Line data Source code
1 : // Copyright 2013 the V8 project authors. All rights reserved.
2 : // Use of this source code is governed by a BSD-style license that can be
3 : // found in the LICENSE file.
4 :
5 : #ifndef V8_INTL_SUPPORT
6 : #error Internationalization is expected to be enabled.
7 : #endif // V8_INTL_SUPPORT
8 :
9 : #include "src/objects/intl-objects.h"
10 :
11 : #include <algorithm>
12 : #include <memory>
13 : #include <string>
14 : #include <vector>
15 :
16 : #include "src/api-inl.h"
17 : #include "src/global-handles.h"
18 : #include "src/heap/factory.h"
19 : #include "src/isolate.h"
20 : #include "src/objects-inl.h"
21 : #include "src/objects/js-collator-inl.h"
22 : #include "src/objects/js-date-time-format-inl.h"
23 : #include "src/objects/js-locale-inl.h"
24 : #include "src/objects/js-number-format-inl.h"
25 : #include "src/objects/string.h"
26 : #include "src/property-descriptor.h"
27 : #include "src/string-case.h"
28 : #include "unicode/basictz.h"
29 : #include "unicode/brkiter.h"
30 : #include "unicode/calendar.h"
31 : #include "unicode/coll.h"
32 : #include "unicode/datefmt.h"
33 : #include "unicode/decimfmt.h"
34 : #include "unicode/locid.h"
35 : #include "unicode/normalizer2.h"
36 : #include "unicode/numfmt.h"
37 : #include "unicode/numsys.h"
38 : #include "unicode/timezone.h"
39 : #include "unicode/ustring.h"
40 : #include "unicode/uvernum.h" // U_ICU_VERSION_MAJOR_NUM
41 :
42 : #define XSTR(s) STR(s)
43 : #define STR(s) #s
44 : static_assert(
45 : V8_MINIMUM_ICU_VERSION <= U_ICU_VERSION_MAJOR_NUM,
46 : "v8 is required to build with ICU " XSTR(V8_MINIMUM_ICU_VERSION) " and up");
47 : #undef STR
48 : #undef XSTR
49 :
50 : namespace v8 {
51 : namespace internal {
52 :
53 : namespace {
54 355 : inline bool IsASCIIUpper(uint16_t ch) { return ch >= 'A' && ch <= 'Z'; }
55 :
56 : const uint8_t kToLower[256] = {
57 : 0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, 0x08, 0x09, 0x0A, 0x0B,
58 : 0x0C, 0x0D, 0x0E, 0x0F, 0x10, 0x11, 0x12, 0x13, 0x14, 0x15, 0x16, 0x17,
59 : 0x18, 0x19, 0x1A, 0x1B, 0x1C, 0x1D, 0x1E, 0x1F, 0x20, 0x21, 0x22, 0x23,
60 : 0x24, 0x25, 0x26, 0x27, 0x28, 0x29, 0x2A, 0x2B, 0x2C, 0x2D, 0x2E, 0x2F,
61 : 0x30, 0x31, 0x32, 0x33, 0x34, 0x35, 0x36, 0x37, 0x38, 0x39, 0x3A, 0x3B,
62 : 0x3C, 0x3D, 0x3E, 0x3F, 0x40, 0x61, 0x62, 0x63, 0x64, 0x65, 0x66, 0x67,
63 : 0x68, 0x69, 0x6A, 0x6B, 0x6C, 0x6D, 0x6E, 0x6F, 0x70, 0x71, 0x72, 0x73,
64 : 0x74, 0x75, 0x76, 0x77, 0x78, 0x79, 0x7A, 0x5B, 0x5C, 0x5D, 0x5E, 0x5F,
65 : 0x60, 0x61, 0x62, 0x63, 0x64, 0x65, 0x66, 0x67, 0x68, 0x69, 0x6A, 0x6B,
66 : 0x6C, 0x6D, 0x6E, 0x6F, 0x70, 0x71, 0x72, 0x73, 0x74, 0x75, 0x76, 0x77,
67 : 0x78, 0x79, 0x7A, 0x7B, 0x7C, 0x7D, 0x7E, 0x7F, 0x80, 0x81, 0x82, 0x83,
68 : 0x84, 0x85, 0x86, 0x87, 0x88, 0x89, 0x8A, 0x8B, 0x8C, 0x8D, 0x8E, 0x8F,
69 : 0x90, 0x91, 0x92, 0x93, 0x94, 0x95, 0x96, 0x97, 0x98, 0x99, 0x9A, 0x9B,
70 : 0x9C, 0x9D, 0x9E, 0x9F, 0xA0, 0xA1, 0xA2, 0xA3, 0xA4, 0xA5, 0xA6, 0xA7,
71 : 0xA8, 0xA9, 0xAA, 0xAB, 0xAC, 0xAD, 0xAE, 0xAF, 0xB0, 0xB1, 0xB2, 0xB3,
72 : 0xB4, 0xB5, 0xB6, 0xB7, 0xB8, 0xB9, 0xBA, 0xBB, 0xBC, 0xBD, 0xBE, 0xBF,
73 : 0xE0, 0xE1, 0xE2, 0xE3, 0xE4, 0xE5, 0xE6, 0xE7, 0xE8, 0xE9, 0xEA, 0xEB,
74 : 0xEC, 0xED, 0xEE, 0xEF, 0xF0, 0xF1, 0xF2, 0xF3, 0xF4, 0xF5, 0xF6, 0xD7,
75 : 0xF8, 0xF9, 0xFA, 0xFB, 0xFC, 0xFD, 0xFE, 0xDF, 0xE0, 0xE1, 0xE2, 0xE3,
76 : 0xE4, 0xE5, 0xE6, 0xE7, 0xE8, 0xE9, 0xEA, 0xEB, 0xEC, 0xED, 0xEE, 0xEF,
77 : 0xF0, 0xF1, 0xF2, 0xF3, 0xF4, 0xF5, 0xF6, 0xF7, 0xF8, 0xF9, 0xFA, 0xFB,
78 : 0xFC, 0xFD, 0xFE, 0xFF,
79 : };
80 :
81 : inline uint16_t ToLatin1Lower(uint16_t ch) {
82 1710 : return static_cast<uint16_t>(kToLower[ch]);
83 : }
84 :
85 : inline uint16_t ToASCIIUpper(uint16_t ch) {
86 0 : return ch & ~((ch >= 'a' && ch <= 'z') << 5);
87 : }
88 :
89 : // Does not work for U+00DF (sharp-s), U+00B5 (micron), U+00FF.
90 : inline uint16_t ToLatin1Upper(uint16_t ch) {
91 : DCHECK(ch != 0xDF && ch != 0xB5 && ch != 0xFF);
92 : return ch &
93 1337 : ~(((ch >= 'a' && ch <= 'z') || (((ch & 0xE0) == 0xE0) && ch != 0xF7))
94 1337 : << 5);
95 : }
96 :
97 : template <typename Char>
98 0 : bool ToUpperFastASCII(const Vector<const Char>& src,
99 : Handle<SeqOneByteString> result) {
100 : // Do a faster loop for the case where all the characters are ASCII.
101 : uint16_t ored = 0;
102 : int32_t index = 0;
103 0 : for (auto it = src.begin(); it != src.end(); ++it) {
104 0 : uint16_t ch = static_cast<uint16_t>(*it);
105 0 : ored |= ch;
106 0 : result->SeqOneByteStringSet(index++, ToASCIIUpper(ch));
107 : }
108 0 : return !(ored & ~0x7F);
109 : }
110 :
111 : const uint16_t sharp_s = 0xDF;
112 :
113 : template <typename Char>
114 424 : bool ToUpperOneByte(const Vector<const Char>& src, uint8_t* dest,
115 : int* sharp_s_count) {
116 : // Still pretty-fast path for the input with non-ASCII Latin-1 characters.
117 :
118 : // There are two special cases.
119 : // 1. U+00B5 and U+00FF are mapped to a character beyond U+00FF.
120 : // 2. Lower case sharp-S converts to "SS" (two characters)
121 212 : *sharp_s_count = 0;
122 2306 : for (auto it = src.begin(); it != src.end(); ++it) {
123 1004 : uint16_t ch = static_cast<uint16_t>(*it);
124 1004 : if (V8_UNLIKELY(ch == sharp_s)) {
125 99 : ++(*sharp_s_count);
126 99 : continue;
127 : }
128 905 : if (V8_UNLIKELY(ch == 0xB5 || ch == 0xFF)) {
129 : // Since this upper-cased character does not fit in an 8-bit string, we
130 : // need to take the 16-bit path.
131 : return false;
132 : }
133 1684 : *dest++ = ToLatin1Upper(ch);
134 : }
135 :
136 : return true;
137 : }
138 :
139 : template <typename Char>
140 81 : void ToUpperWithSharpS(const Vector<const Char>& src,
141 : Handle<SeqOneByteString> result) {
142 : int32_t dest_index = 0;
143 1350 : for (auto it = src.begin(); it != src.end(); ++it) {
144 594 : uint16_t ch = static_cast<uint16_t>(*it);
145 594 : if (ch == sharp_s) {
146 99 : result->SeqOneByteStringSet(dest_index++, 'S');
147 99 : result->SeqOneByteStringSet(dest_index++, 'S');
148 : } else {
149 495 : result->SeqOneByteStringSet(dest_index++, ToLatin1Upper(ch));
150 : }
151 : }
152 81 : }
153 :
154 41 : inline int FindFirstUpperOrNonAscii(String s, int length) {
155 396 : for (int index = 0; index < length; ++index) {
156 : uint16_t ch = s->Get(index);
157 355 : if (V8_UNLIKELY(IsASCIIUpper(ch) || ch & ~0x7F)) {
158 : return index;
159 : }
160 : }
161 : return length;
162 : }
163 :
164 693166 : const UChar* GetUCharBufferFromFlat(const String::FlatContent& flat,
165 : std::unique_ptr<uc16[]>* dest,
166 : int32_t length) {
167 : DCHECK(flat.IsFlat());
168 346619 : if (flat.IsOneByte()) {
169 338029 : if (!*dest) {
170 337957 : dest->reset(NewArray<uc16>(length));
171 : CopyChars(dest->get(), flat.ToOneByteVector().start(), length);
172 : }
173 338029 : return reinterpret_cast<const UChar*>(dest->get());
174 : } else {
175 : return reinterpret_cast<const UChar*>(flat.ToUC16Vector().start());
176 : }
177 : }
178 :
179 : template <typename T>
180 6127 : MaybeHandle<T> New(Isolate* isolate, Handle<JSFunction> constructor,
181 : Handle<Object> locales, Handle<Object> options) {
182 : Handle<JSObject> result;
183 12254 : ASSIGN_RETURN_ON_EXCEPTION(
184 : isolate, result,
185 : JSObject::New(constructor, constructor, Handle<AllocationSite>::null()),
186 : T);
187 6127 : return T::Initialize(isolate, Handle<T>::cast(result), locales, options);
188 : }
189 : } // namespace
190 :
191 61104 : const uint8_t* Intl::ToLatin1LowerTable() { return &kToLower[0]; }
192 :
193 344409 : icu::UnicodeString Intl::ToICUUnicodeString(Isolate* isolate,
194 : Handle<String> string) {
195 344409 : string = String::Flatten(isolate, string);
196 : {
197 : DisallowHeapAllocation no_gc;
198 344409 : std::unique_ptr<uc16[]> sap;
199 : return icu::UnicodeString(
200 688818 : GetUCharBufferFromFlat(string->GetFlatContent(no_gc), &sap,
201 : string->length()),
202 1033227 : string->length());
203 : }
204 : }
205 :
206 : namespace {
207 2030 : MaybeHandle<String> LocaleConvertCase(Isolate* isolate, Handle<String> s,
208 : bool is_to_upper, const char* lang) {
209 2030 : auto case_converter = is_to_upper ? u_strToUpper : u_strToLower;
210 : int32_t src_length = s->length();
211 : int32_t dest_length = src_length;
212 : UErrorCode status;
213 : Handle<SeqTwoByteString> result;
214 2030 : std::unique_ptr<uc16[]> sap;
215 :
216 2030 : if (dest_length == 0) return ReadOnlyRoots(isolate).empty_string_handle();
217 :
218 : // This is not a real loop. It'll be executed only once (no overflow) or
219 : // twice (overflow).
