beenull/ladybird
1
0
Fork 0
mirror of https://github.com/LadybirdBrowser/ladybird.git synced 2024-09-30 08:41:15 +00:00
Code Issues Actions 5 Packages Projects Releases Wiki Activity

LibJS: Reject structurally invalid Unicode locale extensions

Browse source
This commit is contained in:
Timothy Flynn 2021-08-28 14:46:36 -04:00 committed by Linus Groh
parent f897c2edb3
commit 94e66f500c
Notes: sideshowbarker 2024-07-18 05:04:18 +09:00 
Author: https://github.com/trflynn89
Commit: https://github.com/SerenityOS/serenity/commit/94e66f500cc
Pull-request: https://github.com/SerenityOS/serenity/pull/9668
Reviewed-by: https://github.com/linusg
2 changed files with 59 additions and 12 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

48
Userland/Libraries/LibJS/Runtime/Intl/AbstractOperations.cpp
View file

@ -18,6 +18,20 @@ namespace JS::Intl {
// 6.2.2 IsStructurallyValidLanguageTag ( locale ), https://tc39.es/ecma402/#sec-isstructurallyvalidlanguagetag
static Optional<Unicode::LocaleID> is_structurally_valid_language_tag(StringView locale)
{
auto contains_duplicate_variant = [](Vector<StringView>& variants) {
if (variants.is_empty())
return false;
quick_sort(variants);
for (size_t i = 0; i < variants.size() - 1; ++i) {
if (variants[i] == variants[i + 1])
return true;
}
return false;
};
// IsStructurallyValidLanguageTag returns true if all of the following conditions hold, false otherwise:
// locale can be generated from the EBNF grammar for unicode_locale_id in Unicode Technical Standard #35 LDML § 3.2 Unicode Locale Identifier;
@ -31,23 +45,33 @@ static Optional<Unicode::LocaleID> is_structurally_valid_language_tag(StringView
return {};
// the unicode_language_id within locale contains no duplicate unicode_variant_subtag subtags; and
if (auto& variants = locale_id->language_id.variants; !variants.is_empty()) {
quick_sort(variants);
if (contains_duplicate_variant(locale_id->language_id.variants))
return {};
for (size_t i = 0; i < variants.size() - 1; ++i) {
if (variants[i] == variants[i + 1])
// if locale contains an extensions* component, that component
Vector<char> unique_keys;
for (auto& extension : locale_id->extensions) {
// does not contain any other_extensions components with duplicate [alphanum-[tTuUxX]] subtags,
// contains at most one unicode_locale_extensions component,
// contains at most one transformed_extensions component, and
char key = extension.visit(
[](Unicode::LocaleExtension const&) { return 'u'; },
[](Unicode::TransformedExtension const&) { return 't'; },
[](Unicode::OtherExtension const& ext) { return static_cast<char>(to_ascii_lowercase(ext.key)); });
if (unique_keys.contains_slow(key))
return {};
unique_keys.append(key);
// if a transformed_extensions component that contains a tlang component is present, then
// the tlang component contains no duplicate unicode_variant_subtag subtags.
if (auto* transformed = extension.get_pointer<Unicode::TransformedExtension>()) {
auto& language = transformed->language;
if (language.has_value() && contains_duplicate_variant(language->variants))
return {};
}
}
// FIXME: Handle extensions.
// if locale contains an extensions* component, that component
// does not contain any other_extensions components with duplicate [alphanum-[tTuUxX]] subtags,
// contains at most one unicode_locale_extensions component,
// contains at most one transformed_extensions component, and
// if a transformed_extensions component that contains a tlang component is present, then
// the tlang component contains no duplicate unicode_variant_subtag subtags.
return locale_id;
}

23
Userland/Libraries/LibJS/Tests/builtins/Intl/Intl.getCanonicalLocales.js
View file

@ -45,6 +45,29 @@ describe("errors", () => {
Intl.getCanonicalLocales([true]);
}).toThrowWithMessage(TypeError, "true is neither an object nor a string");
});
test("duplicate extension components", () => {
expect(() => {
Intl.getCanonicalLocales("en-u-aa-U-aa");
}).toThrowWithMessage(RangeError, "en-u-aa-U-aa is not a structurally valid language tag");
expect(() => {
Intl.getCanonicalLocales("en-t-aa-T-aa");
}).toThrowWithMessage(RangeError, "en-t-aa-T-aa is not a structurally valid language tag");
expect(() => {
Intl.getCanonicalLocales("en-z-aa-Z-aa");
}).toThrowWithMessage(RangeError, "en-z-aa-Z-aa is not a structurally valid language tag");
});
test("duplicate transformed extension variant subtags", () => {
expect(() => {
Intl.getCanonicalLocales("en-t-en-POSIX-POSIX");
}).toThrowWithMessage(
RangeError,
"en-t-en-POSIX-POSIX is not a structurally valid language tag"
);
});
});
describe("normal behavior", () => {