Skip to content

V
V8
Commit 7f614962 authored by Brian Stell's avatar Brian Stell Committed by Commit Bot
Browse files
Options

[Intl] Use C++ versions of CanonicalizeLanguageTag and isStructuallyValidLanguageTag. 

Remove the JS versions.

Bug: v8:7954
Cq-Include-Trybots: luci.v8.try:v8_linux_noi18n_rel_ng
Change-Id: I1dc90932e570c4acc7ca32a807b452148a6b2d88
Reviewed-on: https://chromium-review.googlesource.com/1144524
Commit-Queue: Brian Stell <bstell@chromium.org>
Reviewed-by: 's avatarJakob Kummerow <jkummerow@chromium.org>
Cr-Commit-Position: refs/heads/master@{#54756}
parent ed8d35ce
Hide whitespace changes
Inline Side-by-side
Showing with 5 additions and 213 deletions
src/js/intl.js
View file @ 7f614962
......@@ -202,43 +202,6 @@ function GetServiceRE() {
return SERVICE_RE;
}
/**
* Validates a language tag against bcp47 spec.
* Actual value is assigned on first run.
*/
var LANGUAGE_TAG_RE = UNDEFINED;
function GetLanguageTagRE() {
if (IS_UNDEFINED(LANGUAGE_TAG_RE)) {
BuildLanguageTagREs();
}
return LANGUAGE_TAG_RE;
}
/**
* Helps find duplicate variants in the language tag.
*/
var LANGUAGE_VARIANT_RE = UNDEFINED;
function GetLanguageVariantRE() {
if (IS_UNDEFINED(LANGUAGE_VARIANT_RE)) {
BuildLanguageTagREs();
}
return LANGUAGE_VARIANT_RE;
}
/**
* Helps find duplicate singletons in the language tag.
*/
var LANGUAGE_SINGLETON_RE = UNDEFINED;
function GetLanguageSingletonRE() {
if (IS_UNDEFINED(LANGUAGE_SINGLETON_RE)) {
BuildLanguageTagREs();
}
return LANGUAGE_SINGLETON_RE;
}
/**
* Matches valid IANA time zone names.
*/
......@@ -775,37 +738,6 @@ function toTitleCaseTimezoneLocation(location) {
return result;
}
/**
* Canonicalizes the language tag, or throws in case the tag is invalid.
* ECMA 402 9.2.1 steps 7.c ii ~ v.
*/
function canonicalizeLanguageTag(localeID) {
// null is typeof 'object' so we have to do extra check.
if ((!IS_STRING(localeID) && !IS_RECEIVER(localeID)) ||
IS_NULL(localeID)) {
throw %make_type_error(kLanguageID);
}
var localeString = TO_STRING(localeID);
// Optimize for the most common case; a 2-letter language code in the
// canonical form/lowercase that is not one of deprecated codes
// (in, iw, ji, jw). Don't check for ~70 of 3-letter deprecated language
// codes. Instead, let them be handled by ICU in the slow path. Besides,
// fast-track 'fil' (3-letter canonical code).
if ((!IS_NULL(%regexp_internal_match(/^[a-z]{2}$/, localeString)) &&
IS_NULL(%regexp_internal_match(/^(in|iw|ji|jw)$/, localeString))) ||
localeString === "fil") {
return localeString;
}
if (isStructuallyValidLanguageTag(localeString) === false) {
throw %make_range_error(kInvalidLanguageTag, localeString);
}
// ECMA 402 6.2.3
return %CanonicalizeLanguageTag(localeString);
}
/**
* Returns an InternalArray where all locales are canonicalized and duplicates
......@@ -818,7 +750,7 @@ function canonicalizeLocaleList(locales) {
if (!IS_UNDEFINED(locales)) {
// We allow single string localeID.