220 180 : for (int i = 0; i < 2; ++i) {
221 : // Case conversion can increase the string length (e.g. sharp-S => SS) so
222 : // that we have to handle RangeError exceptions here.
223 4420 : ASSIGN_RETURN_ON_EXCEPTION(
224 : isolate, result, isolate->factory()->NewRawTwoByteString(dest_length),
225 : String);
226 : DisallowHeapAllocation no_gc;
227 : DCHECK(s->IsFlat());
228 2210 : String::FlatContent flat = s->GetFlatContent(no_gc);
229 2210 : const UChar* src = GetUCharBufferFromFlat(flat, &sap, src_length);
230 2210 : status = U_ZERO_ERROR;
231 : dest_length =
232 : case_converter(reinterpret_cast<UChar*>(result->GetChars(no_gc)),
233 2210 : dest_length, src, src_length, lang, &status);
234 2210 : if (status != U_BUFFER_OVERFLOW_ERROR) break;
235 : }
236 :
237 : // In most cases, the output will fill the destination buffer completely
238 : // leading to an unterminated string (U_STRING_NOT_TERMINATED_WARNING).
239 : // Only in rare cases, it'll be shorter than the destination buffer and
240 : // |result| has to be truncated.
241 : DCHECK(U_SUCCESS(status));
242 2030 : if (V8_LIKELY(status == U_STRING_NOT_TERMINATED_WARNING)) {
243 : DCHECK(dest_length == result->length());
244 1895 : return result;
245 : }
246 : DCHECK(dest_length < result->length());
247 135 : return SeqString::Truncate(result, dest_length);
248 : }
249 :
250 : } // namespace
251 :
252 : // A stripped-down version of ConvertToLower that can only handle flat one-byte
253 : // strings and does not allocate. Note that {src} could still be, e.g., a
254 : // one-byte sliced string with a two-byte parent string.
255 : // Called from TF builtins.
256 2301 : String Intl::ConvertOneByteToLower(String src, String dst) {
257 : DCHECK_EQ(src->length(), dst->length());
258 : DCHECK(src->HasOnlyOneByteChars());
259 : DCHECK(src->IsFlat());
260 : DCHECK(dst->IsSeqOneByteString());
261 :
262 : DisallowHeapAllocation no_gc;
263 :
264 : const int length = src->length();
265 2301 : String::FlatContent src_flat = src->GetFlatContent(no_gc);
266 : uint8_t* dst_data = SeqOneByteString::cast(dst)->GetChars(no_gc);
267 :
268 2301 : if (src_flat.IsOneByte()) {
269 : const uint8_t* src_data = src_flat.ToOneByteVector().start();
270 :
271 2296 : bool has_changed_character = false;
272 : int index_to_first_unprocessed =
273 : FastAsciiConvert<true>(reinterpret_cast<char*>(dst_data),
274 : reinterpret_cast<const char*>(src_data), length,
275 2296 : &has_changed_character);
276 :
277 2296 : if (index_to_first_unprocessed == length) {
278 2278 : return has_changed_character ? dst : src;
279 : }
280 :
281 : // If not ASCII, we keep the result up to index_to_first_unprocessed and
282 : // process the rest.
283 1710 : for (int index = index_to_first_unprocessed; index < length; ++index) {
284 3420 : dst_data[index] = ToLatin1Lower(static_cast<uint16_t>(src_data[index]));
285 : }
286 : } else {
287 : DCHECK(src_flat.IsTwoByte());
288 5 : int index_to_first_unprocessed = FindFirstUpperOrNonAscii(src, length);
289 5 : if (index_to_first_unprocessed == length) return src;
290 :
291 : const uint16_t* src_data = src_flat.ToUC16Vector().start();
292 0 : CopyChars(dst_data, src_data, index_to_first_unprocessed);
293 0 : for (int index = index_to_first_unprocessed; index < length; ++index) {
294 0 : dst_data[index] = ToLatin1Lower(static_cast<uint16_t>(src_data[index]));
295 : }
296 : }
297 :
298 18 : return dst;
299 : }
300 :
301 2836 : MaybeHandle<String> Intl::ConvertToLower(Isolate* isolate, Handle<String> s) {
302 2836 : if (!s->HasOnlyOneByteChars()) {
303 : // Use a slower implementation for strings with characters beyond U+00FF.
304 1310 : return LocaleConvertCase(isolate, s, false, "");
305 : }
306 :
307 : int length = s->length();
308 :
309 : // We depend here on the invariant that the length of a Latin1
310 : // string is invariant under ToLowerCase, and the result always
311 : // fits in the Latin1 range in the *root locale*. It does not hold
312 : // for ToUpperCase even in the root locale.
313 :
314 : // Scan the string for uppercase and non-ASCII characters for strings
315 : // shorter than a machine-word without any memory allocation overhead.
316 : // TODO(jshin): Apply this to a longer input by breaking FastAsciiConvert()
317 : // to two parts, one for scanning the prefix with no change and the other for
318 : // handling ASCII-only characters.
319 :
320 : bool is_short = length < static_cast<int>(sizeof(uintptr_t));
321 1526 : if (is_short) {
322 36 : bool is_lower_ascii = FindFirstUpperOrNonAscii(*s, length) == length;
323 36 : if (is_lower_ascii) return s;
324 : }
325 :
326 : Handle<SeqOneByteString> result =
327 2980 : isolate->factory()->NewRawOneByteString(length).ToHandleChecked();
328 :
329 4470 : return Handle<String>(Intl::ConvertOneByteToLower(*s, *result), isolate);
330 : }
331 :
332 4828 : MaybeHandle<String> Intl::ConvertToUpper(Isolate* isolate, Handle<String> s) {
333 : int32_t length = s->length();
334 4828 : if (s->HasOnlyOneByteChars() && length > 0) {
335 : Handle<SeqOneByteString> result =
336 9116 : isolate->factory()->NewRawOneByteString(length).ToHandleChecked();
337 :
338 : DCHECK(s->IsFlat());
339 : int sharp_s_count;
340 : bool is_result_single_byte;
341 : {
342 : DisallowHeapAllocation no_gc;
343 4558 : String::FlatContent flat = s->GetFlatContent(no_gc);
344 : uint8_t* dest = result->GetChars(no_gc);
345 4558 : if (flat.IsOneByte()) {
346 : Vector<const uint8_t> src = flat.ToOneByteVector();
347 4558 : bool has_changed_character = false;
348 : int index_to_first_unprocessed = FastAsciiConvert<false>(
349 : reinterpret_cast<char*>(result->GetChars(no_gc)),
350 : reinterpret_cast<const char*>(src.start()), length,
351 4558 : &has_changed_character);
352 4558 : if (index_to_first_unprocessed == length) {
353 4346 : return has_changed_character ? result : s;
354 : }
355 : // If not ASCII, we keep the result up to index_to_first_unprocessed and
356 : // process the rest.
357 : is_result_single_byte =
358 212 : ToUpperOneByte(src.SubVector(index_to_first_unprocessed, length),
359 424 : dest + index_to_first_unprocessed, &sharp_s_count);
360 : } else {
361 : DCHECK(flat.IsTwoByte());
362 0 : Vector<const uint16_t> src = flat.ToUC16Vector();
363 0 : if (ToUpperFastASCII(src, result)) return result;
364 0 : is_result_single_byte = ToUpperOneByte(src, dest, &sharp_s_count);
365 : }
366 : }
367 :
368 : // Go to the full Unicode path if there are characters whose uppercase
369 : // is beyond the Latin-1 range (cannot be represented in OneByteString).
370 212 : if (V8_UNLIKELY(!is_result_single_byte)) {
371 63 : return LocaleConvertCase(isolate, s, true, "");
372 : }
373 :
374 149 : if (sharp_s_count == 0) return result;
375 :
376 : // We have sharp_s_count sharp-s characters, but the result is still
377 : // in the Latin-1 range.
378 162 : ASSIGN_RETURN_ON_EXCEPTION(
379 : isolate, result,
380 : isolate->factory()->NewRawOneByteString(length + sharp_s_count),
381 : String);
382 : DisallowHeapAllocation no_gc;
383 81 : String::FlatContent flat = s->GetFlatContent(no_gc);
384 81 : if (flat.IsOneByte()) {
385 81 : ToUpperWithSharpS(flat.ToOneByteVector(), result);
386 : } else {
387 0 : ToUpperWithSharpS(flat.ToUC16Vector(), result);
388 : }
389 :
390 81 : return result;
391 : }
392 :
393 270 : return LocaleConvertCase(isolate, s, true, "");
394 : }
395 :
396 3771 : std::string Intl::GetNumberingSystem(const icu::Locale& icu_locale) {
397 : // Ugly hack. ICU doesn't expose numbering system in any way, so we have
398 : // to assume that for given locale NumberingSystem constructor produces the
399 : // same digits as NumberFormat/Calendar would.
400 3771 : UErrorCode status = U_ZERO_ERROR;
401 : std::unique_ptr<icu::NumberingSystem> numbering_system(
402 3771 : icu::NumberingSystem::createInstance(icu_locale, status));
403 3771 : if (U_SUCCESS(status)) return numbering_system->getName();
404 0 : return "latn";
405 : }
406 :
407 31470 : icu::Locale Intl::CreateICULocale(const std::string& bcp47_locale) {
408 : DisallowHeapAllocation no_gc;
409 :
410 : // Convert BCP47 into ICU locale format.
411 15735 : UErrorCode status = U_ZERO_ERROR;
412 :
413 15735 : icu::Locale icu_locale = icu::Locale::forLanguageTag(bcp47_locale, status);
414 31470 : CHECK(U_SUCCESS(status));
415 15735 : if (icu_locale.isBogus()) {
416 0 : FATAL("Failed to create ICU locale, are ICU data files missing?");
417 : }
418 :
419 15735 : return icu_locale;
420 : }
421 :
422 : // static
423 :
424 41598 : MaybeHandle<String> Intl::ToString(Isolate* isolate,
425 : const icu::UnicodeString& string) {
426 : return isolate->factory()->NewStringFromTwoByte(Vector<const uint16_t>(
427 83196 : reinterpret_cast<const uint16_t*>(string.getBuffer()), string.length()));
428 : }
429 :
430 35811 : MaybeHandle<String> Intl::ToString(Isolate* isolate,
431 : const icu::UnicodeString& string,
432 : int32_t begin, int32_t end) {
433 35811 : return Intl::ToString(isolate, string.tempSubStringBetween(begin, end));
434 : }
435 :
436 : namespace {
437 :
438 18900 : Handle<JSObject> InnerAddElement(Isolate* isolate, Handle<JSArray> array,
439 : int index, Handle<String> field_type_string,
440 : Handle<String> value) {
441 : // let element = $array[$index] = {
442 : // type: $field_type_string,
443 : // value: $value
444 : // }
445 : // return element;
446 : Factory* factory = isolate->factory();
447 18900 : Handle<JSObject> element = factory->NewJSObject(isolate->object_function());
448 : JSObject::AddProperty(isolate, element, factory->type_string(),
449 18900 : field_type_string, NONE);
450 :
451 18900 : JSObject::AddProperty(isolate, element, factory->value_string(), value, NONE);
452 37800 : JSObject::AddDataElement(array, index, element, NONE);
453 18900 : return element;
454 : }
455 :
456 : } // namespace
457 :
458 18450 : void Intl::AddElement(Isolate* isolate, Handle<JSArray> array, int index,
459 : Handle<String> field_type_string, Handle<String> value) {
460 : // Same as $array[$index] = {type: $field_type_string, value: $value};
461 18450 : InnerAddElement(isolate, array, index, field_type_string, value);
462 18450 : }
463 :
464 450 : void Intl::AddElement(Isolate* isolate, Handle<JSArray> array, int index,
465 : Handle<String> field_type_string, Handle<String> value,
466 : Handle<String> additional_property_name,
467 : Handle<String> additional_property_value) {
468 : // Same as $array[$index] = {
469 : // type: $field_type_string, value: $value,
470 : // $additional_property_name: $additional_property_value
471 : // }
472 : Handle<JSObject> element =
473 450 : InnerAddElement(isolate, array, index, field_type_string, value);
474 : JSObject::AddProperty(isolate, element, additional_property_name,
475 450 : additional_property_value, NONE);
476 450 : }
477 :
478 : namespace {
479 :
480 : // Build the shortened locale; eg, convert xx_Yyyy_ZZ to xx_ZZ.
