Parser.h 14 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359
  1. /*
  2. * Copyright (c) 2020, Stephan Unverwerth <s.unverwerth@serenityos.org>
  3. * Copyright (c) 2021-2022, David Tuin <davidot@serenityos.org>
  4. * Copyright (c) 2023, Andreas Kling <kling@serenityos.org>
  5. *
  6. * SPDX-License-Identifier: BSD-2-Clause
  7. */
  8. #pragma once
  9. #include <AK/Assertions.h>
  10. #include <AK/HashTable.h>
  11. #include <AK/NonnullRefPtr.h>
  12. #include <AK/StringBuilder.h>
  13. #include <LibJS/AST.h>
  14. #include <LibJS/Lexer.h>
  15. #include <LibJS/ParserError.h>
  16. #include <LibJS/Runtime/FunctionConstructor.h>
  17. #include <LibJS/SourceRange.h>
  18. #include <LibJS/Token.h>
  19. #include <initializer_list>
  20. #include <stdio.h>
  21. namespace JS {
  22. enum class Associativity {
  23. Left,
  24. Right
  25. };
  26. struct FunctionNodeParseOptions {
  27. enum : u16 {
  28. CheckForFunctionAndName = 1 << 0,
  29. AllowSuperPropertyLookup = 1 << 1,
  30. AllowSuperConstructorCall = 1 << 2,
  31. IsGetterFunction = 1 << 3,
  32. IsSetterFunction = 1 << 4,
  33. IsArrowFunction = 1 << 5,
  34. IsGeneratorFunction = 1 << 6,
  35. IsAsyncFunction = 1 << 7,
  36. HasDefaultExportName = 1 << 8,
  37. };
  38. };
  39. class ScopePusher;
  40. class Parser {
  41. public:
  42. struct EvalInitialState {
  43. bool in_eval_function_context { false };
  44. bool allow_super_property_lookup { false };
  45. bool allow_super_constructor_call { false };
  46. bool in_class_field_initializer { false };
  47. };
  48. explicit Parser(Lexer lexer, Program::Type program_type = Program::Type::Script, Optional<EvalInitialState> initial_state_for_eval = {});
  49. NonnullRefPtr<Program> parse_program(bool starts_in_strict_mode = false);
  50. template<typename FunctionNodeType>
  51. NonnullRefPtr<FunctionNodeType> parse_function_node(u16 parse_options = FunctionNodeParseOptions::CheckForFunctionAndName, Optional<Position> const& function_start = {});
  52. Vector<FunctionParameter> parse_formal_parameters(int& function_length, u16 parse_options = 0);
  53. enum class AllowDuplicates {
  54. Yes,
  55. No
  56. };
  57. enum class AllowMemberExpressions {
  58. Yes,
  59. No
  60. };
  61. RefPtr<BindingPattern const> parse_binding_pattern(AllowDuplicates is_var_declaration = AllowDuplicates::No, AllowMemberExpressions allow_member_expressions = AllowMemberExpressions::No);
  62. struct PrimaryExpressionParseResult {
  63. NonnullRefPtr<Expression const> result;
  64. bool should_continue_parsing_as_expression { true };
  65. };
  66. NonnullRefPtr<Declaration const> parse_declaration();
  67. enum class AllowLabelledFunction {
  68. No,
  69. Yes
  70. };
  71. NonnullRefPtr<Statement const> parse_statement(AllowLabelledFunction allow_labelled_function = AllowLabelledFunction::No);
  72. NonnullRefPtr<BlockStatement const> parse_block_statement();
  73. NonnullRefPtr<FunctionBody const> parse_function_body(Vector<FunctionParameter> const& parameters, FunctionKind function_kind, bool& contains_direct_call_to_eval);
  74. NonnullRefPtr<ReturnStatement const> parse_return_statement();
  75. enum class IsForLoopVariableDeclaration {
  76. No,
  77. Yes
  78. };
  79. NonnullRefPtr<VariableDeclaration const> parse_variable_declaration(IsForLoopVariableDeclaration is_for_loop_variable_declaration = IsForLoopVariableDeclaration::No);
  80. RefPtr<Identifier const> parse_lexical_binding();
  81. NonnullRefPtr<UsingDeclaration const> parse_using_declaration(IsForLoopVariableDeclaration is_for_loop_variable_declaration = IsForLoopVariableDeclaration::No);
  82. NonnullRefPtr<Statement const> parse_for_statement();
  83. enum class IsForAwaitLoop {
  84. No,
  85. Yes
  86. };
