|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
|
|
4
|
|
|
namespace App; |
|
5
|
|
|
|
|
6
|
|
|
|
|
7
|
|
|
use App\Exception\LexerException; |
|
8
|
|
|
use App\Exception\UnknownIdentifier; |
|
9
|
|
|
|
|
10
|
|
|
class Lexer |
|
11
|
|
|
{ |
|
12
|
|
|
/** |
|
13
|
|
|
* @var string |
|
14
|
|
|
*/ |
|
15
|
|
|
private $input; |
|
16
|
|
|
/** |
|
17
|
|
|
* @var string|null |
|
18
|
|
|
*/ |
|
19
|
|
|
private $currentChar; |
|
20
|
|
|
/** |
|
21
|
|
|
* @var int |
|
22
|
|
|
*/ |
|
23
|
|
|
private $pos = 0; |
|
24
|
|
|
|
|
25
|
|
|
/** |
|
26
|
|
|
* Lexer constructor. |
|
27
|
|
|
* @param string $input |
|
28
|
|
|
*/ |
|
29
|
|
|
public function __construct( |
|
30
|
|
|
string $input |
|
31
|
|
|
) |
|
32
|
|
|
{ |
|
33
|
|
|
$this->input = $input; |
|
34
|
|
|
$this->currentChar = $this->input[$this->pos]; |
|
35
|
|
|
} |
|
36
|
|
|
|
|
37
|
|
|
private function isAlpha(string $char): bool |
|
38
|
|
|
{ |
|
39
|
|
|
return (bool)preg_match('/[a-zA-Z_]/', $char); |
|
40
|
|
|
} |
|
41
|
|
|
|
|
42
|
|
|
/** |
|
43
|
|
|
* @return Token |
|
44
|
|
|
* @throws UnknownIdentifier |
|
45
|
|
|
*/ |
|
46
|
|
|
private function identifier(): Token |
|
47
|
|
|
{ |
|
48
|
|
|
$identifier = ''; |
|
49
|
|
|
while ($this->currentChar !== null && $this->isAlpha($this->currentChar)) { |
|
50
|
|
|
$identifier .= $this->currentChar; |
|
51
|
|
|
$this->advance(); |
|
52
|
|
|
} |
|
53
|
|
|
|
|
54
|
|
|
$startPos = $this->pos; |
|
55
|
|
|
while ($this->input[$startPos] === ' ') { |
|
56
|
|
|
$startPos++; |
|
57
|
|
|
} |
|
58
|
|
|
$nextChar = $this->input[$startPos] ?? null; |
|
59
|
|
|
|
|
60
|
|
|
if ($nextChar === '(') { |
|
61
|
|
|
return new Token(Token::FUNCTION_CALL, $identifier); |
|
62
|
|
|
} |
|
63
|
|
|
|
|
64
|
|
|
throw new UnknownIdentifier('Unknown identifier ' . $identifier); |
|
65
|
|
|
} |
|
66
|
|
|
|
|
67
|
|
|
/** |
|
68
|
|
|
* @return Token |
|
69
|
|
|
* @throws LexerException |
|
70
|
|
|
* @throws UnknownIdentifier |
|
71
|
|
|
*/ |
|
72
|
|
|
public function getNextToken(): Token |
|
73
|
|
|
{ |
|
74
|
|
|
while ($this->currentChar !== null) { |
|
75
|
|
|
$this->skipWhitespace(); |
|
76
|
|
|
if ($this->isAlpha($this->currentChar)) { |
|
77
|
|
|
return $this->identifier(); |
|
78
|
|
|
} |
|
79
|
|
|
if ($this->currentChar === ',') { |
|
80
|
|
|
$this->advance(); |
|
81
|
|
|
return new Token(Token::COMMA, ','); |
|
82
|
|
|
} |
|
83
|
|
|
if ($this->currentChar === '+') { |
|
84
|
|
|
$this->advance(); |
|
85
|
|
|
return new Token(Token::PLUS, '+'); |
|
86
|
|
|
} |
|
87
|
|
|
if ($this->currentChar === '-') { |
|
88
|
|
|
$this->advance(); |
|
89
|
|
|
return new Token(Token::MINUS, '-'); |
|
90
|
|
|
} |
|
91
|
|
|
if ($this->currentChar === '^') { |
|
92
|
|
|
$this->advance(); |
|
93
|
|
|
return new Token(Token::POWER, '^'); |
|
94
|
|
|
} |
|
95
|
|
|
if ($this->currentChar === '*') { |
|
96
|
|
|
return $this->analyzeAsterisk(); |
|
97
|
|
|
} |
|
98
|
|
|
if ($this->currentChar === '/') { |
|
99
|
|
|
$this->advance(); |
|
100
|
|
|
return new Token(Token::REALDIV, '/'); |
|
101
|
|
|
} |
|
102
|
|
|
if ($this->currentChar === '(') { |
|
103
|
|
|
$this->advance(); |
|
104
|
|
|
return new Token(Token::LPAREN, '('); |
|
105
|
|
|
} |
|
106
|
|
|
if ($this->currentChar === ')') { |
|
107
|
|
|
$this->advance(); |
|
108
|
|
|
return new Token(Token::RPAREN, ')'); |
|
109
|
|
|
} |
|
110
|
|
|
|
|
111
|
|
|
if (is_numeric($this->currentChar)) { |
|
112
|
|
|
return $this->parseNumber(); |
|
113
|
|
|
} |
|
114
|
|
|
} |
|
115
|
|
|
|
|
116
|
|
|
return new Token(Token::EOF, Token::EOF); |
|
117
|
|
|
} |
|
118
|
|
|
|
|
119
|
|
|
private function skipWhitespace(): void |
|
120
|
|
|
{ |
|
121
|
|
|
if ($this->currentChar === ' ') { |
|
122
|
|
|
$this->advance(); |
|
123
|
|
|
} |
|
124
|
|
|
} |
|
125
|
|
|
|
|
126
|
|
|
private function advance(): void |
|
127
|
|
|
{ |
|
128
|
|
|
$this->pos++; |
|
129
|
|
|
$this->currentChar = $this->input[$this->pos] ?? null; |
|
130
|
|
|
} |
|
131
|
|
|
|
|
132
|
|
|
private function peek(): ?string |
|
133
|
|
|
{ |
|
134
|
|
|
$pos = $this->pos + 1; |
|
135
|
|
|
return $this->input[$pos] ?? null; |
|
136
|
|
|
} |
|
137
|
|
|
|
|
138
|
|
|
/** |
|
139
|
|
|
* @return Token |
|
140
|
|
|
* @throws LexerException |
|
141
|
|
|
*/ |
|
142
|
|
|
private function parseNumber(): Token |
|
143
|
|
|
{ |
|
144
|
|
|
if ($this->currentChar === null) { |
|
145
|
|
|
throw new LexerException('Cannot found current char'); |
|
146
|
|
|
} |
|
147
|
|
|
$number = ''; |
|
148
|
|
|
while (is_numeric($this->currentChar) || ($number !== '' && $this->currentChar === '.')) { |
|
149
|
|
|
$number .= $this->currentChar; |
|
150
|
|
|
$this->advance(); |
|
151
|
|
|
} |
|
152
|
|
|
|
|
153
|
|
|
if ($this->validateNumber($number) === false) { |
|
154
|
|
|
throw new LexerException('Parsed number not valid'); |
|
155
|
|
|
} |
|
156
|
|
|
|
|
157
|
|
|
if (strpos($number, '.')) { |
|
158
|
|
|
return new Token(Token::REAL, (float)$number); |
|
159
|
|
|
} |
|
160
|
|
|
|
|
161
|
|
|
return new Token(Token::INT, (int)$number); |
|
162
|
|
|
} |
|
163
|
|
|
|
|
164
|
|
|
private function validateNumber(string $number): bool |
|
165
|
|
|
{ |
|
166
|
|
|
return (bool)preg_match('/^[\d]+\.?[\d]*/', $number); |
|
167
|
|
|
} |
|
168
|
|
|
|
|
169
|
|
|
/** |
|
170
|
|
|
* @return Token |
|
171
|
|
|
*/ |
|
172
|
|
|
private function analyzeAsterisk(): Token |
|
173
|
|
|
{ |
|
174
|
|
|
if ($this->peek() === '*') { |
|
175
|
|
|
$this->advance(); |
|
176
|
|
|
$this->advance(); |
|
177
|
|
|
return new Token(Token::POWER, '**'); |
|
178
|
|
|
} |
|
179
|
|
|
|
|
180
|
|
|
$this->advance(); |
|
181
|
|
|
return new Token(Token::MUL, '*'); |
|
182
|
|
|
} |
|
183
|
|
|
} |
drumser /
arithmetic-parser