481 : //
482 : // If locale has a script tag then return true and the locale without the
483 : // script else return false and an empty string.
484 317906 : bool RemoveLocaleScriptTag(const std::string& icu_locale,
485 : std::string* locale_less_script) {
486 317906 : icu::Locale new_locale = icu::Locale::createCanonical(icu_locale.c_str());
487 : const char* icu_script = new_locale.getScript();
488 317906 : if (icu_script == nullptr || strlen(icu_script) == 0) {
489 302658 : *locale_less_script = std::string();
490 302658 : return false;
491 : }
492 :
493 : const char* icu_language = new_locale.getLanguage();
494 : const char* icu_country = new_locale.getCountry();
495 30496 : icu::Locale short_locale = icu::Locale(icu_language, icu_country);
496 15248 : *locale_less_script = short_locale.getName();
497 317906 : return true;
498 : }
499 :
500 : } // namespace
501 :
502 984 : std::set<std::string> Intl::BuildLocaleSet(
503 : const icu::Locale* icu_available_locales, int32_t count) {
504 : std::set<std::string> locales;
505 318890 : for (int32_t i = 0; i < count; ++i) {
506 : std::string locale =
507 635812 : Intl::ToLanguageTag(icu_available_locales[i]).FromJust();
508 : locales.insert(locale);
509 :
510 : std::string shortened_locale;
511 317906 : if (RemoveLocaleScriptTag(locale, &shortened_locale)) {
512 15248 : std::replace(shortened_locale.begin(), shortened_locale.end(), '_', '-');
513 : locales.insert(shortened_locale);
514 : }
515 : }
516 :
517 984 : return locales;
518 : }
519 :
520 596579 : Maybe<std::string> Intl::ToLanguageTag(const icu::Locale& locale) {
521 596579 : UErrorCode status = U_ZERO_ERROR;
522 596579 : std::string res = locale.toLanguageTag<std::string>(status);
523 596579 : if (U_FAILURE(status)) {
524 : return Nothing<std::string>();
525 : }
526 596570 : CHECK(U_SUCCESS(status));
527 :
528 : // Hack to remove -true and -yes from unicode extensions
529 : // Address https://crbug.com/v8/8565
530 : // TODO(ftang): Move the following "remove true" logic into ICU toLanguageTag
531 : // by fixing ICU-20310.
532 596570 : size_t u_ext_start = res.find("-u-");
533 596570 : if (u_ext_start != std::string::npos) {
534 : // remove "-true" and "-yes" after -u-
535 17818 : const std::vector<std::string> remove_items({"-true", "-yes"});
536 35636 : for (auto item = remove_items.begin(); item != remove_items.end(); item++) {
537 35669 : for (size_t sep_remove =
538 17818 : res.find(*item, u_ext_start + 5 /* strlen("-u-xx") == 5 */);
539 : sep_remove != std::string::npos; sep_remove = res.find(*item)) {
540 33 : size_t end_of_sep_remove = sep_remove + item->length();
541 51 : if (res.length() == end_of_sep_remove ||
542 18 : res.at(end_of_sep_remove) == '-') {
543 33 : res.erase(sep_remove, item->length());
544 : }
545 : }
546 8909 : }
547 : }
548 : return Just(res);
549 : }
550 :
551 : namespace {
552 2008 : std::string DefaultLocale(Isolate* isolate) {
553 2008 : if (isolate->default_locale().empty()) {
554 474 : icu::Locale default_locale;
555 : // Translate ICU's fallback locale to a well-known locale.
556 474 : if (strcmp(default_locale.getName(), "en_US_POSIX") == 0) {
557 0 : isolate->set_default_locale("en-US");
558 : } else {
559 : // Set the locale
560 : isolate->set_default_locale(
561 474 : default_locale.isBogus()
562 : ? "und"
563 1422 : : Intl::ToLanguageTag(default_locale).FromJust());
564 : }
565 474 : DCHECK(!isolate->default_locale().empty());
566 : }
567 2008 : return isolate->default_locale();
568 : }
569 : } // namespace
570 :
571 : // See ecma402/#legacy-constructor.
572 5994 : MaybeHandle<Object> Intl::LegacyUnwrapReceiver(Isolate* isolate,
573 : Handle<JSReceiver> receiver,
574 : Handle<JSFunction> constructor,
575 : bool has_initialized_slot) {
576 : Handle<Object> obj_is_instance_of;
577 11988 : ASSIGN_RETURN_ON_EXCEPTION(isolate, obj_is_instance_of,
578 : Object::InstanceOf(isolate, receiver, constructor),
579 : Object);
580 5994 : bool is_instance_of = obj_is_instance_of->BooleanValue(isolate);
581 :
582 : // 2. If receiver does not have an [[Initialized...]] internal slot
583 : // and ? InstanceofOperator(receiver, constructor) is true, then
584 5994 : if (!has_initialized_slot && is_instance_of) {
585 : // 2. a. Let new_receiver be ? Get(receiver, %Intl%.[[FallbackSymbol]]).
586 : Handle<Object> new_receiver;
587 180 : ASSIGN_RETURN_ON_EXCEPTION(
588 : isolate, new_receiver,
589 : JSReceiver::GetProperty(isolate, receiver,
590 : isolate->factory()->intl_fallback_symbol()),
591 : Object);
592 90 : return new_receiver;
593 : }
594 :
595 5904 : return receiver;
596 : }
597 :
598 505322 : Maybe<bool> Intl::GetStringOption(Isolate* isolate, Handle<JSReceiver> options,
599 : const char* property,
600 13496 : std::vector<const char*> values,
601 : const char* service,
602 : std::unique_ptr<char[]>* result) {
603 : Handle<String> property_str =
604 505322 : isolate->factory()->NewStringFromAsciiChecked(property);
605 :
606 : // 1. Let value be ? Get(options, property).
607 : Handle<Object> value;
608 1010644 : ASSIGN_RETURN_ON_EXCEPTION_VALUE(
609 : isolate, value,
610 : Object::GetPropertyOrElement(isolate, options, property_str),
611 : Nothing<bool>());
612 :
613 1010446 : if (value->IsUndefined(isolate)) {
614 : return Just(false);
615 : }
616 :
617 : // 2. c. Let value be ? ToString(value).
618 : Handle<String> value_str;
619 27010 : ASSIGN_RETURN_ON_EXCEPTION_VALUE(
620 : isolate, value_str, Object::ToString(isolate, value), Nothing<bool>());
621 13496 : std::unique_ptr<char[]> value_cstr = value_str->ToCString();
622 :
623 : // 2. d. if values is not undefined, then
624 13496 : if (values.size() > 0) {
625 : // 2. d. i. If values does not contain an element equal to value,
626 : // throw a RangeError exception.
627 53445 : for (size_t i = 0; i < values.size(); i++) {
628 32725 : if (strcmp(values.at(i), value_cstr.get()) == 0) {
629 : // 2. e. return value
630 : *result = std::move(value_cstr);
631 : return Just(true);
632 : }
633 : }
634 :
635 : Handle<String> service_str =
636 365 : isolate->factory()->NewStringFromAsciiChecked(service);
637 730 : THROW_NEW_ERROR_RETURN_VALUE(
638 : isolate,
639 : NewRangeError(MessageTemplate::kValueOutOfRange, value, service_str,
640 : property_str),
641 : Nothing<bool>());
642 : }
643 :
644 : // 2. e. return value
645 : *result = std::move(value_cstr);
646 : return Just(true);
647 : }
648 :
649 70685 : V8_WARN_UNUSED_RESULT Maybe<bool> Intl::GetBoolOption(
650 : Isolate* isolate, Handle<JSReceiver> options, const char* property,
651 : const char* service, bool* result) {
652 : Handle<String> property_str =
653 70685 : isolate->factory()->NewStringFromAsciiChecked(property);
654 :
655 : // 1. Let value be ? Get(options, property).
656 : Handle<Object> value;
657 141370 : ASSIGN_RETURN_ON_EXCEPTION_VALUE(
658 : isolate, value,
659 : Object::GetPropertyOrElement(isolate, options, property_str),
660 : Nothing<bool>());
661 :
662 : // 2. If value is not undefined, then
663 141352 : if (!value->IsUndefined(isolate)) {
664 : // 2. b. i. Let value be ToBoolean(value).
665 100 : *result = value->BooleanValue(isolate);
666 :
667 : // 2. e. return value
668 : return Just(true);
669 : }
670 :
671 : return Just(false);
672 : }
673 :
674 : namespace {
675 :
676 768346 : char AsciiToLower(char c) {
677 768346 : if (c < 'A' || c > 'Z') {
678 : return c;
679 : }
680 205470 : return c | (1 << 5);
681 : }
682 :
683 51564 : bool IsLowerAscii(char c) { return c >= 'a' && c < 'z'; }
684 :
685 : bool IsTwoLetterLanguage(const std::string& locale) {
686 : // Two letters, both in range 'a'-'z'...
687 171896 : return locale.length() == 2 && IsLowerAscii(locale[0]) &&
688 25566 : IsLowerAscii(locale[1]);
689 : }
690 :
691 25008 : bool IsDeprecatedLanguage(const std::string& locale) {
692 : // Check if locale is one of the deprecated language tags:
693 100005 : return locale == "in" || locale == "iw" || locale == "ji" || locale == "jw";
694 : }
695 :
696 : // Reference:
697 : // https://www.iana.org/assignments/language-subtag-registry/language-subtag-registry
698 95162 : bool IsGrandfatheredTagWithoutPreferredVaule(const std::string& locale) {
699 190297 : if (V8_UNLIKELY(locale == "zh-min" || locale == "cel-gaulish")) return true;
700 147871 : if (locale.length() > 6 /* i-mingo is 7 chars long */ &&
701 52745 : V8_UNLIKELY(locale[0] == 'i' && locale[1] == '-')) {
702 486 : return locale.substr(2) == "default" || locale.substr(2) == "enochian" ||
703 135 : locale.substr(2) == "mingo";
704 : }
705 : return false;
706 : }
707 :
708 : } // anonymous namespace
709 :
710 67484 : Maybe<std::string> Intl::CanonicalizeLanguageTag(Isolate* isolate,
711 : Handle<Object> locale_in) {
712 : Handle<String> locale_str;
713 : // This does part of the validity checking spec'ed in CanonicalizeLocaleList:
714 : // 7c ii. If Type(kValue) is not String or Object, throw a TypeError
715 : // exception.
716 : // 7c iii. Let tag be ? ToString(kValue).
717 : // 7c iv. If IsStructurallyValidLanguageTag(tag) is false, throw a
718 : // RangeError exception.