  87. struct ForbiddenTokens {
  88. ForbiddenTokens(std::initializer_list<TokenType> const& forbidden);
  89. ForbiddenTokens merge(ForbiddenTokens other) const;
  90. bool allows(TokenType token) const;
  91. ForbiddenTokens forbid(std::initializer_list<TokenType> const& forbidden) const;
  92. private:
  93. void forbid_tokens(std::initializer_list<TokenType> const& forbidden);
  94. bool m_forbid_in_token : 1 { false };
  95. bool m_forbid_logical_tokens : 1 { false };
  96. bool m_forbid_coalesce_token : 1 { false };
  97. bool m_forbid_paren_open : 1 { false };
  98. bool m_forbid_question_mark_period : 1 { false };
  99. bool m_forbid_equals : 1 { false };
  100. };
  101. struct ExpressionResult {
  102. template<typename T>
  103. ExpressionResult(NonnullRefPtr<T const> expression, ForbiddenTokens forbidden = {})
  104. : expression(move(expression))
  105. , forbidden(forbidden)
  106. {
  107. }
  108. template<typename T>
  109. ExpressionResult(NonnullRefPtr<T> expression, ForbiddenTokens forbidden = {})
  110. : expression(move(expression))
  111. , forbidden(forbidden)
  112. {
  113. }
  114. NonnullRefPtr<Expression const> expression;
  115. ForbiddenTokens forbidden;
  116. };
  117. NonnullRefPtr<Statement const> parse_for_in_of_statement(NonnullRefPtr<ASTNode const> lhs, IsForAwaitLoop is_await);
  118. NonnullRefPtr<IfStatement const> parse_if_statement();
  119. NonnullRefPtr<ThrowStatement const> parse_throw_statement();
  120. NonnullRefPtr<TryStatement const> parse_try_statement();
  121. NonnullRefPtr<CatchClause const> parse_catch_clause();
  122. NonnullRefPtr<SwitchStatement const> parse_switch_statement();
  123. NonnullRefPtr<SwitchCase const> parse_switch_case();
  124. NonnullRefPtr<BreakStatement const> parse_break_statement();
  125. NonnullRefPtr<ContinueStatement const> parse_continue_statement();
  126. NonnullRefPtr<DoWhileStatement const> parse_do_while_statement();
  127. NonnullRefPtr<WhileStatement const> parse_while_statement();
  128. NonnullRefPtr<WithStatement const> parse_with_statement();
  129. NonnullRefPtr<DebuggerStatement const> parse_debugger_statement();
  130. NonnullRefPtr<ConditionalExpression const> parse_conditional_expression(NonnullRefPtr<Expression const> test, ForbiddenTokens);
  131. NonnullRefPtr<OptionalChain const> parse_optional_chain(NonnullRefPtr<Expression const> base);
  132. NonnullRefPtr<Expression const> parse_expression(int min_precedence, Associativity associate = Associativity::Right, ForbiddenTokens forbidden = {});
  133. PrimaryExpressionParseResult parse_primary_expression();
  134. NonnullRefPtr<Expression const> parse_unary_prefixed_expression();
  135. NonnullRefPtr<RegExpLiteral const> parse_regexp_literal();
  136. NonnullRefPtr<ObjectExpression const> parse_object_expression();
  137. NonnullRefPtr<ArrayExpression const> parse_array_expression();
  138. enum class StringLiteralType {
  139. Normal,
  140. NonTaggedTemplate,
  141. TaggedTemplate
  142. };
  143. NonnullRefPtr<StringLiteral const> parse_string_literal(Token const& token, StringLiteralType string_literal_type = StringLiteralType::Normal, bool* contains_invalid_escape = nullptr);
  144. NonnullRefPtr<TemplateLiteral const> parse_template_literal(bool is_tagged);
  145. ExpressionResult parse_secondary_expression(NonnullRefPtr<Expression const>, int min_precedence, Associativity associate = Associativity::Right, ForbiddenTokens forbidden = {});
  146. NonnullRefPtr<Expression const> parse_call_expression(NonnullRefPtr<Expression const>);
  147. NonnullRefPtr<NewExpression const> parse_new_expression();
  148. NonnullRefPtr<ClassDeclaration const> parse_class_declaration();
  149. NonnullRefPtr<ClassExpression const> parse_class_expression(bool expect_class_name);
