Parser.h 8.8 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229
  1. /*
  2. * Copyright (c) 2020, Stephan Unverwerth <s.unverwerth@serenityos.org>
  3. *
  4. * SPDX-License-Identifier: BSD-2-Clause
  5. */
  6. #pragma once
  7. #include <AK/HashTable.h>
  8. #include <AK/NonnullRefPtr.h>
  9. #include <AK/StringBuilder.h>
  10. #include <LibJS/AST.h>
  11. #include <LibJS/Lexer.h>
  12. #include <LibJS/SourceRange.h>
  13. #include <stdio.h>
  14. namespace JS {
  15. enum class Associativity {
  16. Left,
  17. Right
  18. };
  19. struct FunctionNodeParseOptions {
  20. enum {
  21. CheckForFunctionAndName = 1 << 0,
  22. AllowSuperPropertyLookup = 1 << 1,
  23. AllowSuperConstructorCall = 1 << 2,
  24. IsGetterFunction = 1 << 3,
  25. IsSetterFunction = 1 << 4,
  26. IsArrowFunction = 1 << 5,
  27. };
  28. };
  29. class Parser {
  30. public:
  31. explicit Parser(Lexer lexer);
  32. NonnullRefPtr<Program> parse_program();
  33. template<typename FunctionNodeType>
  34. NonnullRefPtr<FunctionNodeType> parse_function_node(u8 parse_options = FunctionNodeParseOptions::CheckForFunctionAndName);
  35. Vector<FunctionNode::Parameter> parse_function_parameters(int& function_length, u8 parse_options = 0);
  36. NonnullRefPtr<Declaration> parse_declaration();
  37. NonnullRefPtr<Statement> parse_statement();
  38. NonnullRefPtr<BlockStatement> parse_block_statement();
  39. NonnullRefPtr<BlockStatement> parse_block_statement(bool& is_strict);
  40. NonnullRefPtr<ReturnStatement> parse_return_statement();
  41. NonnullRefPtr<VariableDeclaration> parse_variable_declaration(bool for_loop_variable_declaration = false);
  42. NonnullRefPtr<Statement> parse_for_statement();
  43. NonnullRefPtr<Statement> parse_for_in_of_statement(NonnullRefPtr<ASTNode> lhs);
  44. NonnullRefPtr<IfStatement> parse_if_statement();
  45. NonnullRefPtr<ThrowStatement> parse_throw_statement();
  46. NonnullRefPtr<TryStatement> parse_try_statement();
  47. NonnullRefPtr<CatchClause> parse_catch_clause();
  48. NonnullRefPtr<SwitchStatement> parse_switch_statement();
  49. NonnullRefPtr<SwitchCase> parse_switch_case();
  50. NonnullRefPtr<BreakStatement> parse_break_statement();
  51. NonnullRefPtr<ContinueStatement> parse_continue_statement();
  52. NonnullRefPtr<DoWhileStatement> parse_do_while_statement();
  53. NonnullRefPtr<WhileStatement> parse_while_statement();
  54. NonnullRefPtr<WithStatement> parse_with_statement();
  55. NonnullRefPtr<DebuggerStatement> parse_debugger_statement();
  56. NonnullRefPtr<ConditionalExpression> parse_conditional_expression(NonnullRefPtr<Expression> test);
  57. NonnullRefPtr<Expression> parse_expression(int min_precedence, Associativity associate = Associativity::Right, const Vector<TokenType>& forbidden = {});
  58. NonnullRefPtr<Expression> parse_primary_expression();
  59. NonnullRefPtr<Expression> parse_unary_prefixed_expression();
  60. NonnullRefPtr<RegExpLiteral> parse_regexp_literal();
  61. NonnullRefPtr<ObjectExpression> parse_object_expression();
  62. NonnullRefPtr<ArrayExpression> parse_array_expression();
  63. NonnullRefPtr<StringLiteral> parse_string_literal(const Token& token, bool in_template_literal = false);
  64. NonnullRefPtr<TemplateLiteral> parse_template_literal(bool is_tagged);
  65. NonnullRefPtr<Expression> parse_secondary_expression(NonnullRefPtr<Expression>, int min_precedence, Associativity associate = Associativity::Right);
  66. NonnullRefPtr<CallExpression> parse_call_expression(NonnullRefPtr<Expression>);
  67. NonnullRefPtr<NewExpression> parse_new_expression();
  68. NonnullRefPtr<ClassDeclaration> parse_class_declaration();
  69. NonnullRefPtr<ClassExpression> parse_class_expression(bool expect_class_name);
  70. NonnullRefPtr<Expression> parse_property_key();
  71. NonnullRefPtr<AssignmentExpression> parse_assignment_expression(AssignmentOp, NonnullRefPtr<Expression> lhs, int min_precedence, Associativity);
  72. RefPtr<FunctionExpression> try_parse_arrow_function_expression(bool expect_parens);
  73. RefPtr<Statement> try_parse_labelled_statement();
  74. RefPtr<MetaProperty> try_parse_new_target_expression();
  75. struct Error {
  76. String message;
  77. Optional<Position> position;
  78. String to_string() const
  79. {
  80. if (!position.has_value())
  81. return message;
  82. return String::formatted("{} (line: {}, column: {})", message, position.value().line, position.value().column);
  83. }
  84. String source_location_hint(const StringView& source, const char spacer = ' ', const char indicator = '^') const
  85. {
  86. if (!position.has_value())
  87. return {};
  88. // We need to modify the source to match what the lexer considers one line - normalizing
  89. // line terminators to \n is easier than splitting using all different LT characters.
