13af6ab5fSopenharmony_ci/** 23af6ab5fSopenharmony_ci * Copyright (c) 2021-2024 Huawei Device Co., Ltd. 33af6ab5fSopenharmony_ci * Licensed under the Apache License, Version 2.0 (the "License"); 43af6ab5fSopenharmony_ci * you may not use this file except in compliance with the License. 53af6ab5fSopenharmony_ci * You may obtain a copy of the License at 63af6ab5fSopenharmony_ci * 73af6ab5fSopenharmony_ci * http://www.apache.org/licenses/LICENSE-2.0 83af6ab5fSopenharmony_ci * 93af6ab5fSopenharmony_ci * Unless required by applicable law or agreed to in writing, software 103af6ab5fSopenharmony_ci * distributed under the License is distributed on an "AS IS" BASIS, 113af6ab5fSopenharmony_ci * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 123af6ab5fSopenharmony_ci * See the License for the specific language governing permissions and 133af6ab5fSopenharmony_ci * limitations under the License. 143af6ab5fSopenharmony_ci */ 153af6ab5fSopenharmony_ci 163af6ab5fSopenharmony_ci#ifndef ES2PANDA_PARSER_CORE_REGEXP_H 173af6ab5fSopenharmony_ci#define ES2PANDA_PARSER_CORE_REGEXP_H 183af6ab5fSopenharmony_ci 193af6ab5fSopenharmony_ci#include "util/enumbitops.h" 203af6ab5fSopenharmony_ci#include "util/ustring.h" 213af6ab5fSopenharmony_ci 223af6ab5fSopenharmony_ci#include "../../parser/parserImpl.h" 233af6ab5fSopenharmony_ci 243af6ab5fSopenharmony_ci#include <unordered_set> 253af6ab5fSopenharmony_ci 263af6ab5fSopenharmony_cinamespace ark::es2panda::lexer { 273af6ab5fSopenharmony_ci 283af6ab5fSopenharmony_ciusing ENUMBITOPS_OPERATORS; 293af6ab5fSopenharmony_ci 303af6ab5fSopenharmony_cienum class RegExpFlags : uint32_t { 313af6ab5fSopenharmony_ci EMPTY = 0U, 323af6ab5fSopenharmony_ci GLOBAL = 1U << 0U, 333af6ab5fSopenharmony_ci IGNORE_CASE = 1U << 1U, 343af6ab5fSopenharmony_ci MULTILINE = 1U << 2U, 353af6ab5fSopenharmony_ci DOTALL = 1U << 3U, 363af6ab5fSopenharmony_ci UNICODE = 1U << 4U, 373af6ab5fSopenharmony_ci STICKY = 1U << 5U, 383af6ab5fSopenharmony_ci}; 393af6ab5fSopenharmony_ci 403af6ab5fSopenharmony_cistruct RegExp { 413af6ab5fSopenharmony_ci RegExp(util::StringView p, util::StringView f, RegExpFlags reFlags); 423af6ab5fSopenharmony_ci 433af6ab5fSopenharmony_ci // NOLINTBEGIN(misc-non-private-member-variables-in-classes) 443af6ab5fSopenharmony_ci util::StringView patternStr; 453af6ab5fSopenharmony_ci util::StringView flagsStr; 463af6ab5fSopenharmony_ci RegExpFlags flags; 473af6ab5fSopenharmony_ci // NOLINTEND(misc-non-private-member-variables-in-classes) 483af6ab5fSopenharmony_ci}; 493af6ab5fSopenharmony_ci 503af6ab5fSopenharmony_ciclass RegExpParser { 513af6ab5fSopenharmony_cipublic: 523af6ab5fSopenharmony_ci explicit RegExpParser(const RegExp &re, ArenaAllocator *allocator, const parser::ParserImpl &parser); 533af6ab5fSopenharmony_ci void ParsePattern(); 543af6ab5fSopenharmony_ci 553af6ab5fSopenharmony_ciprivate: 