if (typeof locales === 'string') {
%_Call(ArrayPush, seen, canonicalizeLanguageTag(locales));
%_Call(ArrayPush, seen, %CanonicalizeLanguageTag(locales));
return seen;
}
......@@ -829,7 +761,7 @@ function canonicalizeLocaleList(locales) {
if (k in o) {
var value = o[k];
var tag = canonicalizeLanguageTag(value);
var tag = %CanonicalizeLanguageTag(value);
if (%ArrayIndexOf(seen, tag, 0) === -1) {
%_Call(ArrayPush, seen, tag);
......@@ -852,109 +784,6 @@ function initializeLocaleList(locales) {
"initialize_locale_list", initializeLocaleList
]);
/**
* Check the structural Validity of the language tag per ECMA 402 6.2.2:
* - Well-formed per RFC 5646 2.1
* - There are no duplicate variant subtags
* - There are no duplicate singletion (extension) subtags
*
* One extra-check is done (from RFC 5646 2.2.9): the tag is compared
* against the list of grandfathered tags. However, subtags for
* primary/extended language, script, region, variant are not checked
* against the IANA language subtag registry.
*
* ICU is too permissible and lets invalid tags, like
* hant-cmn-cn, through.
*
* Returns false if the language tag is invalid.
*/
function isStructuallyValidLanguageTag(locale) {
// Check if it's well-formed, including grandfadered tags.
if (IS_NULL(%regexp_internal_match(GetLanguageTagRE(), locale))) {
return false;
}
locale = %StringToLowerCaseIntl(locale);
// Just return if it's a x- form. It's all private.
if (%StringIndexOf(locale, 'x-', 0) === 0) {
return true;
}
// Check if there are any duplicate variants or singletons (extensions).
// Remove private use section.
locale = %StringSplit(locale, '-x-', kMaxUint32)[0];
// Skip language since it can match variant regex, so we start from 1.
// We are matching i-klingon here, but that's ok, since i-klingon-klingon
// is not valid and would fail LANGUAGE_TAG_RE test.
var variants = new InternalArray();
var extensions = new InternalArray();
var parts = %StringSplit(locale, '-', kMaxUint32);
for (var i = 1; i < parts.length; i++) {
var value = parts[i];
if (!IS_NULL(%regexp_internal_match(GetLanguageVariantRE(), value)) &&
extensions.length === 0) {
if (%ArrayIndexOf(variants, value, 0) === -1) {
%_Call(ArrayPush, variants, value);
} else {
return false;
}
}
if (!IS_NULL(%regexp_internal_match(GetLanguageSingletonRE(), value))) {
if (%ArrayIndexOf(extensions, value, 0) === -1) {
%_Call(ArrayPush, extensions, value);
} else {
return false;
}
}
}
return true;
}
/**
* Builds a regular expresion that validates the language tag
* against bcp47 spec.
* Uses http://tools.ietf.org/html/bcp47, section 2.1, ABNF.
* Runs on load and initializes the global REs.
*/
function BuildLanguageTagREs() {
var alpha = '[a-zA-Z]';
var digit = '[0-9]';
var alphanum = '(' + alpha + '|' + digit + ')';
var regular = '(art-lojban|cel-gaulish|no-bok|no-nyn|zh-guoyu|zh-hakka|' +
'zh-min|zh-min-nan|zh-xiang)';
var irregular = '(en-GB-oed|i-ami|i-bnn|i-default|i-enochian|i-hak|' +
'i-klingon|i-lux|i-mingo|i-navajo|i-pwn|i-tao|i-tay|' +
'i-tsu|sgn-BE-FR|sgn-BE-NL|sgn-CH-DE)';
var grandfathered = '(' + irregular + '|' + regular + ')';
var privateUse = '(x(-' + alphanum + '{1,8})+)';
var singleton = '(' + digit + '|[A-WY-Za-wy-z])';
LANGUAGE_SINGLETON_RE = new GlobalRegExp('^' + singleton + '$', 'i');
var extension = '(' + singleton + '(-' + alphanum + '{2,8})+)';
var variant = '(' + alphanum + '{5,8}|(' + digit + alphanum + '{3}))';
LANGUAGE_VARIANT_RE = new GlobalRegExp('^' + variant + '$', 'i');
var region = '(' + alpha + '{2}|' + digit + '{3})';
var script = '(' + alpha + '{4})';
var extLang = '(' + alpha + '{3}(-' + alpha + '{3}){0,2})';
var language = '(' + alpha + '{2,3}(-' + extLang + ')?|' + alpha + '{4}|' +
alpha + '{5,8})';
var langTag = language + '(-' + script + ')?(-' + region + ')?(-' +
variant + ')*(-' + extension + ')*(-' + privateUse + ')?';
var languageTag =
'^(' + langTag + '|' + privateUse + '|' + grandfathered + ')$';
LANGUAGE_TAG_RE = new GlobalRegExp(languageTag, 'i');
}
// ECMA 402 section 8.2.1
DEFINE_METHOD(
GlobalIntl,
......
