1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
/* |
4
|
|
|
* This file is part of Jitamin. |
5
|
|
|
* |
6
|
|
|
* Copyright (C) Jitamin Team |
7
|
|
|
* |
8
|
|
|
* For the full copyright and license information, please view the LICENSE |
9
|
|
|
* file that was distributed with this source code. |
10
|
|
|
*/ |
11
|
|
|
|
12
|
|
|
namespace Jitamin\Foundation\Filter; |
13
|
|
|
|
14
|
|
|
/** |
15
|
|
|
* Lexer. |
16
|
|
|
*/ |
17
|
|
|
class Lexer |
18
|
|
|
{ |
19
|
|
|
/** |
20
|
|
|
* Current position. |
21
|
|
|
* |
22
|
|
|
* @var int |
23
|
|
|
*/ |
24
|
|
|
private $offset = 0; |
25
|
|
|
|
26
|
|
|
/** |
27
|
|
|
* Token map. |
28
|
|
|
* |
29
|
|
|
* @var array |
30
|
|
|
*/ |
31
|
|
|
private $tokenMap = [ |
32
|
|
|
'/^(\s+)/' => 'T_WHITESPACE', |
33
|
|
|
'/^([<=>]{0,2}[0-9]{4}-[0-9]{2}-[0-9]{2})/' => 'T_STRING', |
34
|
|
|
'/^([<=>]{1,2}\w+)/u' => 'T_STRING', |
35
|
|
|
'/^([<=>]{1,2}".+")/' => 'T_STRING', |
36
|
|
|
'/^("(.+)")/' => 'T_STRING', |
37
|
|
|
'/^(\S+)/u' => 'T_STRING', |
38
|
|
|
'/^(#\d+)/' => 'T_STRING', |
39
|
|
|
]; |
40
|
|
|
|
41
|
|
|
/** |
42
|
|
|
* Default token. |
43
|
|
|
* |
44
|
|
|
* @var string |
45
|
|
|
*/ |
46
|
|
|
private $defaultToken = ''; |
47
|
|
|
|
48
|
|
|
/** |
49
|
|
|
* Add token. |
50
|
|
|
* |
51
|
|
|
* @param string $regex |
52
|
|
|
* @param string $token |
53
|
|
|
* |
54
|
|
|
* @return $this |
55
|
|
|
*/ |
56
|
|
|
public function addToken($regex, $token) |
57
|
|
|
{ |
58
|
|
|
$this->tokenMap = [$regex => $token] + $this->tokenMap; |
59
|
|
|
|
60
|
|
|
return $this; |
61
|
|
|
} |
62
|
|
|
|
63
|
|
|
/** |
64
|
|
|
* Set default token. |
65
|
|
|
* |
66
|
|
|
* @param string $token |
67
|
|
|
* |
68
|
|
|
* @return $this |
69
|
|
|
*/ |
70
|
|
|
public function setDefaultToken($token) |
71
|
|
|
{ |
72
|
|
|
$this->defaultToken = $token; |
73
|
|
|
|
74
|
|
|
return $this; |
75
|
|
|
} |
76
|
|
|
|
77
|
|
|
/** |
78
|
|
|
* Tokenize input string. |
79
|
|
|
* |
80
|
|
|
* @param string $input |
81
|
|
|
* |
82
|
|
|
* @return array |
83
|
|
|
*/ |
84
|
|
|
public function tokenize($input) |
85
|
|
|
{ |
86
|
|
|
$tokens = []; |
87
|
|
|
$this->offset = 0; |
88
|
|
|
$input_length = mb_strlen($input, 'UTF-8'); |
89
|
|
|
|
90
|
|
|
while ($this->offset < $input_length) { |
91
|
|
|
$result = $this->match(mb_substr($input, $this->offset, $input_length, 'UTF-8')); |
92
|
|
|
|
93
|
|
|
if ($result === false) { |
94
|
|
|
return []; |
95
|
|
|
} |
96
|
|
|
|
97
|
|
|
$tokens[] = $result; |
98
|
|
|
} |
99
|
|
|
|
100
|
|
|
return $this->map($tokens); |
101
|
|
|
} |
102
|
|
|
|
103
|
|
|
/** |
104
|
|
|
* Find a token that match and move the offset. |
105
|
|
|
* |
106
|
|
|
* @param string $string |
107
|
|
|
* |
108
|
|
|
* @return array|bool |
109
|
|
|
*/ |
110
|
|
|
protected function match($string) |
111
|
|
|
{ |
112
|
|
|
foreach ($this->tokenMap as $pattern => $name) { |
113
|
|
|
if (preg_match($pattern, $string, $matches)) { |
114
|
|
|
$this->offset += mb_strlen($matches[1], 'UTF-8'); |
115
|
|
|
|
116
|
|
|
return [ |
117
|
|
|
'match' => str_replace('"', '', $matches[1]), |
118
|
|
|
'token' => $name, |
119
|
|
|
]; |
120
|
|
|
} |
121
|
|
|
} |
122
|
|
|
|
123
|
|
|
return false; |
124
|
|
|
} |
125
|
|
|
|
126
|
|
|
/** |
127
|
|
|
* Build map of tokens and matches. |
128
|
|
|
* |
129
|
|
|
* @param array $tokens |
130
|
|
|
* |
131
|
|
|
* @return array |
132
|
|
|
*/ |
133
|
|
|
protected function map(array $tokens) |
134
|
|
|
{ |
135
|
|
|
$map = []; |
136
|
|
|
$leftOver = ''; |
137
|
|
|
|
138
|
|
|
while (false !== ($token = current($tokens))) { |
139
|
|
|
if ($token['token'] === 'T_STRING' || $token['token'] === 'T_WHITESPACE') { |
140
|
|
|
$leftOver .= $token['match']; |
141
|
|
|
} else { |
142
|
|
|
$next = next($tokens); |
143
|
|
|
|
144
|
|
|
if ($next !== false && $next['token'] === 'T_STRING') { |
145
|
|
|
$map[$token['token']][] = $next['match']; |
146
|
|
|
} |
147
|
|
|
} |
148
|
|
|
|
149
|
|
|
next($tokens); |
150
|
|
|
} |
151
|
|
|
|
152
|
|
|
$leftOver = trim($leftOver); |
153
|
|
|
|
154
|
|
|
if ($this->defaultToken !== '' && $leftOver !== '') { |
155
|
|
|
$map[$this->defaultToken] = [$leftOver]; |
156
|
|
|
} |
157
|
|
|
|
158
|
|
|
return $map; |
159
|
|
|
} |
160
|
|
|
} |
161
|
|
|
|