563af6ab5fSopenharmony_ci void ParseDisjunction(); 573af6ab5fSopenharmony_ci void ParseAlternatives(); 583af6ab5fSopenharmony_ci void ParseAlternative(); 593af6ab5fSopenharmony_ci bool ParseAlternativeCharLeftParen(); 603af6ab5fSopenharmony_ci 613af6ab5fSopenharmony_ci void ParseNonCapturingGroup(); 623af6ab5fSopenharmony_ci void ParseNamedCapturingGroup(); 633af6ab5fSopenharmony_ci void ParseCapturingGroup(); 643af6ab5fSopenharmony_ci 653af6ab5fSopenharmony_ci void ParseAssertion(); 663af6ab5fSopenharmony_ci char32_t ParseClassAtom(); 673af6ab5fSopenharmony_ci void ParseCharacterClass(); 683af6ab5fSopenharmony_ci void ParseAtomEscape(); 693af6ab5fSopenharmony_ci void ParseAtomEscapeSwitch(char32_t cp); 703af6ab5fSopenharmony_ci 713af6ab5fSopenharmony_ci uint32_t ParseControlEscape(); 723af6ab5fSopenharmony_ci uint32_t ParseDecimalEscape(); 733af6ab5fSopenharmony_ci uint32_t ParseLegacyOctalEscape(); 743af6ab5fSopenharmony_ci uint32_t ParseHexEscape(); 753af6ab5fSopenharmony_ci uint32_t ParseUnicodeDigits(); 763af6ab5fSopenharmony_ci uint32_t ParseUnicodeEscape(); 773af6ab5fSopenharmony_ci 783af6ab5fSopenharmony_ci void ParseUnicodePropertyEscape(); 793af6ab5fSopenharmony_ci void ValidateNamedBackreference(bool isUnicode); 803af6ab5fSopenharmony_ci void ValidateGroupNameElement(char32_t cp); 813af6ab5fSopenharmony_ci void ParseNamedBackreference(); 823af6ab5fSopenharmony_ci 833af6ab5fSopenharmony_ci void ParseQuantifier(); 843af6ab5fSopenharmony_ci bool ParseBracedQuantifier(); 853af6ab5fSopenharmony_ci 863af6ab5fSopenharmony_ci bool IsSyntaxCharacter(char32_t cp) const; 873af6ab5fSopenharmony_ci bool ParsePatternCharacter(); 883af6ab5fSopenharmony_ci 893af6ab5fSopenharmony_ci util::StringView ParseIdent(); 903af6ab5fSopenharmony_ci 913af6ab5fSopenharmony_ci bool Unicode() const; 923af6ab5fSopenharmony_ci 933af6ab5fSopenharmony_ci char32_t Peek() const; 943af6ab5fSopenharmony_ci char32_t Next(); 953af6ab5fSopenharmony_ci void Advance(); 963af6ab5fSopenharmony_ci bool Eos() const; 973af6ab5fSopenharmony_ci 983af6ab5fSopenharmony_ci RegExp re_; 993af6ab5fSopenharmony_ci ArenaAllocator *allocator_ {}; 1003af6ab5fSopenharmony_ci util::StringView::Iterator iter_; 1013af6ab5fSopenharmony_ci uint32_t capturingGroupCount_ {}; 1023af6ab5fSopenharmony_ci std::unordered_set<util::StringView> groupNames_; 1033af6ab5fSopenharmony_ci std::unordered_set<util::StringView> backReferences_; 1043af6ab5fSopenharmony_ci const es2panda::parser::ParserImpl &parser_; 1053af6ab5fSopenharmony_ci}; 1063af6ab5fSopenharmony_ci} // namespace ark::es2panda::lexer 1073af6ab5fSopenharmony_ci 1083af6ab5fSopenharmony_citemplate <> 1093af6ab5fSopenharmony_cistruct enumbitops::IsAllowedType<ark::es2panda::lexer::RegExpFlags> : std::true_type { 1103af6ab5fSopenharmony_ci}; 1113af6ab5fSopenharmony_ci 1123af6ab5fSopenharmony_ci#endif 113