  90. String source_string { source };
  91. source_string.replace("\r\n", "\n");
  92. source_string.replace("\r", "\n");
  93. source_string.replace(LINE_SEPARATOR, "\n");
  94. source_string.replace(PARAGRAPH_SEPARATOR, "\n");
  95. StringBuilder builder;
  96. builder.append(source_string.split_view('\n', true)[position.value().line - 1]);
  97. builder.append('\n');
  98. for (size_t i = 0; i < position.value().column - 1; ++i)
  99. builder.append(spacer);
  100. builder.append(indicator);
  101. return builder.build();
  102. }
  103. };
  104. bool has_errors() const { return m_parser_state.m_errors.size(); }
  105. const Vector<Error>& errors() const { return m_parser_state.m_errors; }
  106. void print_errors() const
  107. {
  108. for (auto& error : m_parser_state.m_errors) {
  109. auto hint = error.source_location_hint(m_parser_state.m_lexer.source());
  110. if (!hint.is_empty())
  111. warnln("{}", hint);
  112. warnln("SyntaxError: {}", error.to_string());
  113. }
  114. }
  115. struct TokenMemoization {
  116. bool try_parse_arrow_function_expression_failed;
  117. };
  118. private:
  119. friend class ScopePusher;
  120. Associativity operator_associativity(TokenType) const;
  121. bool match_expression() const;
  122. bool match_unary_prefixed_expression() const;
  123. bool match_secondary_expression(const Vector<TokenType>& forbidden = {}) const;
  124. bool match_statement() const;
  125. bool match_declaration() const;
  126. bool match_variable_declaration() const;
  127. bool match_identifier_name() const;
  128. bool match_property_key() const;
  129. bool match(TokenType type) const;
  130. bool done() const;
  131. void expected(const char* what);
  132. void syntax_error(const String& message, Optional<Position> = {});
  133. Token consume();
  134. Token consume(TokenType type);
  135. Token consume_and_validate_numeric_literal();
  136. void consume_or_insert_semicolon();
  137. void save_state();
  138. void load_state();
  139. void discard_saved_state();
  140. Position position() const;
  141. bool try_parse_arrow_function_expression_failed_at_position(const Position&) const;
  142. void set_try_parse_arrow_function_expression_failed_at_position(const Position&, bool);
  143. struct RulePosition {
  144. AK_MAKE_NONCOPYABLE(RulePosition);
  145. AK_MAKE_NONMOVABLE(RulePosition);
  146. public:
  147. RulePosition(Parser& parser, Position position)
  148. : m_parser(parser)
  149. , m_position(position)
  150. {
  151. m_parser.m_rule_starts.append(position);
  152. }
  153. ~RulePosition()
  154. {
  155. auto last = m_parser.m_rule_starts.take_last();
  156. VERIFY(last.line == m_position.line);
  157. VERIFY(last.column == m_position.column);
  158. }
  159. const Position& position() const { return m_position; }
  160. private:
  161. Parser& m_parser;
  162. Position m_position;
  163. };
  164. [[nodiscard]] RulePosition push_start() { return { *this, position() }; }
  165. struct ParserState {
  166. Lexer m_lexer;
  167. Token m_current_token;
  168. Vector<Error> m_errors;
  169. Vector<NonnullRefPtrVector<VariableDeclaration>> m_var_scopes;
  170. Vector<NonnullRefPtrVector<VariableDeclaration>> m_let_scopes;
  171. Vector<NonnullRefPtrVector<FunctionDeclaration>> m_function_scopes;
  172. HashTable<StringView> m_labels_in_scope;
  173. bool m_strict_mode { false };
  174. bool m_allow_super_property_lookup { false };
  175. bool m_allow_super_constructor_call { false };
  176. bool m_in_function_context { false };
  177. bool m_in_arrow_function_context { false };
  178. bool m_in_break_context { false };
  179. bool m_in_continue_context { false };
  180. bool m_string_legacy_octal_escape_sequence_in_scope { false };
  181. explicit ParserState(Lexer);
  182. };
  183. class PositionKeyTraits {
  184. public:
  185. static int hash(const Position& position)
  186. {
  187. return int_hash(position.line) ^ int_hash(position.column);
  188. }
  189. static bool equals(const Position& a, const Position& b)
  190. {
  191. return a.column == b.column && a.line == b.line;
  192. }
  193. };
  194. Vector<Position> m_rule_starts;
  195. ParserState m_parser_state;
  196. FlyString m_filename;
  197. Vector<ParserState> m_saved_state;
  198. HashMap<Position, TokenMemoization, PositionKeyTraits> m_token_memoizations;
  199. };
  200. }