RGBCube/serenity
1
Fork 0
mirror of https://github.com/RGBCube/serenity synced 2025-07-27 03:37:45 +00:00
Code Issues Activity Actions

LibJS: Implement RegExp.prototype [ @@split ] with UTF-16 code units

Browse source
This commit is contained in:
Timothy Flynn 2021-07-22 10:24:40 -04:00 • committed by Linus Groh
parent 66c31a0c07
commit ee7b04f7bb
2 changed files with 19 additions and 14 deletions
Download patch file Download diff file Expand all files Collapse all files

24
Userland/Libraries/LibJS/Runtime/RegExpPrototype.cpp
View file

@ -838,9 +838,11 @@ JS_DEFINE_NATIVE_FUNCTION(RegExpPrototype::symbol_split)
auto* regexp_object = this_object_from(vm, global_object); auto* regexp_object = this_object_from(vm, global_object);
if (!regexp_object) if (!regexp_object)
return {}; return {};
auto string = vm.argument(0).to_string(global_object);
auto string = vm.argument(0).to_utf16_string(global_object);
if (vm.exception()) if (vm.exception())
return {}; return {};
Utf16View string_view { string };
auto* constructor = species_constructor(global_object, *regexp_object, *global_object.regexp_constructor()); auto* constructor = species_constructor(global_object, *regexp_object, *global_object.regexp_constructor());
if (vm.exception()) if (vm.exception())
@ -879,28 +881,28 @@ JS_DEFINE_NATIVE_FUNCTION(RegExpPrototype::symbol_split)
if (limit == 0) if (limit == 0)
return array; return array;
if (string.is_empty()) { if (string_view.is_empty()) {
auto result = regexp_exec(global_object, *splitter, string); auto result = regexp_exec(global_object, *splitter, string_view);
if (!result.is_null()) if (!result.is_null())
return array; return array;
array->create_data_property_or_throw(0, js_string(vm, string)); array->create_data_property_or_throw(0, js_string(vm, string_view));
return array; return array;
} }
size_t last_match_end = 0; // 'p' in the spec. size_t last_match_end = 0; // 'p' in the spec.
size_t next_search_from = 0; // 'q' in the spec. size_t next_search_from = 0; // 'q' in the spec.
while (next_search_from < string.length()) { while (next_search_from < string_view.length_in_code_units()) {
splitter->set(vm.names.lastIndex, Value(next_search_from), Object::ShouldThrowExceptions::Yes); splitter->set(vm.names.lastIndex, Value(next_search_from), Object::ShouldThrowExceptions::Yes);
if (vm.exception()) if (vm.exception())
return {}; return {};
auto result = regexp_exec(global_object, *splitter, string); auto result = regexp_exec(global_object, *splitter, string_view);
if (vm.exception()) if (vm.exception())
return {}; return {};
if (result.is_null()) { if (result.is_null()) {
next_search_from = advance_string_index(string, next_search_from, unicode); next_search_from = advance_string_index(string_view, next_search_from, unicode);
continue; continue;
} }
@ -910,14 +912,14 @@ JS_DEFINE_NATIVE_FUNCTION(RegExpPrototype::symbol_split)
auto last_index = last_index_value.to_length(global_object); // 'e' in the spec. auto last_index = last_index_value.to_length(global_object); // 'e' in the spec.
if (vm.exception()) if (vm.exception())
return {}; return {};
last_index = min(last_index, string.length()); last_index = min(last_index, string_view.length_in_code_units());
if (last_index == last_match_end) { if (last_index == last_match_end) {
next_search_from = advance_string_index(string, next_search_from, unicode); next_search_from = advance_string_index(string_view, next_search_from, unicode);
continue; continue;
} }
auto substring = string.substring(last_match_end, next_search_from - last_match_end); auto substring = string_view.substring_view(last_match_end, next_search_from - last_match_end);
array->create_data_property_or_throw(array_length, js_string(vm, move(substring))); array->create_data_property_or_throw(array_length, js_string(vm, move(substring)));
if (++array_length == limit) if (++array_length == limit)
@ -946,7 +948,7 @@ JS_DEFINE_NATIVE_FUNCTION(RegExpPrototype::symbol_split)
next_search_from = last_index; next_search_from = last_index;
} }
auto substring = string.substring(last_match_end); auto substring = string_view.substring_view(last_match_end);
array->create_data_property_or_throw(array_length, js_string(vm, move(substring))); array->create_data_property_or_throw(array_length, js_string(vm, move(substring)));
return array; return array;

9
Userland/Libraries/LibJS/Tests/builtins/String/String.prototype.split.js
View file

@ -73,7 +73,10 @@ test("UTF-16", () => {
expect(s.split("\ud83d")).toEqual(["", "\ude00"]); expect(s.split("\ud83d")).toEqual(["", "\ude00"]);
expect(s.split("\ude00")).toEqual(["\ud83d", ""]); expect(s.split("\ude00")).toEqual(["\ud83d", ""]);
// FIXME: RegExp.prototype [ @@split ] also needs to support UTF-16. expect(s.split(/\ud83d/)).toEqual(["", "\ude00"]);
// expect(s.split(/\ud83d/)).toEqual(["", "\ude00"]); expect(s.split(/\ude00/)).toEqual(["\ud83d", ""]);
// expect(s.split(/\ude00/)).toEqual(["\ud83d", ""]);
s = "😀😀😀";
expect(s.split(/\ud83d/)).toEqual(["", "\ude00", "\ude00", "\ude00"]);
expect(s.split(/\ude00/)).toEqual(["\ud83d", "\ud83d", "\ud83d", ""]);
}); });