src/objects/intl-objects.cc
View file @ 7f614962
......@@ -1627,8 +1627,6 @@ bool IsStructurallyValidLanguageTag(Isolate* isolate,
std::vector<std::string> extensions;
for (auto it = parts.begin() + 1; it != parts.end(); it++) {
icu::UnicodeString part(it->data(), -1, US_INV);
// Note: icu::RegexMatcher::reset does not make a copy of the input string
// so cannot use a temp value; ie: cannot create it as a call parameter.
language_variant_regexp_matcher->reset(part);
bool is_language_variant = language_variant_regexp_matcher->matches(status);
if (V8_UNLIKELY(U_FAILURE(status))) {
......
src/runtime/runtime-intl.cc
View file @ 7f614962
......@@ -58,45 +58,10 @@ RUNTIME_FUNCTION(Runtime_CanonicalizeLanguageTag) {
HandleScope scope(isolate);
DCHECK_EQ(1, args.length());
CONVERT_ARG_HANDLE_CHECKED(String, locale_id_str, 0);
CONVERT_ARG_HANDLE_CHECKED(Object, locale, 0);
v8::Isolate* v8_isolate = reinterpret_cast<v8::Isolate*>(isolate);
Factory* factory = isolate->factory();
v8::String::Utf8Value locale_id(v8_isolate,
v8::Utils::ToLocal(locale_id_str));
// TODO(jshin): uloc_{for,to}TanguageTag can fail even for a structually valid
// language tag if it's too long (much longer than 100 chars). Even if we
// allocate a longer buffer, ICU will still fail if it's too long. Either
// propose to Ecma 402 to put a limit on the locale length or change ICU to
// handle long locale names better. See
// https://ssl.icu-project.org/trac/ticket/13417 .
UErrorCode error = U_ZERO_ERROR;
char icu_result[ULOC_FULLNAME_CAPACITY];
uloc_forLanguageTag(*locale_id, icu_result, ULOC_FULLNAME_CAPACITY, nullptr,
&error);
if (U_FAILURE(error) || error == U_STRING_NOT_TERMINATED_WARNING) {
THROW_NEW_ERROR_RETURN_FAILURE(
isolate,
NewRangeError(MessageTemplate::kInvalidLanguageTag, locale_id_str));
}
char result[ULOC_FULLNAME_CAPACITY];
// Force strict BCP47 rules.
uloc_toLanguageTag(icu_result, result, ULOC_FULLNAME_CAPACITY, TRUE, &error);
if (U_FAILURE(error) || error == U_STRING_NOT_TERMINATED_WARNING) {
THROW_NEW_ERROR_RETURN_FAILURE(
isolate,
NewRangeError(MessageTemplate::kInvalidLanguageTag, locale_id_str));
}
DCHECK_NOT_NULL(result);
return *factory->NewStringFromAsciiChecked(result);
RETURN_RESULT_OR_FAILURE(isolate,
Intl::CanonicalizeLanguageTag(isolate, locale));
}
RUNTIME_FUNCTION(Runtime_AvailableLocalesOf) {
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment