Serenity Operating System
1/*
2 * Copyright (c) 2021, Itamar S. <itamar8910@gmail.com>
3 *
4 * SPDX-License-Identifier: BSD-2-Clause
5 */
6
7#pragma once
8
9#include <AK/Noncopyable.h>
10#include <AK/NonnullRefPtr.h>
11#include <LibCodeComprehension/Types.h>
12#include <LibCpp/AST.h>
13#include <LibCpp/Lexer.h>
14#include <LibCpp/Preprocessor.h>
15
16namespace Cpp {
17
18class Parser final {
19 AK_MAKE_NONCOPYABLE(Parser);
20
21public:
22 explicit Parser(Vector<Token> tokens, DeprecatedString const& filename);
23 ~Parser() = default;
24
25 NonnullRefPtr<TranslationUnit> parse();
26 bool eof() const;
27
28 RefPtr<ASTNode const> node_at(Position) const;
29 Optional<size_t> index_of_node_at(Position) const;
30 Optional<Token> token_at(Position) const;
31 Optional<size_t> index_of_token_at(Position) const;
32 RefPtr<TranslationUnit const> root_node() const { return m_root_node; }
33 DeprecatedString text_of_node(ASTNode const&) const;
34 StringView text_of_token(Cpp::Token const& token) const;
35 void print_tokens() const;
36 Vector<Token> const& tokens() const { return m_tokens; }
37 Vector<DeprecatedString> const& errors() const { return m_errors; }
38
39 Vector<CodeComprehension::TodoEntry> get_todo_entries() const;
40
41 Vector<Token> tokens_in_range(Position start, Position end) const;
42
43private:
44 enum class DeclarationType {
45 Function,
46 Variable,
47 Enum,
48 Class,
49 Namespace,
50 Constructor,
51 Destructor,
52 UsingNamespace,
53 };
54
55 Optional<DeclarationType> match_declaration_in_translation_unit();
56 Optional<Parser::DeclarationType> match_class_member(StringView class_name);
57
58 bool match_function_declaration();
59 bool match_comment();
60 bool match_preprocessor();
61 bool match_whitespace();
62 bool match_variable_declaration();
63 bool match_expression();
64 bool match_secondary_expression();
65 bool match_enum_declaration();
66 bool match_class_declaration();
67 bool match_literal();
68 bool match_unary_expression();
69 bool match_boolean_literal();
70 bool match_keyword(DeprecatedString const&);
71 bool match_block_statement();
72 bool match_namespace_declaration();
73 bool match_template_arguments();
74 bool match_name();
75 bool match_cpp_cast_expression();
76 bool match_c_style_cast_expression();
77 bool match_sizeof_expression();
78 bool match_braced_init_list();
79 bool match_type();
80 bool match_named_type();
81 bool match_access_specifier();
82 bool match_constructor(StringView class_name);
83 bool match_destructor(StringView class_name);
84 bool match_using_namespace_declaration();
85
86 Optional<Vector<NonnullRefPtr<Parameter const>>> parse_parameter_list(ASTNode const& parent);
87 Optional<Token> consume_whitespace();
88 void consume_preprocessor();
89
90 NonnullRefPtr<Declaration const> parse_declaration(ASTNode const& parent, DeclarationType);
91 NonnullRefPtr<FunctionDeclaration const> parse_function_declaration(ASTNode const& parent);
92 NonnullRefPtr<FunctionDefinition const> parse_function_definition(ASTNode const& parent);
93 NonnullRefPtr<Statement const> parse_statement(ASTNode const& parent);
94 NonnullRefPtr<VariableDeclaration const> parse_variable_declaration(ASTNode const& parent, bool expect_semicolon = true);
95 NonnullRefPtr<Expression const> parse_expression(ASTNode const& parent);
96 NonnullRefPtr<Expression const> parse_primary_expression(ASTNode const& parent);
97 NonnullRefPtr<Expression const> parse_secondary_expression(ASTNode const& parent, NonnullRefPtr<Expression const> lhs);
98 NonnullRefPtr<StringLiteral const> parse_string_literal(ASTNode const& parent);
99 NonnullRefPtr<ReturnStatement const> parse_return_statement(ASTNode const& parent);
100 NonnullRefPtr<EnumDeclaration const> parse_enum_declaration(ASTNode const& parent);
101 NonnullRefPtr<StructOrClassDeclaration const> parse_class_declaration(ASTNode const& parent);
102 NonnullRefPtr<Expression const> parse_literal(ASTNode const& parent);
103 NonnullRefPtr<UnaryExpression const> parse_unary_expression(ASTNode const& parent);