719 :
720 134968 : if (locale_in->IsString()) {
721 67448 : locale_str = Handle<String>::cast(locale_in);
722 72 : } else if (locale_in->IsJSReceiver()) {
723 0 : ASSIGN_RETURN_ON_EXCEPTION_VALUE(isolate, locale_str,
724 : Object::ToString(isolate, locale_in),
725 : Nothing<std::string>());
726 : } else {
727 72 : THROW_NEW_ERROR_RETURN_VALUE(isolate,
728 : NewTypeError(MessageTemplate::kLanguageID),
729 : Nothing<std::string>());
730 : }
731 202344 : std::string locale(locale_str->ToCString().get());
732 :
733 67448 : return Intl::CanonicalizeLanguageTag(isolate, locale);
734 : }
735 :
736 120332 : Maybe<std::string> Intl::CanonicalizeLanguageTag(Isolate* isolate,
737 : const std::string& locale_in) {
738 120332 : std::string locale = locale_in;
739 :
740 240664 : if (locale.length() == 0 ||
741 120332 : !String::IsAscii(locale.data(), static_cast<int>(locale.length()))) {
742 0 : THROW_NEW_ERROR_RETURN_VALUE(
743 : isolate,
744 : NewRangeError(
745 : MessageTemplate::kInvalidLanguageTag,
746 : isolate->factory()->NewStringFromAsciiChecked(locale.c_str())),
747 : Nothing<std::string>());
748 : }
749 :
750 : // Optimize for the most common case: a 2-letter language code in the
751 : // canonical form/lowercase that is not one of the deprecated codes
752 : // (in, iw, ji, jw). Don't check for ~70 of 3-letter deprecated language
753 : // codes. Instead, let them be handled by ICU in the slow path. However,
754 : // fast-track 'fil' (3-letter canonical code).
755 215674 : if ((IsTwoLetterLanguage(locale) && !IsDeprecatedLanguage(locale)) ||
756 : locale == "fil") {
757 : return Just(locale);
758 : }
759 :
760 : // Because per BCP 47 2.1.1 language tags are case-insensitive, lowercase
761 : // the input before any more check.
762 95162 : std::transform(locale.begin(), locale.end(), locale.begin(), AsciiToLower);
763 :
764 : // ICU maps a few grandfathered tags to what looks like a regular language
765 : // tag even though IANA language tag registry does not have a preferred
766 : // entry map for them. Return them as they're with lowercasing.
767 95162 : if (IsGrandfatheredTagWithoutPreferredVaule(locale)) {
768 : return Just(locale);
769 : }
770 :
771 : // // ECMA 402 6.2.3
772 : // TODO(jshin): uloc_{for,to}TanguageTag can fail even for a structually valid
773 : // language tag if it's too long (much longer than 100 chars). Even if we
774 : // allocate a longer buffer, ICU will still fail if it's too long. Either
775 : // propose to Ecma 402 to put a limit on the locale length or change ICU to
776 : // handle long locale names better. See
777 : // https://unicode-org.atlassian.net/browse/ICU-13417
778 95036 : UErrorCode error = U_ZERO_ERROR;
779 : // uloc_forLanguageTag checks the structrual validity. If the input BCP47
780 : // language tag is parsed all the way to the end, it indicates that the input
781 : // is structurally valid. Due to a couple of bugs, we can't use it
782 : // without Chromium patches or ICU 62 or earlier.
783 190072 : icu::Locale icu_locale = icu::Locale::forLanguageTag(locale.c_str(), error);
784 95036 : if (U_FAILURE(error) || icu_locale.isBogus()) {
785 378 : THROW_NEW_ERROR_RETURN_VALUE(
786 : isolate,
787 : NewRangeError(
788 : MessageTemplate::kInvalidLanguageTag,
789 : isolate->factory()->NewStringFromAsciiChecked(locale.c_str())),
790 : Nothing<std::string>());
791 : }
792 94910 : Maybe<std::string> maybe_to_language_tag = Intl::ToLanguageTag(icu_locale);
793 94910 : if (maybe_to_language_tag.IsNothing()) {
794 27 : THROW_NEW_ERROR_RETURN_VALUE(
795 : isolate,
796 : NewRangeError(
797 : MessageTemplate::kInvalidLanguageTag,
798 : isolate->factory()->NewStringFromAsciiChecked(locale.c_str())),
799 : Nothing<std::string>());
800 : }
801 :
802 : return maybe_to_language_tag;
803 : }
804 :
805 16437 : Maybe<std::vector<std::string>> Intl::CanonicalizeLocaleList(
806 : Isolate* isolate, Handle<Object> locales, bool only_return_one_result) {
807 : // 1. If locales is undefined, then
808 32874 : if (locales->IsUndefined(isolate)) {
809 : // 1a. Return a new empty List.
810 1612 : return Just(std::vector<std::string>());
811 : }
812 : // 2. Let seen be a new empty List.
813 : std::vector<std::string> seen;
814 : // 3. If Type(locales) is String or locales has an [[InitializedLocale]]
815 : // internal slot, then
816 29650 : if (locales->IsJSLocale()) {
817 : // Since this value came from JSLocale, which is already went though the
818 : // CanonializeLanguageTag process once, therefore there are no need to
819 : // call CanonializeLanguageTag again.
820 54 : seen.push_back(JSLocale::ToString(Handle<JSLocale>::cast(locales)));
821 : return Just(seen);
822 : }
823 29596 : if (locales->IsString()) {
824 : // 3a. Let O be CreateArrayFromList(« locales »).
825 : // Instead of creating a one-element array and then iterating over it,
826 : // we inline the body of the iteration:
827 : std::string canonicalized_tag;
828 20450 : if (!CanonicalizeLanguageTag(isolate, locales).To(&canonicalized_tag)) {
829 : return Nothing<std::vector<std::string>>();
830 : }
831 10126 : seen.push_back(canonicalized_tag);
832 : return Just(seen);
833 : }
834 : // 4. Else,
835 : // 4a. Let O be ? ToObject(locales).
836 : Handle<JSReceiver> o;
837 9146 : ASSIGN_RETURN_ON_EXCEPTION_VALUE(isolate, o,
838 : Object::ToObject(isolate, locales),
839 : Nothing<std::vector<std::string>>());
840 : // 5. Let len be ? ToLength(? Get(O, "length")).
841 : Handle<Object> length_obj;
842 9146 : ASSIGN_RETURN_ON_EXCEPTION_VALUE(isolate, length_obj,
843 : Object::GetLengthFromArrayLike(isolate, o),
844 : Nothing<std::vector<std::string>>());
845 : // TODO(jkummerow): Spec violation: strictly speaking, we have to iterate
846 : // up to 2^53-1 if {length_obj} says so. Since cases above 2^32 probably
847 : // don't happen in practice (and would be very slow if they do), we'll keep
848 : // the code simple for now by using a saturating to-uint32 conversion.
849 4573 : double raw_length = length_obj->Number();
850 : uint32_t len =
851 4573 : raw_length >= kMaxUInt32 ? kMaxUInt32 : static_cast<uint32_t>(raw_length);
852 : // 6. Let k be 0.
853 : // 7. Repeat, while k < len
854 8840 : for (uint32_t k = 0; k < len; k++) {
855 : // 7a. Let Pk be ToString(k).
856 : // 7b. Let kPresent be ? HasProperty(O, Pk).
857 4420 : LookupIterator it(isolate, o, k);
858 4420 : Maybe<bool> maybe_found = JSReceiver::HasProperty(&it);
859 4573 : MAYBE_RETURN(maybe_found, Nothing<std::vector<std::string>>());
860 : // 7c. If kPresent is true, then
861 4420 : if (!maybe_found.FromJust()) continue;
862 : // 7c i. Let kValue be ? Get(O, Pk).
863 : Handle<Object> k_value;
864 8840 : ASSIGN_RETURN_ON_EXCEPTION_VALUE(isolate, k_value, Object::GetProperty(&it),
865 : Nothing<std::vector<std::string>>());
866 : // 7c ii. If Type(kValue) is not String or Object, throw a TypeError
867 : // exception.
868 : // 7c iii. If Type(kValue) is Object and kValue has an [[InitializedLocale]]
869 : // internal slot, then
870 : std::string canonicalized_tag;
871 8840 : if (k_value->IsJSLocale()) {
872 : // 7c iii. 1. Let tag be kValue.[[Locale]].
873 90 : canonicalized_tag = JSLocale::ToString(Handle<JSLocale>::cast(k_value));
874 : // 7c iv. Else,
875 : } else {
876 : // 7c iv 1. Let tag be ? ToString(kValue).
877 : // 7c v. If IsStructurallyValidLanguageTag(tag) is false, throw a
878 : // RangeError exception.
879 : // 7c vi. Let canonicalizedTag be CanonicalizeLanguageTag(tag).
880 8750 : if (!CanonicalizeLanguageTag(isolate, k_value).To(&canonicalized_tag)) {
881 : return Nothing<std::vector<std::string>>();
882 : }
883 : }
884 : // 7c vi. If canonicalizedTag is not an element of seen, append
885 : // canonicalizedTag as the last element of seen.
886 4348 : if (std::find(seen.begin(), seen.end(), canonicalized_tag) == seen.end()) {
887 4348 : seen.push_back(canonicalized_tag);
888 : }
889 : // 7d. Increase k by 1. (See loop header.)
890 : // Optimization: some callers only need one result.
891 4348 : if (only_return_one_result) return Just(seen);
892 : }
893 : // 8. Return seen.
894 14825 : return Just(seen);
895 : }
896 :
897 : // ecma402 #sup-string.prototype.tolocalelowercase
898 : // ecma402 #sup-string.prototype.tolocaleuppercase
899 801 : MaybeHandle<String> Intl::StringLocaleConvertCase(Isolate* isolate,
900 : Handle<String> s,
901 : bool to_upper,
902 : Handle<Object> locales) {
903 : std::vector<std::string> requested_locales;
904 1602 : if (!CanonicalizeLocaleList(isolate, locales, true).To(&requested_locales)) {
905 36 : return MaybeHandle<String>();
906 : }
907 765 : std::string requested_locale = requested_locales.size() == 0
908 : ? DefaultLocale(isolate)
909 765 : : requested_locales[0];
910 765 : size_t dash = requested_locale.find('-');
911 765 : if (dash != std::string::npos) {
912 360 : requested_locale = requested_locale.substr(0, dash);
913 : }
914 :
915 : // Primary language tag can be up to 8 characters long in theory.
916 : // https://tools.ietf.org/html/bcp47#section-2.2.1
917 : DCHECK_LE(requested_locale.length(), 8);
918 765 : s = String::Flatten(isolate, s);
919 :
920 : // All the languages requiring special-handling have two-letter codes.
921 : // Note that we have to check for '!= 2' here because private-use language
922 : // tags (x-foo) or grandfathered irregular tags (e.g. i-enochian) would have
923 : // only 'x' or 'i' when they get here.
924 765 : if (V8_UNLIKELY(requested_locale.length() != 2)) {
925 135 : if (to_upper) {
926 45 : return ConvertToUpper(isolate, s);
927 : }
928 90 : return ConvertToLower(isolate, s);
929 : }
930 : // TODO(jshin): Consider adding a fast path for ASCII or Latin-1. The fastpath
931 : // in the root locale needs to be adjusted for az, lt and tr because even case
932 : // mapping of ASCII range characters are different in those locales.
933 : // Greek (el) does not require any adjustment.
934 1728 : if (V8_UNLIKELY((requested_locale == "tr") || (requested_locale == "el") ||
935 : (requested_locale == "lt") || (requested_locale == "az"))) {
936 387 : return LocaleConvertCase(isolate, s, to_upper, requested_locale.c_str());
937 : } else {
938 243 : if (to_upper) {
939 81 : return ConvertToUpper(isolate, s);
940 : }
941 162 : return ConvertToLower(isolate, s);
942 801 : }
943 : }
944 :
945 68751 : MaybeHandle<Object> Intl::StringLocaleCompare(Isolate* isolate,
946 : Handle<String> string1,
947 : Handle<String> string2,
948 : Handle<Object> locales,
949 : Handle<Object> options) {
950 : // We only cache the instance when both locales and options are undefined,
951 : // as that is the only case when the specified side-effects of examining
952 : // those arguments are unobservable.
