for strings - // shorter than a machine-word without any memory allocation overhead. - // TODO(jshin): Apply this to a longer input by breaking FastAsciiConvert() - // to two parts, one for scanning the prefix with no change and the other for - // handling ASCII-only characters. - int index_to_first_unprocessed = length; - const bool is_short = length < static_cast<int>(sizeof(uintptr_t)); - if (is_short) { - index_to_first_unprocessed = FindFirstUpperOrNonAscii(s, length); - // Nothing to do if the string is all ASCII with no uppercase. - if (index_to_first_unprocessed == length) return *s; - } - - Handle<SeqOneByteString> result = - isolate->factory()->NewRawOneByteString(length).ToHandleChecked(); - - DisallowHeapAllocation no_gc; - DCHECK(s->IsFlat()); - String::FlatContent flat = s->GetFlatContent(); - uint8_t* dest = result->GetChars(); - if (flat.IsOneByte()) { - const uint8_t* src = flat.ToOneByteVector().start(); - bool has_changed_character = false; - index_to_first_unprocessed = FastAsciiConvert<true>( - reinterpret_cast<char*>(dest), reinterpret_cast<const char*>(src), - length, &has_changed_character); - // If not ASCII, we keep the result up to index_to_first_unprocessed and - // process the rest. - if (index_to_first_unprocessed == length) - return has_changed_character ? *result : *s; - - for (int index = index_to_first_unprocessed; index < length; ++index) { - dest[index] = ToLatin1Lower(static_cast<uint16_t>(src[index])); - } - } else { - if (index_to_first_unprocessed == length) { - DCHECK(!is_short); - index_to_first_unprocessed = FindFirstUpperOrNonAscii(s, length); - } - // Nothing to do if the string is all ASCII with no uppercase. - if (index_to_first_unprocessed == length) return *s; - const uint16_t* src = flat.ToUC16Vector().start(); - CopyChars(dest, src, index_to_first_unprocessed); - for (int index = index_to_first_unprocessed; index < length; ++index) { - dest[index] = ToLatin1Lower(static_cast<uint16_t>(src[index])); - } - } - - return *result; -} - -MUST_USE_RESULT Object* ConvertToUpper(Handle<String> s, Isolate* isolate) { - int32_t length = s->length(); - if (s->HasOnlyOneByteChars() && length > 0) { - Handle<SeqOneByteString> result = - isolate->factory()->NewRawOneByteString(length).ToHandleChecked(); - - DCHECK(s->IsFlat()); - int sharp_s_count; - bool is_result_single_byte; - { - DisallowHeapAllocation no_gc; - String::FlatContent flat = s->GetFlatContent(); - uint8_t* dest = result->GetChars(); - if (flat.IsOneByte()) { - Vector<const uint8_t> src = flat.ToOneByteVector(); - bool has_changed_character = false; - int index_to_first_unprocessed = - FastAsciiConvert<false>(reinterpret_cast<char*>(result->GetChars()), - reinterpret_cast<const char*>(src.start()), - length, &has_changed_character); - if (index_to_first_unprocessed == length) - return has_changed_character ? *result : *s; - // If not ASCII, we keep the result up to index_to_first_unprocessed and - // process the rest. - is_result_single_byte = - ToUpperOneByte(src.SubVector(index_to_first_unprocessed, length), - dest + index_to_first_unprocessed, &sharp_s_count); - } else { - DCHECK(flat.IsTwoByte()); - Vector<const uint16_t> src = flat.ToUC16Vector(); - if (ToUpperFastASCII(src, result)) return *result; - is_result_single_byte = ToUpperOneByte(src, dest, &sharp_s_count); - } - } - - // Go to the full Unicode path if there are characters whose uppercase - // is beyond the Latin-1 range (cannot be represented in OneByteString). - if (V8_UNLIKELY(!is_result_single_byte)) { - return LocaleConvertCase(s, isolate, true, ""); - } - - if (sharp_s_count == 0) return *result; - - // We have sharp_s_count sharp-s characters, but the result is still - // in the Latin-1 range. - ASSIGN_RETURN_FAILURE_ON_EXCEPTION( - isolate, result, - isolate->factory()->NewRawOneByteString(length + sharp_s_count)); - DisallowHeapAllocation no_gc; - String::FlatContent flat = s->GetFlatContent(); - if (flat.IsOneByte()) { - ToUpperWithSharpS(flat.ToOneByteVector(), result); - } else { - ToUpperWithSharpS(flat.ToUC16Vector(), result); - } - - return *result; - } - - return LocaleConvertCase(s, isolate, true, ""); -} - -MUST_USE_RESULT Object* ConvertCase(Handle<String> s, bool is_upper, - Isolate* isolate) { - return is_upper ? ConvertToUpper(s, isolate) : ConvertToLower(s, isolate); -} - -} // namespace - RUNTIME_FUNCTION(Runtime_StringToLowerCaseI18N) { HandleScope scope(isolate); DCHECK_EQ(args.length(), 1); |