  150. NonnullRefPtr<YieldExpression const> parse_yield_expression();
  151. NonnullRefPtr<AwaitExpression const> parse_await_expression();
  152. NonnullRefPtr<Expression const> parse_property_key();
  153. NonnullRefPtr<AssignmentExpression const> parse_assignment_expression(AssignmentOp, NonnullRefPtr<Expression const> lhs, int min_precedence, Associativity, ForbiddenTokens forbidden = {});
  154. NonnullRefPtr<Identifier const> parse_identifier();
  155. NonnullRefPtr<ImportStatement const> parse_import_statement(Program& program);
  156. NonnullRefPtr<ExportStatement const> parse_export_statement(Program& program);
  157. RefPtr<FunctionExpression const> try_parse_arrow_function_expression(bool expect_parens, bool is_async = false);
  158. RefPtr<LabelledStatement const> try_parse_labelled_statement(AllowLabelledFunction allow_function);
  159. RefPtr<MetaProperty const> try_parse_new_target_expression();
  160. RefPtr<MetaProperty const> try_parse_import_meta_expression();
  161. NonnullRefPtr<ImportCall const> parse_import_call();
  162. Vector<CallExpression::Argument> parse_arguments();
  163. bool has_errors() const { return m_state.errors.size(); }
  164. Vector<ParserError> const& errors() const { return m_state.errors; }
  165. void print_errors(bool print_hint = true) const
  166. {
  167. for (auto& error : m_state.errors) {
  168. if (print_hint) {
  169. auto hint = error.source_location_hint(m_state.lexer.source());
  170. if (!hint.is_empty())
  171. warnln("{}", hint);
  172. }
  173. warnln("SyntaxError: {}", error.to_deprecated_string());
  174. }
  175. }
  176. struct TokenMemoization {
  177. bool try_parse_arrow_function_expression_failed;
  178. };
  179. // Needs to mess with m_state, and we're not going to expose a non-const getter for that :^)
  180. friend ThrowCompletionOr<ECMAScriptFunctionObject*> FunctionConstructor::create_dynamic_function(VM&, FunctionObject&, FunctionObject*, FunctionKind, MarkedVector<Value> const&);
  181. static Parser parse_function_body_from_string(DeprecatedString const& body_string, u16 parse_options, Vector<FunctionParameter> const& parameters, FunctionKind kind, bool& contains_direct_call_to_eval);
  182. private:
  183. friend class ScopePusher;
  184. void parse_script(Program& program, bool starts_in_strict_mode);
  185. void parse_module(Program& program);
  186. Associativity operator_associativity(TokenType) const;
  187. bool match_expression() const;
  188. bool match_unary_prefixed_expression() const;
  189. bool match_secondary_expression(ForbiddenTokens forbidden = {}) const;
  190. bool match_statement() const;
  191. bool match_export_or_import() const;
  192. bool match_assert_clause() const;
  193. enum class AllowUsingDeclaration {
  194. No,
  195. Yes
  196. };
  197. bool match_declaration(AllowUsingDeclaration allow_using = AllowUsingDeclaration::No) const;
  198. bool try_match_let_declaration() const;
  199. bool try_match_using_declaration() const;
  200. bool match_variable_declaration() const;
  201. bool match_identifier() const;
  202. bool token_is_identifier(Token const&) const;
  203. bool match_identifier_name() const;
  204. bool match_property_key() const;
  205. bool is_private_identifier_valid() const;
  206. bool match(TokenType type) const;
  207. bool done() const;
  208. void expected(char const* what);
  209. void syntax_error(DeprecatedString const& message, Optional<Position> = {});
  210. Token consume();
  211. Token consume_and_allow_division();
  212. Token consume_identifier();
  213. Token consume_identifier_reference();
  214. Token consume(TokenType type);
  215. Token consume_and_validate_numeric_literal();
  216. void consume_or_insert_semicolon();
  217. void save_state();
  218. void load_state();
