RGBCube/serenity
1
Fork 0
mirror of https://github.com/RGBCube/serenity synced 2025-07-26 18:57:34 +00:00
Code Issues Activity Actions

LibRegex: Allow a '?' suffix for brace quantifiers

Browse source 
This fixes another compat point in #6042.
This commit is contained in:
AnotherTest 2021-04-10 08:31:53 +04:30 • committed by Andreas Kling
parent 8d7bcc2476
commit e9279d1790
3 changed files with 15 additions and 9 deletions
Download patch file Download diff file Expand all files Collapse all files

8
Userland/Libraries/LibJS/Tests/builtins/RegExp/RegExp.prototype.exec.js
View file

@ -119,3 +119,11 @@ test("named capture group with two '?' qualifiers", () => {
expect(res[1]).toBeUndefined(); expect(res[1]).toBeUndefined();
expect(res.groups.foo).toBeUndefined(); expect(res.groups.foo).toBeUndefined();
}); });
// #6042
test("non-greedy brace quantifier", () => {
let res = /a[a-z]{2,4}?/.exec("abcdefghi");
expect(res.length).toBe(1);
expect(res[0]).toBe("abc");
});

10
Userland/Libraries/LibRegex/RegexByteCode.h
View file

@ -362,26 +362,28 @@ public:
// LABEL _END = alterantive_bytecode.size // LABEL _END = alterantive_bytecode.size
} }
void insert_bytecode_repetition_min_max(ByteCode& bytecode_to_repeat, size_t minimum, Optional<size_t> maximum) void insert_bytecode_repetition_min_max(ByteCode& bytecode_to_repeat, size_t minimum, Optional<size_t> maximum, bool greedy = true)
{ {
ByteCode new_bytecode; ByteCode new_bytecode;
new_bytecode.insert_bytecode_repetition_n(bytecode_to_repeat, minimum); new_bytecode.insert_bytecode_repetition_n(bytecode_to_repeat, minimum);
if (maximum.has_value()) { if (maximum.has_value()) {
auto jump_kind = static_cast<ByteCodeValueType>(greedy ? OpCodeId::ForkStay : OpCodeId::ForkJump);
if (maximum.value() > minimum) { if (maximum.value() > minimum) {
auto diff = maximum.value() - minimum; auto diff = maximum.value() - minimum;
new_bytecode.empend(static_cast<ByteCodeValueType>(OpCodeId::ForkStay)); new_bytecode.empend(jump_kind);
new_bytecode.empend(diff * (bytecode_to_repeat.size() + 2)); // Jump to the _END label new_bytecode.empend(diff * (bytecode_to_repeat.size() + 2)); // Jump to the _END label
for (size_t i = 0; i < diff; ++i) { for (size_t i = 0; i < diff; ++i) {
new_bytecode.append(bytecode_to_repeat); new_bytecode.append(bytecode_to_repeat);
new_bytecode.empend(static_cast<ByteCodeValueType>(OpCodeId::ForkStay)); new_bytecode.empend(jump_kind);
new_bytecode.empend((diff - i - 1) * (bytecode_to_repeat.size() + 2)); // Jump to the _END label new_bytecode.empend((diff - i - 1) * (bytecode_to_repeat.size() + 2)); // Jump to the _END label
} }
} }
} else { } else {
// no maximum value set, repeat finding if possible // no maximum value set, repeat finding if possible
new_bytecode.empend(static_cast<ByteCodeValueType>(OpCodeId::ForkJump)); auto jump_kind = static_cast<ByteCodeValueType>(greedy ? OpCodeId::ForkJump : OpCodeId::ForkStay);
new_bytecode.empend(jump_kind);
new_bytecode.empend(-bytecode_to_repeat.size() - 2); // Jump to the last iteration new_bytecode.empend(-bytecode_to_repeat.size() - 2); // Jump to the last iteration
} }

6
Userland/Libraries/LibRegex/RegexParser.cpp
View file

@ -968,10 +968,6 @@ bool ECMA262Parser::parse_quantifier(ByteCode& stack, size_t& match_length_minim
} }
if (match(TokenType::Questionmark)) { if (match(TokenType::Questionmark)) {
if (repetition_mark == Repetition::Explicit) {
set_error(Error::InvalidRepetitionMarker);
return false;
}
consume(); consume();
ungreedy = true; ungreedy = true;
} }
@ -990,7 +986,7 @@ bool ECMA262Parser::parse_quantifier(ByteCode& stack, size_t& match_length_minim
match_length_minimum = 0; match_length_minimum = 0;
break; break;
case Repetition::Explicit: case Repetition::Explicit:
new_bytecode.insert_bytecode_repetition_min_max(stack, repeat_min.value(), repeat_max); new_bytecode.insert_bytecode_repetition_min_max(stack, repeat_min.value(), repeat_max, !ungreedy);
match_length_minimum *= repeat_min.value(); match_length_minimum *= repeat_min.value();
break; break;
case Repetition::None: case Repetition::None: