jcs.org
Serenity Operating System
1/*
2 * Copyright (c) 2021, Tim Flynn <trflynn89@serenityos.org>
3 * Copyright (c) 2021, Jan de Visser <jan@de-visser.net>
4 *
5 * SPDX-License-Identifier: BSD-2-Clause
6 */
7
8#pragma once
9
10#include "Token.h"
11#include <AK/DeprecatedString.h>
12#include <AK/HashMap.h>
13#include <AK/StringView.h>
14
15namespace SQL::AST {
16
17class Lexer {
18public:
19 explicit Lexer(StringView source);
20
21 Token next();
22
23private:
24 void consume(StringBuilder* = nullptr);
25
26 bool consume_whitespace_and_comments();
27 bool consume_numeric_literal(StringBuilder&);
28 bool consume_string_literal(StringBuilder&);
29 bool consume_quoted_identifier(StringBuilder&);
30 bool consume_blob_literal(StringBuilder&);
31 bool consume_exponent(StringBuilder&);
32 bool consume_hexadecimal_number(StringBuilder&);
33
34 bool match(char a, char b) const;
35 bool is_identifier_start() const;
36 bool is_identifier_middle() const;
37 bool is_numeric_literal_start() const;
38 bool is_string_literal_start() const;
39 bool is_string_literal_end() const;
40 bool is_quoted_identifier_start() const;
41 bool is_quoted_identifier_end() const;
42 bool is_blob_literal_start() const;
43 bool is_line_comment_start() const;
44 bool is_block_comment_start() const;
45 bool is_block_comment_end() const;
46 bool is_line_break() const;
47 bool is_eof() const;
48
49 static HashMap<DeprecatedString, TokenType> s_keywords;
50 static HashMap<char, TokenType> s_one_char_tokens;
51 static HashMap<DeprecatedString, TokenType> s_two_char_tokens;
52
53 StringView m_source;
54 size_t m_line_number { 1 };
55 size_t m_line_column { 0 };
56 char m_current_char { 0 };
57 bool m_eof { false };
58 size_t m_position { 0 };
59};
60
61}