RGBCube/serenity
1
Fork 0
mirror of https://github.com/RGBCube/serenity synced 2025-07-25 17:37:37 +00:00
Code Issues Activity Actions

LibRegex: Fix parsing identity escape sequences

Browse source 
Also fixes the propagation of default options (the previous
implementation reset them to zero before parsing...).
Partially deals with #4189.
This commit is contained in:
AnotherTest 2020-11-28 10:46:30 +03:30 • committed by Andreas Kling
parent e83e7a03c2
commit 801750b95a
1 changed files with 24 additions and 7 deletions
Download patch file Download diff file Expand all files Collapse all files

31
Libraries/LibRegex/RegexParser.cpp
View file

@ -115,7 +115,6 @@ ALWAYS_INLINE void Parser::reset()
m_parser_state.current_token = m_parser_state.lexer.next(); m_parser_state.current_token = m_parser_state.lexer.next();
m_parser_state.error = Error::NoError; m_parser_state.error = Error::NoError;
m_parser_state.error_token = { TokenType::Eof, 0, StringView(nullptr) }; m_parser_state.error_token = { TokenType::Eof, 0, StringView(nullptr) };
m_parser_state.regex_options = {};
} }
Parser::Result Parser::parse(Optional<AllOptions> regex_options) Parser::Result Parser::parse(Optional<AllOptions> regex_options)
@ -621,7 +620,7 @@ bool PosixExtendedParser::parse_root(ByteCode& stack, size_t& match_length_minim
bool ECMA262Parser::parse_internal(ByteCode& stack, size_t& match_length_minimum) bool ECMA262Parser::parse_internal(ByteCode& stack, size_t& match_length_minimum)
{ {
if (m_parser_state.regex_options & AllFlags::Unicode) { if (m_parser_state.regex_options.has_flag_set(AllFlags::Unicode)) {
return parse_pattern(stack, match_length_minimum, true, true); return parse_pattern(stack, match_length_minimum, true, true);
} else { } else {
ByteCode new_stack; ByteCode new_stack;
@ -918,6 +917,13 @@ bool ECMA262Parser::parse_quantifier(ByteCode& stack, size_t& match_length_minim
bool ECMA262Parser::parse_atom(ByteCode& stack, size_t& match_length_minimum, bool unicode, bool named) bool ECMA262Parser::parse_atom(ByteCode& stack, size_t& match_length_minimum, bool unicode, bool named)
{ {
if (match(TokenType::EscapeSequence)) {
// Also part of AtomEscape.
auto token = consume();
match_length_minimum += 1;
stack.insert_bytecode_compare_values({ { CharacterCompareType::Char, (ByteCodeValueType)token.value()[0] } });
return true;
}
if (try_skip("\\")) { if (try_skip("\\")) {
// AtomEscape. // AtomEscape.
return parse_atom_escape(stack, match_length_minimum, unicode, named); return parse_atom_escape(stack, match_length_minimum, unicode, named);
@ -1035,11 +1041,20 @@ bool ECMA262Parser::parse_atom_escape(ByteCode& stack, size_t& match_length_mini
} }
// IdentityEscape // IdentityEscape
if (match(TokenType::EscapeSequence)) { for (auto ch : StringView { "^$\\.*+?()[]{}|" }) {
match_length_minimum += 1; if (try_skip({ &ch, 1 })) {
auto token = consume().value(); match_length_minimum += 1;
stack.insert_bytecode_compare_values({ { CharacterCompareType::Char, (ByteCodeValueType)token[token.length() - 1] } }); stack.insert_bytecode_compare_values({ { CharacterCompareType::Char, (ByteCodeValueType)ch } });
return true; return true;
}
}
if (unicode) {
if (try_skip("/")) {
match_length_minimum += 1;
stack.insert_bytecode_compare_values({ { CharacterCompareType::Char, (ByteCodeValueType)'/' } });
return true;
}
} }
if (named && try_skip("k")) { if (named && try_skip("k")) {
@ -1171,6 +1186,8 @@ bool ECMA262Parser::parse_nonempty_class_ranges(Vector<CompareTypeAndValuePair>&
return { { .code_point = '\v', .is_character_class = false } }; return { { .code_point = '\v', .is_character_class = false } };
if (try_skip("b")) if (try_skip("b"))
return { { .code_point = '\b', .is_character_class = false } }; return { { .code_point = '\b', .is_character_class = false } };
if (try_skip("/"))
return { { .code_point = '/', .is_character_class = false } };
// CharacterEscape > ControlLetter // CharacterEscape > ControlLetter
if (try_skip("c")) { if (try_skip("c")) {