953 : bool can_cache =
954 264176 : locales->IsUndefined(isolate) && options->IsUndefined(isolate);
955 68751 : if (can_cache) {
956 : // Both locales and options are undefined, check the cache.
957 : icu::Collator* cached_icu_collator =
958 : static_cast<icu::Collator*>(isolate->get_cached_icu_object(
959 63337 : Isolate::ICUObjectCacheType::kDefaultCollator));
960 : // We may use the cached icu::Collator for a fast path.
961 63337 : if (cached_icu_collator != nullptr) {
962 : return Intl::CompareStrings(isolate, *cached_icu_collator, string1,
963 63298 : string2);
964 : }
965 : }
966 :
967 : Handle<JSFunction> constructor = Handle<JSFunction>(
968 : JSFunction::cast(
969 10906 : isolate->context()->native_context()->intl_collator_function()),
970 10906 : isolate);
971 :
972 : Handle<JSCollator> collator;
973 10906 : ASSIGN_RETURN_ON_EXCEPTION(
974 : isolate, collator,
975 : New<JSCollator>(isolate, constructor, locales, options), Object);
976 5453 : if (can_cache) {
977 : isolate->set_icu_object_in_cache(
978 : Isolate::ICUObjectCacheType::kDefaultCollator,
979 : std::static_pointer_cast<icu::UObject>(
980 156 : collator->icu_collator()->get()));
981 : }
982 10906 : icu::Collator* icu_collator = collator->icu_collator()->raw();
983 5453 : return Intl::CompareStrings(isolate, *icu_collator, string1, string2);
984 : }
985 :
986 : // ecma402/#sec-collator-comparestrings
987 74930 : Handle<Object> Intl::CompareStrings(Isolate* isolate,
988 : const icu::Collator& icu_collator,
989 : Handle<String> string1,
990 : Handle<String> string2) {
991 : Factory* factory = isolate->factory();
992 :
993 74930 : string1 = String::Flatten(isolate, string1);
994 74930 : string2 = String::Flatten(isolate, string2);
995 :
996 : UCollationResult result;
997 74930 : UErrorCode status = U_ZERO_ERROR;
998 74930 : icu::UnicodeString string_val1 = Intl::ToICUUnicodeString(isolate, string1);
999 149860 : icu::UnicodeString string_val2 = Intl::ToICUUnicodeString(isolate, string2);
1000 74930 : result = icu_collator.compare(string_val1, string_val2, status);
1001 : DCHECK(U_SUCCESS(status));
1002 :
1003 149860 : return factory->NewNumberFromInt(result);
1004 : }
1005 :
1006 : // ecma402/#sup-properties-of-the-number-prototype-object
1007 1899 : MaybeHandle<String> Intl::NumberToLocaleString(Isolate* isolate,
1008 : Handle<Object> num,
1009 : Handle<Object> locales,
1010 : Handle<Object> options) {
1011 : Handle<Object> numeric_obj;
1012 1899 : if (FLAG_harmony_intl_bigint) {
1013 1188 : ASSIGN_RETURN_ON_EXCEPTION(isolate, numeric_obj,
1014 : Object::ToNumeric(isolate, num), String);
1015 : } else {
1016 2610 : ASSIGN_RETURN_ON_EXCEPTION(isolate, numeric_obj,
1017 : Object::ToNumber(isolate, num), String);
1018 : }
1019 :
1020 : // We only cache the instance when both locales and options are undefined,
1021 : // as that is the only case when the specified side-effects of examining
1022 : // those arguments are unobservable.
1023 : bool can_cache =
1024 6318 : locales->IsUndefined(isolate) && options->IsUndefined(isolate);
1025 1899 : if (can_cache) {
1026 : icu::NumberFormat* cached_number_format =
1027 : static_cast<icu::NumberFormat*>(isolate->get_cached_icu_object(
1028 1260 : Isolate::ICUObjectCacheType::kDefaultNumberFormat));
1029 : // We may use the cached icu::NumberFormat for a fast path.
1030 1260 : if (cached_number_format != nullptr) {
1031 : return JSNumberFormat::FormatNumeric(isolate, *cached_number_format,
1032 1225 : numeric_obj);
1033 : }
1034 : }
1035 :
1036 : Handle<JSFunction> constructor = Handle<JSFunction>(
1037 : JSFunction::cast(
1038 1348 : isolate->context()->native_context()->intl_number_format_function()),
1039 1348 : isolate);
1040 : Handle<JSNumberFormat> number_format;
1041 : // 2. Let numberFormat be ? Construct(%NumberFormat%, « locales, options »).
1042 1348 : ASSIGN_RETURN_ON_EXCEPTION(
1043 : isolate, number_format,
1044 : New<JSNumberFormat>(isolate, constructor, locales, options), String);
1045 :
1046 665 : if (can_cache) {
1047 : isolate->set_icu_object_in_cache(
1048 : Isolate::ICUObjectCacheType::kDefaultNumberFormat,
1049 : std::static_pointer_cast<icu::UObject>(
1050 140 : number_format->icu_number_format()->get()));
1051 : }
1052 :
1053 : // Return FormatNumber(numberFormat, x).
1054 : icu::NumberFormat* icu_number_format =
1055 1330 : number_format->icu_number_format()->raw();
1056 : return JSNumberFormat::FormatNumeric(isolate, *icu_number_format,
1057 665 : numeric_obj);
1058 : }
1059 :
1060 : namespace {
1061 :
1062 : // ecma402/#sec-defaultnumberoption
1063 6423 : Maybe<int> DefaultNumberOption(Isolate* isolate, Handle<Object> value, int min,
1064 : int max, int fallback, Handle<String> property) {
1065 : // 2. Else, return fallback.
1066 12846 : if (value->IsUndefined()) return Just(fallback);
1067 :
1068 : // 1. If value is not undefined, then
1069 : // a. Let value be ? ToNumber(value).
1070 : Handle<Object> value_num;
1071 936 : ASSIGN_RETURN_ON_EXCEPTION_VALUE(
1072 : isolate, value_num, Object::ToNumber(isolate, value), Nothing<int>());
1073 : DCHECK(value_num->IsNumber());
1074 :
1075 : // b. If value is NaN or less than minimum or greater than maximum, throw a
1076 : // RangeError exception.
1077 2277 : if (value_num->IsNaN() || value_num->Number() < min ||
1078 873 : value_num->Number() > max) {
1079 216 : THROW_NEW_ERROR_RETURN_VALUE(
1080 : isolate,
1081 : NewRangeError(MessageTemplate::kPropertyValueOutOfRange, property),
1082 : Nothing<int>());
1083 : }
1084 :
1085 : // The max and min arguments are integers and the above check makes
1086 : // sure that we are within the integer range making this double to
1087 : // int conversion safe.
1088 : //
1089 : // c. Return floor(value).
1090 720 : return Just(FastD2I(floor(value_num->Number())));
1091 : }
1092 :
1093 : // ecma402/#sec-getnumberoption
1094 6297 : Maybe<int> GetNumberOption(Isolate* isolate, Handle<JSReceiver> options,
1095 : Handle<String> property, int min, int max,
1096 : int fallback) {
1097 : // 1. Let value be ? Get(options, property).
1098 : Handle<Object> value;
1099 12594 : ASSIGN_RETURN_ON_EXCEPTION_VALUE(
1100 : isolate, value, JSReceiver::GetProperty(isolate, options, property),
1101 : Nothing<int>());
1102 :
1103 : // Return ? DefaultNumberOption(value, minimum, maximum, fallback).
1104 6297 : return DefaultNumberOption(isolate, value, min, max, fallback, property);
1105 : }
1106 :
1107 6297 : Maybe<int> GetNumberOption(Isolate* isolate, Handle<JSReceiver> options,
1108 : const char* property, int min, int max,
1109 : int fallback) {
1110 : Handle<String> property_str =
1111 6297 : isolate->factory()->NewStringFromAsciiChecked(property);
1112 6297 : return GetNumberOption(isolate, options, property_str, min, max, fallback);
1113 : }
1114 :
1115 : } // namespace
1116 :
1117 2132 : Maybe<bool> Intl::SetNumberFormatDigitOptions(Isolate* isolate,
1118 : icu::DecimalFormat* number_format,
1119 : Handle<JSReceiver> options,
1120 : int mnfd_default,
1121 : int mxfd_default) {
1122 2132 : CHECK_NOT_NULL(number_format);
1123 :
1124 : // 5. Let mnid be ? GetNumberOption(options, "minimumIntegerDigits,", 1, 21,
1125 : // 1).
1126 : int mnid;
1127 2132 : if (!GetNumberOption(isolate, options, "minimumIntegerDigits", 1, 21, 1)
1128 4264 : .To(&mnid)) {
1129 : return Nothing<bool>();
1130 : }
1131 :
1132 : // 6. Let mnfd be ? GetNumberOption(options, "minimumFractionDigits", 0, 20,
1133 : // mnfdDefault).
1134 : int mnfd;
1135 2087 : if (!GetNumberOption(isolate, options, "minimumFractionDigits", 0, 20,
1136 : mnfd_default)
1137 4174 : .To(&mnfd)) {
1138 : return Nothing<bool>();
1139 : }
1140 :
1141 : // 7. Let mxfdActualDefault be max( mnfd, mxfdDefault ).
1142 2078 : int mxfd_actual_default = std::max(mnfd, mxfd_default);
1143 :
1144 : // 8. Let mxfd be ? GetNumberOption(options,
1145 : // "maximumFractionDigits", mnfd, 20, mxfdActualDefault).
1146 : int mxfd;
1147 2078 : if (!GetNumberOption(isolate, options, "maximumFractionDigits", mnfd, 20,
1148 : mxfd_actual_default)
1149 4156 : .To(&mxfd)) {
1150 : return Nothing<bool>();
1151 : }
1152 :
1153 : // 9. Let mnsd be ? Get(options, "minimumSignificantDigits").
1154 : Handle<Object> mnsd_obj;
1155 : Handle<String> mnsd_str =
1156 : isolate->factory()->minimumSignificantDigits_string();
1157 4138 : ASSIGN_RETURN_ON_EXCEPTION_VALUE(
1158 : isolate, mnsd_obj, JSReceiver::GetProperty(isolate, options, mnsd_str),
1159 : Nothing<bool>());
1160 :
1161 : // 10. Let mxsd be ? Get(options, "maximumSignificantDigits").
1162 : Handle<Object> mxsd_obj;
1163 : Handle<String> mxsd_str =
1164 : isolate->factory()->maximumSignificantDigits_string();
1165 4138 : ASSIGN_RETURN_ON_EXCEPTION_VALUE(
1166 : isolate, mxsd_obj, JSReceiver::GetProperty(isolate, options, mxsd_str),
1167 : Nothing<bool>());
1168 :
1169 : // 11. Set intlObj.[[MinimumIntegerDigits]] to mnid.
1170 2069 : number_format->setMinimumIntegerDigits(mnid);
1171 :
1172 : // 12. Set intlObj.[[MinimumFractionDigits]] to mnfd.
1173 2069 : number_format->setMinimumFractionDigits(mnfd);
1174 :
1175 : // 13. Set intlObj.[[MaximumFractionDigits]] to mxfd.
1176 2069 : number_format->setMaximumFractionDigits(mxfd);
1177 :
1178 : bool significant_digits_used = false;
1179 : // 14. If mnsd is not undefined or mxsd is not undefined, then
1180 8186 : if (!mnsd_obj->IsUndefined(isolate) || !mxsd_obj->IsUndefined(isolate)) {
1181 : // 14. a. Let mnsd be ? DefaultNumberOption(mnsd, 1, 21, 1).