104 NonnullRefPtr<BooleanLiteral const> parse_boolean_literal(ASTNode const& parent);
105 NonnullRefPtr<Type const> parse_type(ASTNode const& parent);
106 NonnullRefPtr<BinaryExpression const> parse_binary_expression(ASTNode const& parent, NonnullRefPtr<Expression const> lhs, BinaryOp);
107 NonnullRefPtr<AssignmentExpression const> parse_assignment_expression(ASTNode const& parent, NonnullRefPtr<Expression const> lhs, AssignmentOp);
108 NonnullRefPtr<ForStatement const> parse_for_statement(ASTNode const& parent);
109 NonnullRefPtr<BlockStatement const> parse_block_statement(ASTNode const& parent);
110 NonnullRefPtr<Comment const> parse_comment(ASTNode const& parent);
111 NonnullRefPtr<IfStatement const> parse_if_statement(ASTNode const& parent);
112 NonnullRefPtr<NamespaceDeclaration const> parse_namespace_declaration(ASTNode const& parent, bool is_nested_namespace = false);
113 Vector<NonnullRefPtr<Declaration const>> parse_declarations_in_translation_unit(ASTNode const& parent);
114 RefPtr<Declaration const> parse_single_declaration_in_translation_unit(ASTNode const& parent);
115 Vector<NonnullRefPtr<Type const>> parse_template_arguments(ASTNode const& parent);
116 NonnullRefPtr<Name const> parse_name(ASTNode const& parent);
117 NonnullRefPtr<CppCastExpression const> parse_cpp_cast_expression(ASTNode const& parent);
118 NonnullRefPtr<SizeofExpression const> parse_sizeof_expression(ASTNode const& parent);
119 NonnullRefPtr<BracedInitList const> parse_braced_init_list(ASTNode const& parent);
120 NonnullRefPtr<CStyleCastExpression const> parse_c_style_cast_expression(ASTNode const& parent);
121 Vector<NonnullRefPtr<Declaration const>> parse_class_members(StructOrClassDeclaration& parent);
122 NonnullRefPtr<Constructor const> parse_constructor(ASTNode const& parent);
123 NonnullRefPtr<Destructor const> parse_destructor(ASTNode const& parent);
124 NonnullRefPtr<UsingNamespaceDeclaration const> parse_using_namespace_declaration(ASTNode const& parent);
125
126 bool match(Token::Type);
127 Token consume(Token::Type);
128 Token consume();
129 Token consume_keyword(DeprecatedString const&);
130 Token peek(size_t offset = 0) const;
131 Optional<Token> peek(Token::Type) const;
132 Position position() const;
133 Position previous_token_end() const;
134 DeprecatedString text_in_range(Position start, Position end) const;
135
136 void save_state();
137 void load_state();
138
139 struct State {
140 size_t token_index { 0 };
141 Vector<NonnullRefPtr<ASTNode>> state_nodes;
142 };
143
144 void error(StringView message = {});
145
146 template<class T, class... Args>
147 NonnullRefPtr<T>
148 create_ast_node(ASTNode const& parent, Position const& start, Optional<Position> end, Args&&... args)
149 {
150 auto node = adopt_ref(*new T(&parent, start, end, m_filename, forward<Args>(args)...));
151
152 if (m_saved_states.is_empty()) {
153 m_nodes.append(node);
154 } else {
155 m_state.state_nodes.append(node);
156 }
157
158 return node;
159 }
160
161 NonnullRefPtr<TranslationUnit>
162 create_root_ast_node(Position const& start, Position end)
163 {
164 auto node = adopt_ref(*new TranslationUnit(nullptr, start, end, m_filename));
165 m_nodes.append(node);
166 m_root_node = node;
167 return node;
168 }
169
170 DummyAstNode& get_dummy_node()
171 {
172 static NonnullRefPtr<DummyAstNode> dummy = adopt_ref(*new DummyAstNode(nullptr, {}, {}, {}));
173 return dummy;
174 }
175
176 bool match_attribute_specification();
177 void consume_attribute_specification();
178 void consume_access_specifier();
179 bool match_ellipsis();
180 Vector<StringView> parse_type_qualifiers();
181 Vector<StringView> parse_function_qualifiers();
182
183 enum class CtorOrDtor {
184 Ctor,
185 Dtor,
186 };
187 void parse_constructor_or_destructor_impl(FunctionDeclaration&, CtorOrDtor);
188
189 DeprecatedString m_filename;
190 Vector<Token> m_tokens;
191 State m_state;
192 Vector<State> m_saved_states;
193 RefPtr<TranslationUnit> m_root_node;
194 Vector<DeprecatedString> m_errors;
195 Vector<NonnullRefPtr<ASTNode>> m_nodes;
196};
197
198}