  219. void discard_saved_state();
  220. Position position() const;
  221. RefPtr<BindingPattern const> synthesize_binding_pattern(Expression const& expression);
  222. Token next_token(size_t steps = 1) const;
  223. void check_identifier_name_for_assignment_validity(DeprecatedFlyString const&, bool force_strict = false);
  224. bool try_parse_arrow_function_expression_failed_at_position(Position const&) const;
  225. void set_try_parse_arrow_function_expression_failed_at_position(Position const&, bool);
  226. bool match_invalid_escaped_keyword() const;
  227. bool parse_directive(ScopeNode& body);
  228. void parse_statement_list(ScopeNode& output_node, AllowLabelledFunction allow_labelled_functions = AllowLabelledFunction::No);
  229. DeprecatedFlyString consume_string_value();
  230. ModuleRequest parse_module_request();
  231. struct RulePosition {
  232. AK_MAKE_NONCOPYABLE(RulePosition);
  233. AK_MAKE_NONMOVABLE(RulePosition);
  234. public:
  235. RulePosition(Parser& parser, Position position)
  236. : m_parser(parser)
  237. , m_position(position)
  238. {
  239. m_parser.m_rule_starts.append(position);
  240. }
  241. ~RulePosition()
  242. {
  243. auto last = m_parser.m_rule_starts.take_last();
  244. VERIFY(last.line == m_position.line);
  245. VERIFY(last.column == m_position.column);
  246. }
  247. Position const& position() const { return m_position; }
  248. private:
  249. Parser& m_parser;
  250. Position m_position;
  251. };
  252. [[nodiscard]] RulePosition push_start() { return { *this, position() }; }
  253. struct ParserState {
  254. Lexer lexer;
  255. Token current_token;
  256. Vector<ParserError> errors;
  257. ScopePusher* current_scope_pusher { nullptr };
  258. HashMap<StringView, Optional<Position>> labels_in_scope;
  259. HashMap<size_t, Position> invalid_property_range_in_object_expression;
  260. HashTable<StringView>* referenced_private_names { nullptr };
  261. bool strict_mode { false };
  262. bool allow_super_property_lookup { false };
  263. bool allow_super_constructor_call { false };
  264. bool in_function_context { false };
  265. bool initiated_by_eval { false };
  266. bool in_eval_function_context { false }; // This controls if we allow new.target or not. Note that eval("return") is not allowed, so we have to have a separate state variable for eval.
  267. bool in_formal_parameter_context { false };
  268. bool in_catch_parameter_context { false };
  269. bool in_generator_function_context { false };
  270. bool await_expression_is_valid { false };
  271. bool in_arrow_function_context { false };
  272. bool in_break_context { false };
  273. bool in_continue_context { false };
  274. bool string_legacy_octal_escape_sequence_in_scope { false };
  275. bool in_class_field_initializer { false };
  276. bool in_class_static_init_block { false };
  277. bool function_might_need_arguments_object { false };
  278. ParserState(Lexer, Program::Type);
  279. };
  280. class PositionKeyTraits {
  281. public:
  282. static int hash(Position const& position)
  283. {
  284. return int_hash(position.line) ^ int_hash(position.column);
  285. }
  286. static bool equals(Position const& a, Position const& b)
  287. {
  288. return a.column == b.column && a.line == b.line;
  289. }
  290. };
  291. NonnullRefPtr<Identifier const> create_identifier_and_register_in_current_scope(SourceRange range, DeprecatedFlyString string);
  292. NonnullRefPtr<SourceCode const> m_source_code;
  293. Vector<Position> m_rule_starts;
  294. ParserState m_state;
  295. DeprecatedFlyString m_filename;
  296. Vector<ParserState> m_saved_state;
  297. HashMap<Position, TokenMemoization, PositionKeyTraits> m_token_memoizations;
  298. Program::Type m_program_type;
  299. };
  300. }