1182 : int mnsd;
1183 144 : if (!DefaultNumberOption(isolate, mnsd_obj, 1, 21, 1, mnsd_str).To(&mnsd)) {
1184 : return Nothing<bool>();
1185 : }
1186 :
1187 : // 14. b. Let mxsd be ? DefaultNumberOption(mxsd, mnsd, 21, 21).
1188 : int mxsd;
1189 54 : if (!DefaultNumberOption(isolate, mxsd_obj, mnsd, 21, 21, mxsd_str)
1190 108 : .To(&mxsd)) {
1191 : return Nothing<bool>();
1192 : }
1193 :
1194 : significant_digits_used = true;
1195 :
1196 : // 14. c. Set intlObj.[[MinimumSignificantDigits]] to mnsd.
1197 27 : number_format->setMinimumSignificantDigits(mnsd);
1198 :
1199 : // 14. d. Set intlObj.[[MaximumSignificantDigits]] to mxsd.
1200 27 : number_format->setMaximumSignificantDigits(mxsd);
1201 : }
1202 :
1203 2024 : number_format->setSignificantDigitsUsed(significant_digits_used);
1204 2024 : number_format->setRoundingMode(icu::DecimalFormat::kRoundHalfUp);
1205 : return Just(true);
1206 : }
1207 :
1208 : namespace {
1209 :
1210 : // ecma402/#sec-bestavailablelocale
1211 13286 : std::string BestAvailableLocale(const std::set<std::string>& available_locales,
1212 : const std::string& locale) {
1213 : // 1. Let candidate be locale.
1214 13286 : std::string candidate = locale;
1215 :
1216 : // 2. Repeat,
1217 : while (true) {
1218 : // 2.a. If availableLocales contains an element equal to candidate, return
1219 : // candidate.
1220 13718 : if (available_locales.find(candidate) != available_locales.end()) {
1221 : return candidate;
1222 : }
1223 :
1224 : // 2.b. Let pos be the character index of the last occurrence of "-"
1225 : // (U+002D) within candidate. If that character does not occur, return
1226 : // undefined.
1227 675 : size_t pos = candidate.rfind('-');
1228 675 : if (pos == std::string::npos) {
1229 : return std::string();
1230 : }
1231 :
1232 : // 2.c. If pos ≥ 2 and the character "-" occurs at index pos-2 of candidate,
1233 : // decrease pos by 2.
1234 810 : if (pos >= 2 && candidate[pos - 2] == '-') {
1235 90 : pos -= 2;
1236 : }
1237 :
1238 : // 2.d. Let candidate be the substring of candidate from position 0,
1239 : // inclusive, to position pos, exclusive.
1240 864 : candidate = candidate.substr(0, pos);
1241 : }
1242 : }
1243 :
1244 26572 : struct ParsedLocale {
1245 : std::string no_extensions_locale;
1246 : std::string extension;
1247 : };
1248 :
1249 : // Returns a struct containing a bcp47 tag without unicode extensions
1250 : // and the removed unicode extensions.
1251 : //
1252 : // For example, given 'en-US-u-co-emoji' returns 'en-US' and
1253 : // 'u-co-emoji'.
1254 13286 : ParsedLocale ParseBCP47Locale(const std::string& locale) {
1255 : size_t length = locale.length();
1256 : ParsedLocale parsed_locale;
1257 :
1258 : // Privateuse or grandfathered locales have no extension sequences.
1259 26572 : if ((length > 1) && (locale[1] == '-')) {
1260 : // Check to make sure that this really is a grandfathered or
1261 : // privateuse extension. ICU can sometimes mess up the
1262 : // canonicalization.
1263 63 : CHECK(locale[0] == 'x' || locale[0] == 'i');
1264 63 : parsed_locale.no_extensions_locale = locale;
1265 : return parsed_locale;
1266 : }
1267 :
1268 13223 : size_t unicode_extension_start = locale.find("-u-");
1269 :
1270 : // No unicode extensions found.
1271 13223 : if (unicode_extension_start == std::string::npos) {
1272 10194 : parsed_locale.no_extensions_locale = locale;
1273 : return parsed_locale;
1274 : }
1275 :
1276 3029 : size_t private_extension_start = locale.find("-x-");
1277 :
1278 : // Unicode extensions found within privateuse subtags don't count.
1279 6058 : if (private_extension_start != std::string::npos &&
1280 3029 : private_extension_start < unicode_extension_start) {
1281 36 : parsed_locale.no_extensions_locale = locale;
1282 : return parsed_locale;
1283 : }
1284 :
1285 2993 : const std::string beginning = locale.substr(0, unicode_extension_start);
1286 : size_t unicode_extension_end = length;
1287 : DCHECK_GT(length, 2);
1288 :
1289 : // Find the end of the extension production as per the bcp47 grammar
1290 : // by looking for '-' followed by 2 chars and then another '-'.
1291 24315 : for (size_t i = unicode_extension_start + 1; i < length - 2; i++) {
1292 21331 : if (locale[i] != '-') continue;
1293 :
1294 19902 : if (locale[i + 2] == '-') {
1295 : unicode_extension_end = i;
1296 : break;
1297 : }
1298 :
1299 : i += 2;
1300 : }
1301 :
1302 2993 : const std::string end = locale.substr(unicode_extension_end);
1303 5986 : parsed_locale.no_extensions_locale = beginning + end;
1304 5986 : parsed_locale.extension = locale.substr(
1305 : unicode_extension_start, unicode_extension_end - unicode_extension_start);
1306 : return parsed_locale;
1307 : }
1308 :
1309 : // ecma402/#sec-lookupsupportedlocales
1310 585 : std::vector<std::string> LookupSupportedLocales(
1311 : const std::set<std::string>& available_locales,
1312 : const std::vector<std::string>& requested_locales) {
1313 : // 1. Let subset be a new empty List.
1314 : std::vector<std::string> subset;
1315 :
1316 : // 2. For each element locale of requestedLocales in List order, do
1317 1719 : for (const std::string& locale : requested_locales) {
1318 : // 2. a. Let noExtensionsLocale be the String value that is locale
1319 : // with all Unicode locale extension sequences removed.
1320 : std::string no_extension_locale =
1321 1098 : ParseBCP47Locale(locale).no_extensions_locale;
1322 :
1323 : // 2. b. Let availableLocale be
1324 : // BestAvailableLocale(availableLocales, noExtensionsLocale).
1325 : std::string available_locale =
1326 549 : BestAvailableLocale(available_locales, no_extension_locale);
1327 :
1328 : // 2. c. If availableLocale is not undefined, append locale to the
1329 : // end of subset.
1330 549 : if (!available_locale.empty()) {
1331 441 : subset.push_back(locale);
1332 : }
1333 : }
1334 :
1335 : // 3. Return subset.
1336 585 : return subset;
1337 : }
1338 :
1339 : // ECMA 402 9.2.8 BestFitSupportedLocales(availableLocales, requestedLocales)
1340 : // https://tc39.github.io/ecma402/#sec-bestfitsupportedlocales
1341 : std::vector<std::string> BestFitSupportedLocales(
1342 : const std::set<std::string>& available_locales,
1343 : const std::vector<std::string>& requested_locales) {
1344 549 : return LookupSupportedLocales(available_locales, requested_locales);
1345 : }
1346 :
1347 : // ecma262 #sec-createarrayfromlist
1348 918 : Handle<JSArray> CreateArrayFromList(Isolate* isolate,
1349 1710 : std::vector<std::string> elements,
1350 : PropertyAttributes attr) {
1351 : Factory* factory = isolate->factory();
1352 : // Let array be ! ArrayCreate(0).
1353 918 : Handle<JSArray> array = factory->NewJSArray(0);
1354 :
1355 918 : uint32_t length = static_cast<uint32_t>(elements.size());
1356 : // 3. Let n be 0.
1357 : // 4. For each element e of elements, do
1358 1710 : for (uint32_t i = 0; i < length; i++) {
1359 : // a. Let status be CreateDataProperty(array, ! ToString(n), e).
1360 792 : const std::string& part = elements[i];
1361 : Handle<String> value =
1362 1584 : factory->NewStringFromUtf8(CStrVector(part.c_str())).ToHandleChecked();
1363 792 : JSObject::AddDataElement(array, i, value, attr);
1364 : }
1365 : // 5. Return array.
1366 918 : return array;
1367 : }
1368 :
1369 : // ECMA 402 9.2.9 SupportedLocales(availableLocales, requestedLocales, options)
1370 : // https://tc39.github.io/ecma402/#sec-supportedlocales
1371 621 : MaybeHandle<JSObject> SupportedLocales(
1372 : Isolate* isolate, const char* method,
1373 : const std::set<std::string>& available_locales,
1374 : const std::vector<std::string>& requested_locales, Handle<Object> options) {
1375 : std::vector<std::string> supported_locales;
1376 :
1377 : // 2. Else, let matcher be "best fit".
1378 : Intl::MatcherOption matcher = Intl::MatcherOption::kBestFit;
1379 :
1380 : // 1. If options is not undefined, then
1381 1242 : if (!options->IsUndefined(isolate)) {
1382 : // 1. a. Let options be ? ToObject(options).
1383 : Handle<JSReceiver> options_obj;
1384 216 : ASSIGN_RETURN_ON_EXCEPTION(isolate, options_obj,
1385 : Object::ToObject(isolate, options), JSObject);
1386 :
1387 : // 1. b. Let matcher be ? GetOption(options, "localeMatcher", "string",
1388 : // « "lookup", "best fit" », "best fit").
1389 : Maybe<Intl::MatcherOption> maybe_locale_matcher =
1390 108 : Intl::GetLocaleMatcher(isolate, options_obj, method);
1391 108 : MAYBE_RETURN(maybe_locale_matcher, MaybeHandle<JSObject>());
1392 : matcher = maybe_locale_matcher.FromJust();
1393 : }
1394 :
1395 : // 3. If matcher is "best fit", then
1396 : // a. Let supportedLocales be BestFitSupportedLocales(availableLocales,
1397 : // requestedLocales).
1398 585 : if (matcher == Intl::MatcherOption::kBestFit) {
1399 549 : supported_locales =
1400 : BestFitSupportedLocales(available_locales, requested_locales);
1401 : } else {
1402 : // 4. Else,
1403 : // a. Let supportedLocales be LookupSupportedLocales(availableLocales,
1404 : // requestedLocales).
1405 : DCHECK_EQ(matcher, Intl::MatcherOption::kLookup);
1406 72 : supported_locales =
1407 : LookupSupportedLocales(available_locales, requested_locales);
1408 : }
1409 :
1410 : // 5. Return CreateArrayFromList(supportedLocales).
1411 : PropertyAttributes attr = static_cast<PropertyAttributes>(NONE);
1412 585 : return CreateArrayFromList(isolate, supported_locales, attr);
1413 : }
1414 :
1415 : } // namespace
1416 :
1417 : // ecma-402 #sec-intl.getcanonicallocales
1418 342 : MaybeHandle<JSArray> Intl::GetCanonicalLocales(Isolate* isolate,
1419 : Handle<Object> locales) {
1420 : // 1. Let ll be ? CanonicalizeLocaleList(locales).
1421 : Maybe<std::vector<std::string>> maybe_ll =
1422 342 : CanonicalizeLocaleList(isolate, locales, false);
1423 342 : MAYBE_RETURN(maybe_ll, MaybeHandle<JSArray>());
1424 :
1425 : // 2. Return CreateArrayFromList(ll).
1426 : PropertyAttributes attr = static_cast<PropertyAttributes>(NONE);
1427 333 : return CreateArrayFromList(isolate, maybe_ll.FromJust(), attr);
1428 : }
1429 :
1430 : // ECMA 402 Intl.*.supportedLocalesOf
1431 657 : MaybeHandle<JSObject> Intl::SupportedLocalesOf(
1432 : Isolate* isolate, const char* method,
1433 : const std::set<std::string>& available_locales, Handle<Object> locales,
1434 : Handle<Object> options) {
1435 : // Let availableLocales be %Collator%.[[AvailableLocales]].
1436 :
1437 : // Let requestedLocales be ? CanonicalizeLocaleList(locales).
1438 : Maybe<std::vector<std::string>> requested_locales =
1439 657 : CanonicalizeLocaleList(isolate, locales, false);
1440 657 : MAYBE_RETURN(requested_locales, MaybeHandle<JSObject>());
1441 :
1442 : // Return ? SupportedLocales(availableLocales, requestedLocales, options).
1443 : return SupportedLocales(isolate, method, available_locales,
1444 621 : requested_locales.FromJust(), options);
1445 : }
1446 :
1447 : namespace {
1448 : template <typename T>
1449 720 : bool IsValidExtension(const icu::Locale& locale, const char* key,
1450 : const std::string& value) {
1451 720 : UErrorCode status = U_ZERO_ERROR;
1452 : std::unique_ptr<icu::StringEnumeration> enumeration(
1453 : T::getKeywordValuesForLocale(key, icu::Locale(locale.getBaseName()),
1454 720 : false, status));
1455 720 : if (U_SUCCESS(status)) {
1456 : int32_t length;
1457 720 : std::string legacy_type(uloc_toLegacyType(key, value.c_str()));
1458 5868 : for (const char* item = enumeration->next(&length, status); item != nullptr;
1459 5148 : item = enumeration->next(&length, status)) {
1460 11610 : if (U_SUCCESS(status) && legacy_type == item) {
1461 : return true;
1462 : }
1463 : }
1464 : }
1465 : return false;
1466 : }
1467 :
1468 : bool IsValidCalendar(const icu::Locale& locale, const std::string& value) {
1469 603 : return IsValidExtension<icu::Calendar>(locale, "calendar", value);
1470 : }
1471 :
1472 189 : bool IsValidCollation(const icu::Locale& locale, const std::string& value) {
1473 378 : std::set<std::string> invalid_values = {"standard", "search"};
1474 189 : if (invalid_values.find(value) != invalid_values.end()) return false;
1475 117 : return IsValidExtension<icu::Collator>(locale, "collation", value);
1476 : }
1477 :
1478 1386 : bool IsValidNumberingSystem(const std::string& value) {
1479 2772 : std::set<std::string> invalid_values = {"native", "traditio", "finance"};
1480 1386 : if (invalid_values.find(value) != invalid_values.end()) return false;
1481 1332 : UErrorCode status = U_ZERO_ERROR;
1482 : std::unique_ptr<icu::NumberingSystem> numbering_system(
1483 1332 : icu::NumberingSystem::createInstanceByName(value.c_str(), status));
1484 1332 : return U_SUCCESS(status) && numbering_system.get() != nullptr;
1485 : }
1486 :
1487 14493 : std::map<std::string, std::string> LookupAndValidateUnicodeExtensions(
1488 : icu::Locale* icu_locale, const std::set<std::string>& relevant_keys) {
1489 : std::map<std::string, std::string> extensions;
1490 :
1491 14493 : UErrorCode status = U_ZERO_ERROR;
1492 : std::unique_ptr<icu::StringEnumeration> keywords(
1493 14493 : icu_locale->createKeywords(status));
1494 14493 : if (U_FAILURE(status)) return extensions;
1495 :
1496 14493 : if (!keywords) return extensions;
1497 : char value[ULOC_FULLNAME_CAPACITY];
1498 :
1499 : int32_t length;
1500 2957 : status = U_ZERO_ERROR;
1501 7645 : for (const char* keyword = keywords->next(&length, status);
1502 4688 : keyword != nullptr; keyword = keywords->next(&length, status)) {
1503 : // Ignore failures in ICU and skip to the next keyword.
1504 : //
1505 : // This is fine.™
1506 4688 : if (U_FAILURE(status)) {
1507 0 : status = U_ZERO_ERROR;
1508 0 : continue;
1509 : }
1510 :
1511 4688 : icu_locale->getKeywordValue(keyword, value, ULOC_FULLNAME_CAPACITY, status);
1512 :
1513 : // Ignore failures in ICU and skip to the next keyword.
1514 : //
1515 : // This is fine.™
1516 4688 : if (U_FAILURE(status)) {
1517 0 : status = U_ZERO_ERROR;
1518 0 : continue;
1519 : }
1520 :
1521 4688 : const char* bcp47_key = uloc_toUnicodeLocaleKey(keyword);
1522 :
1523 18725 : if (bcp47_key && (relevant_keys.find(bcp47_key) != relevant_keys.end())) {
1524 2426 : const char* bcp47_value = uloc_toUnicodeLocaleType(bcp47_key, value);
1525 : bool is_valid_value = false;
1526 : // 8.h.ii.1.a If keyLocaleData contains requestedValue, then
1527 2426 : if (strcmp("ca", bcp47_key) == 0) {
1528 1206 : is_valid_value = IsValidCalendar(*icu_locale, bcp47_value);
1529 1823 : } else if (strcmp("co", bcp47_key) == 0) {
1530 378 : is_valid_value = IsValidCollation(*icu_locale, bcp47_value);
1531 1634 : } else if (strcmp("hc", bcp47_key) == 0) {
1532 : // https://www.unicode.org/repos/cldr/tags/latest/common/bcp47/calendar.xml
1533 396 : std::set<std::string> valid_values = {"h11", "h12", "h23", "h24"};
1534 396 : is_valid_value = valid_values.find(bcp47_value) != valid_values.end();
1535 1436 : } else if (strcmp("lb", bcp47_key) == 0) {
1536 : // https://www.unicode.org/repos/cldr/tags/latest/common/bcp47/segmentation.xml
1537 0 : std::set<std::string> valid_values = {"strict", "normal", "loose"};
1538 0 : is_valid_value = valid_values.find(bcp47_value) != valid_values.end();
1539 1436 : } else if (strcmp("kn", bcp47_key) == 0) {
1540 : // https://www.unicode.org/repos/cldr/tags/latest/common/bcp47/collation.xml
1541 50 : std::set<std::string> valid_values = {"true", "false"};
1542 50 : is_valid_value = valid_values.find(bcp47_value) != valid_values.end();
1543 1411 : } else if (strcmp("kf", bcp47_key) == 0) {
1544 : // https://www.unicode.org/repos/cldr/tags/latest/common/bcp47/collation.xml
1545 50 : std::set<std::string> valid_values = {"upper", "lower", "false"};
1546 50 : is_valid_value = valid_values.find(bcp47_value) != valid_values.end();
1547 1386 : } else if (strcmp("nu", bcp47_key) == 0) {
1548 2772 : is_valid_value = IsValidNumberingSystem(bcp47_value);
1549 : }
1550 2426 : if (is_valid_value) {
1551 : extensions.insert(
1552 4034 : std::pair<std::string, std::string>(bcp47_key, bcp47_value));
1553 2017 : continue;
1554 : }
1555 : }
1556 2671 : status = U_ZERO_ERROR;
1557 2671 : icu_locale->setKeywordValue(keyword, nullptr, status);
1558 5342 : CHECK(U_SUCCESS(status));
1559 : }
1560 :
1561 : return extensions;
1562 : }
1563 :
1564 : // ecma402/#sec-lookupmatcher
1565 14493 : std::string LookupMatcher(Isolate* isolate,
1566 : const std::set<std::string>& available_locales,
1567 : const std::vector<std::string>& requested_locales) {
1568 : // 1. Let result be a new Record.
1569 : std::string result;
1570 :
1571 : // 2. For each element locale of requestedLocales in List order, do
1572 29121 : for (const std::string& locale : requested_locales) {
1573 : // 2. a. Let noExtensionsLocale be the String value that is locale
1574 : // with all Unicode locale extension sequences removed.
1575 12737 : ParsedLocale parsed_locale = ParseBCP47Locale(locale);
1576 12737 : std::string no_extensions_locale = parsed_locale.no_extensions_locale;
1577 :
1578 : // 2. b. Let availableLocale be
1579 : // BestAvailableLocale(availableLocales, noExtensionsLocale).
1580 : std::string available_locale =
1581 12737 : BestAvailableLocale(available_locales, no_extensions_locale);
1582 :
1583 : // 2. c. If availableLocale is not undefined, append locale to the
1584 : // end of subset.
1585 12737 : if (!available_locale.empty()) {
1586 : // Note: The following steps are not performed here because we
1587 : // can use ICU to parse the unicode locale extension sequence
1588 : // as part of Intl::ResolveLocale.
1589 : //
1590 : // There's no need to separate the unicode locale extensions
1591 : // right here. Instead just return the available locale with the
1592 : // extensions.
1593 : //
1594 : // 2. c. i. Set result.[[locale]] to availableLocale.
1595 : // 2. c. ii. If locale and noExtensionsLocale are not the same
1596 : // String value, then
1597 : // 2. c. ii. 1. Let extension be the String value consisting of
1598 : // the first substring of locale that is a Unicode locale
1599 : // extension sequence.
1600 : // 2. c. ii. 2. Set result.[[extension]] to extension.
1601 : // 2. c. iii. Return result.
1602 : return available_locale + parsed_locale.extension;
1603 : }
1604 135 : }
1605 :
1606 : // 3. Let defLocale be DefaultLocale();
1607 : // 4. Set result.[[locale]] to defLocale.
1608 : // 5. Return result.
1609 1891 : return DefaultLocale(isolate);
1610 : }
1611 :
1612 : } // namespace
1613 :
1614 : // This function doesn't correspond exactly with the spec. Instead
1615 : // we use ICU to do all the string manipulations that the spec
1616 : // peforms.
1617 : //
1618 : // The spec uses this function to normalize values for various
1619 : // relevant extension keys (such as disallowing "search" for
1620 : // collation). Instead of doing this here, we let the callers of
1621 : // this method perform such normalization.
1622 : //
1623 : // ecma402/#sec-resolvelocale
1624 14493 : Intl::ResolvedLocale Intl::ResolveLocale(
1625 : Isolate* isolate, const std::set<std::string>& available_locales,
1626 : const std::vector<std::string>& requested_locales, MatcherOption matcher,
1627 : const std::set<std::string>& relevant_extension_keys) {
1628 : std::string locale;
1629 14493 : if (matcher == Intl::MatcherOption::kLookup) {
1630 28662 : locale = LookupMatcher(isolate, available_locales, requested_locales);
1631 162 : } else if (matcher == Intl::MatcherOption::kBestFit) {
1632 : // TODO(intl): Implement better lookup algorithm.
1633 324 : locale = LookupMatcher(isolate, available_locales, requested_locales);
1634 : }
1635 :
1636 28986 : icu::Locale icu_locale = CreateICULocale(locale);
1637 : std::map<std::string, std::string> extensions =
1638 14493 : LookupAndValidateUnicodeExtensions(&icu_locale, relevant_extension_keys);
1639 :
1640 28986 : std::string canonicalized_locale = Intl::ToLanguageTag(icu_locale).FromJust();
1641 :
1642 : // TODO(gsathya): Remove privateuse subtags from extensions.
1643 :
1644 28986 : return Intl::ResolvedLocale{canonicalized_locale, icu_locale, extensions};
1645 : }
1646 :
1647 2813 : Managed<icu::UnicodeString> Intl::SetTextToBreakIterator(
1648 : Isolate* isolate, Handle<String> text, icu::BreakIterator* break_iterator) {
1649 : icu::UnicodeString* u_text =
1650 2813 : (icu::UnicodeString*)(Intl::ToICUUnicodeString(isolate, text).clone());
1651 :
1652 : Handle<Managed<icu::UnicodeString>> new_u_text =
1653 2813 : Managed<icu::UnicodeString>::FromRawPtr(isolate, 0, u_text);
1654 :
1655 2813 : break_iterator->setText(*u_text);
1656 2813 : return *new_u_text;
1657 : }
1658 :
1659 : // ecma262 #sec-string.prototype.normalize
1660 2313 : MaybeHandle<String> Intl::Normalize(Isolate* isolate, Handle<String> string,
1661 : Handle<Object> form_input) {
1662 : const char* form_name;
1663 : UNormalization2Mode form_mode;
1664 4626 : if (form_input->IsUndefined(isolate)) {
1665 : // default is FNC
1666 : form_name = "nfc";
1667 : form_mode = UNORM2_COMPOSE;
1668 : } else {
1669 : Handle<String> form;
1670 3780 : ASSIGN_RETURN_ON_EXCEPTION(isolate, form,
1671 : Object::ToString(isolate, form_input), String);
1672 :
1673 1890 : if (String::Equals(isolate, form, isolate->factory()->NFC_string())) {
1674 : form_name = "nfc";
1675 : form_mode = UNORM2_COMPOSE;
1676 1476 : } else if (String::Equals(isolate, form,
1677 : isolate->factory()->NFD_string())) {
1678 : form_name = "nfc";
1679 : form_mode = UNORM2_DECOMPOSE;
1680 1071 : } else if (String::Equals(isolate, form,
1681 : isolate->factory()->NFKC_string())) {
1682 : form_name = "nfkc";
1683 : form_mode = UNORM2_COMPOSE;
1684 657 : } else if (String::Equals(isolate, form,
1685 : isolate->factory()->NFKD_string())) {
1686 : form_name = "nfkc";
1687 : form_mode = UNORM2_DECOMPOSE;
1688 : } else {
1689 : Handle<String> valid_forms =
1690 99 : isolate->factory()->NewStringFromStaticChars("NFC, NFD, NFKC, NFKD");
1691 99 : THROW_NEW_ERROR(
1692 : isolate,
1693 : NewRangeError(MessageTemplate::kNormalizationForm, valid_forms),
1694 : String);
1695 : }
1696 : }
1697 :
1698 : int length = string->length();
1699 2214 : string = String::Flatten(isolate, string);
1700 : icu::UnicodeString result;
1701 : std::unique_ptr<uc16[]> sap;
1702 2214 : UErrorCode status = U_ZERO_ERROR;
1703 4428 : icu::UnicodeString input = ToICUUnicodeString(isolate, string);
1704 : // Getting a singleton. Should not free it.
1705 : const icu::Normalizer2* normalizer =
1706 2214 : icu::Normalizer2::getInstance(nullptr, form_name, form_mode, status);
1707 : DCHECK(U_SUCCESS(status));
1708 2214 : CHECK_NOT_NULL(normalizer);
1709 : int32_t normalized_prefix_length =
1710 2214 : normalizer->spanQuickCheckYes(input, status);
1711 : // Quick return if the input is already normalized.
1712 2214 : if (length == normalized_prefix_length) return string;
1713 : icu::UnicodeString unnormalized =
1714 1926 : input.tempSubString(normalized_prefix_length);
1715 : // Read-only alias of the normalized prefix.
1716 1926 : result.setTo(false, input.getBuffer(), normalized_prefix_length);
1717 : // copy-on-write; normalize the suffix and append to |result|.
1718 963 : normalizer->normalizeSecondAndAppend(result, unnormalized, status);
1719 :
1720 963 : if (U_FAILURE(status)) {
1721 0 : THROW_NEW_ERROR(isolate, NewTypeError(MessageTemplate::kIcuError), String);
1722 : }
1723 :
1724 3177 : return Intl::ToString(isolate, result);
1725 : }
1726 :
1727 : // ICUTimezoneCache calls out to ICU for TimezoneCache
1728 : // functionality in a straightforward way.
1729 : class ICUTimezoneCache : public base::TimezoneCache {
1730 : public:
1731 122086 : ICUTimezoneCache() : timezone_(nullptr) { Clear(TimeZoneDetection::kSkip); }
1732 :
1733 183087 : ~ICUTimezoneCache() override { Clear(TimeZoneDetection::kSkip); }
1734 :
1735 : const char* LocalTimezone(double time_ms) override;
1736 :
1737 : double DaylightSavingsOffset(double time_ms) override;
1738 :
1739 : double LocalTimeOffset(double time_ms, bool is_utc) override;
1740 :
1741 : void Clear(TimeZoneDetection time_zone_detection) override;
1742 :
1743 : private:
1744 : icu::TimeZone* GetTimeZone();
1745 :
1746 : bool GetOffsets(double time_ms, bool is_utc, int32_t* raw_offset,
1747 : int32_t* dst_offset);
1748 :
1749 : icu::TimeZone* timezone_;
1750 :
1751 : std::string timezone_name_;
1752 : std::string dst_timezone_name_;
1753 : };
1754 :
1755 130 : const char* ICUTimezoneCache::LocalTimezone(double time_ms) {
1756 130 : bool is_dst = DaylightSavingsOffset(time_ms) != 0;
1757 130 : std::string* name = is_dst ? &dst_timezone_name_ : &timezone_name_;
1758 130 : if (name->empty()) {
1759 : icu::UnicodeString result;
1760 260 : GetTimeZone()->getDisplayName(is_dst, icu::TimeZone::LONG, result);
1761 : result += '\0';
1762 :
1763 : icu::StringByteSink<std::string> byte_sink(name);
1764 260 : result.toUTF8(byte_sink);
1765 : }
1766 : DCHECK(!name->empty());
1767 130 : return name->c_str();
1768 : }
1769 :
1770 0 : icu::TimeZone* ICUTimezoneCache::GetTimeZone() {
1771 135312 : if (timezone_ == nullptr) {
1772 215 : timezone_ = icu::TimeZone::createDefault();
1773 : }
1774 135312 : return timezone_;
1775 : }
1776 :
1777 135182 : bool ICUTimezoneCache::GetOffsets(double time_ms, bool is_utc,
1778 : int32_t* raw_offset, int32_t* dst_offset) {
1779 135182 : UErrorCode status = U_ZERO_ERROR;
1780 : // TODO(jshin): ICU TimeZone class handles skipped time differently from
1781 : // Ecma 262 (https://github.com/tc39/ecma262/pull/778) and icu::TimeZone
1782 : // class does not expose the necessary API. Fixing
1783 : // http://bugs.icu-project.org/trac/ticket/13268 would make it easy to
1784 : // implement the proposed spec change. A proposed fix for ICU is
1785 : // https://chromium-review.googlesource.com/851265 .
1786 : // In the meantime, use an internal (still public) API of icu::BasicTimeZone.
1787 : // Once it's accepted by the upstream, get rid of cast. Note that casting
1788 : // TimeZone to BasicTimeZone is safe because we know that icu::TimeZone used
1789 : // here is a BasicTimeZone.
1790 135182 : if (is_utc) {
1791 132310 : GetTimeZone()->getOffset(time_ms, false, *raw_offset, *dst_offset, status);
1792 : } else {
1793 : static_cast<const icu::BasicTimeZone*>(GetTimeZone())
1794 : ->getOffsetFromLocal(time_ms, icu::BasicTimeZone::kFormer,
1795 : icu::BasicTimeZone::kFormer, *raw_offset,
1796 2872 : *dst_offset, status);
1797 : }
1798 :
1799 270364 : return U_SUCCESS(status);
1800 : }
1801 :
1802 19216 : double ICUTimezoneCache::DaylightSavingsOffset(double time_ms) {
1803 : int32_t raw_offset, dst_offset;
1804 19216 : if (!GetOffsets(time_ms, true, &raw_offset, &dst_offset)) return 0;
1805 19216 : return dst_offset;
1806 : }
1807 :
1808 115966 : double ICUTimezoneCache::LocalTimeOffset(double time_ms, bool is_utc) {
1809 : int32_t raw_offset, dst_offset;
1810 115966 : if (!GetOffsets(time_ms, is_utc, &raw_offset, &dst_offset)) return 0;
1811 115966 : return raw_offset + dst_offset;
1812 : }
1813 :
1814 183116 : void ICUTimezoneCache::Clear(TimeZoneDetection time_zone_detection) {
1815 183116 : delete timezone_;
1816 183116 : timezone_ = nullptr;
1817 183116 : timezone_name_.clear();
1818 183114 : dst_timezone_name_.clear();
1819 183116 : if (time_zone_detection == TimeZoneDetection::kRedetect) {
1820 0 : icu::TimeZone::adoptDefault(icu::TimeZone::detectHostTimeZone());
1821 : }
1822 183116 : }
1823 :
1824 61052 : base::TimezoneCache* Intl::CreateTimeZoneCache() {
1825 : return FLAG_icu_timezone_data ? new ICUTimezoneCache()
1826 61062 : : base::OS::CreateTimezoneCache();
1827 : }
1828 :
1829 6075 : Maybe<Intl::CaseFirst> Intl::GetCaseFirst(Isolate* isolate,
1830 : Handle<JSReceiver> options,
1831 : const char* method) {
1832 : return Intl::GetStringOption<Intl::CaseFirst>(
1833 : isolate, options, "caseFirst", method, {"upper", "lower", "false"},
1834 : {Intl::CaseFirst::kUpper, Intl::CaseFirst::kLower,
1835 : Intl::CaseFirst::kFalse},
1836 18225 : Intl::CaseFirst::kUndefined);
1837 : }
1838 :
1839 3693 : Maybe<Intl::HourCycle> Intl::GetHourCycle(Isolate* isolate,
1840 : Handle<JSReceiver> options,
1841 : const char* method) {
1842 : return Intl::GetStringOption<Intl::HourCycle>(
1843 : isolate, options, "hourCycle", method, {"h11", "h12", "h23", "h24"},
1844 : {Intl::HourCycle::kH11, Intl::HourCycle::kH12, Intl::HourCycle::kH23,
1845 : Intl::HourCycle::kH24},
1846 11079 : Intl::HourCycle::kUndefined);
1847 : }
1848 :
1849 14637 : Maybe<Intl::MatcherOption> Intl::GetLocaleMatcher(Isolate* isolate,
1850 : Handle<JSReceiver> options,
1851 : const char* method) {
1852 : return Intl::GetStringOption<Intl::MatcherOption>(
1853 : isolate, options, "localeMatcher", method, {"best fit", "lookup"},
1854 : {Intl::MatcherOption::kLookup, Intl::MatcherOption::kBestFit},
1855 43911 : Intl::MatcherOption::kLookup);
1856 : }
1857 :
1858 108 : Intl::HourCycle Intl::ToHourCycle(const std::string& hc) {
1859 108 : if (hc == "h11") return Intl::HourCycle::kH11;
1860 81 : if (hc == "h12") return Intl::HourCycle::kH12;
1861 54 : if (hc == "h23") return Intl::HourCycle::kH23;
1862 27 : if (hc == "h24") return Intl::HourCycle::kH24;
1863 0 : return Intl::HourCycle::kUndefined;
1864 : }
1865 :
1866 1221 : const std::set<std::string>& Intl::GetAvailableLocalesForLocale() {
1867 : static base::LazyInstance<Intl::AvailableLocales<icu::Locale>>::type
1868 : available_locales = LAZY_INSTANCE_INITIALIZER;
1869 1221 : return available_locales.Pointer()->Get();
1870 : }
1871 :
1872 4621 : const std::set<std::string>& Intl::GetAvailableLocalesForDateFormat() {
1873 : static base::LazyInstance<Intl::AvailableLocales<icu::DateFormat>>::type
1874 : available_locales = LAZY_INSTANCE_INITIALIZER;
1875 4621 : return available_locales.Pointer()->Get();
1876 : }
1877 :
1878 : } // namespace internal
1879 178779 : } // namespace v8
|
