|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
/** |
|
4
|
|
|
* @package s9e\TextFormatter |
|
5
|
|
|
* @copyright Copyright (c) 2010-2016 The s9e Authors |
|
6
|
|
|
* @license http://www.opensource.org/licenses/mit-license.php The MIT License |
|
7
|
|
|
*/ |
|
8
|
|
|
namespace s9e\TextFormatter\Configurator\Items; |
|
9
|
|
|
|
|
10
|
|
|
use InvalidArgumentException; |
|
11
|
|
|
use s9e\TextFormatter\Configurator\ConfigProvider; |
|
12
|
|
|
use s9e\TextFormatter\Configurator\Helpers\RegexpParser; |
|
13
|
|
|
use s9e\TextFormatter\Configurator\JavaScript\RegexpConvertor; |
|
14
|
|
|
|
|
15
|
|
|
class Regexp implements ConfigProvider |
|
16
|
|
|
{ |
|
17
|
|
|
/** |
|
18
|
|
|
* @var bool Whether this regexp should have the global flag set in JavaScript |
|
19
|
|
|
*/ |
|
20
|
|
|
protected $isGlobal; |
|
21
|
|
|
|
|
22
|
|
|
/** |
|
23
|
|
|
* @var string JavaScript regexp, with delimiters and modifiers, e.g. "/foo/i" |
|
24
|
|
|
*/ |
|
25
|
|
|
protected $jsRegexp; |
|
26
|
|
|
|
|
27
|
|
|
/** |
|
28
|
|
|
* @var string PCRE regexp, with delimiters and modifiers, e.g. "/foo/i" |
|
29
|
|
|
*/ |
|
30
|
|
|
protected $regexp; |
|
31
|
|
|
|
|
32
|
|
|
/** |
|
33
|
|
|
* Constructor |
|
34
|
|
|
* |
|
35
|
|
|
* @param string $regexp PCRE regexp, with delimiters and modifiers, e.g. "/foo/i" |
|
36
|
|
|
*/ |
|
37
|
|
|
public function __construct($regexp, $isGlobal = false) |
|
38
|
|
|
{ |
|
39
|
|
|
if (@preg_match($regexp, '') === false) |
|
40
|
|
|
{ |
|
41
|
|
|
throw new InvalidArgumentException('Invalid regular expression ' . var_export($regexp, true)); |
|
42
|
|
|
} |
|
43
|
|
|
|
|
44
|
|
|
$this->regexp = $regexp; |
|
45
|
|
|
$this->isGlobal = $isGlobal; |
|
46
|
|
|
} |
|
47
|
|
|
|
|
48
|
|
|
/** |
|
49
|
|
|
* Return this regexp as a string |
|
50
|
|
|
* |
|
51
|
|
|
* @return string |
|
52
|
|
|
*/ |
|
53
|
|
|
public function __toString() |
|
54
|
|
|
{ |
|
55
|
|
|
return $this->regexp; |
|
56
|
|
|
} |
|
57
|
|
|
|
|
58
|
|
|
/** |
|
59
|
|
|
* {@inheritdoc} |
|
60
|
|
|
*/ |
|
61
|
|
|
public function asConfig() |
|
62
|
|
|
{ |
|
63
|
|
|
return $this; |
|
64
|
|
|
} |
|
65
|
|
|
|
|
66
|
|
|
/** |
|
67
|
|
|
* Return the name of each capture in this regexp |
|
68
|
|
|
* |
|
69
|
|
|
* @return string[] |
|
70
|
|
|
*/ |
|
71
|
|
|
public function getCaptureNames() |
|
72
|
|
|
{ |
|
73
|
|
|
return RegexpParser::getCaptureNames($this->regexp); |
|
74
|
|
|
} |
|
75
|
|
|
|
|
76
|
|
|
/** |
|
77
|
|
|
* Return this regexp's JavaScript representation |
|
78
|
|
|
* |
|
79
|
|
|
* @return string |
|
80
|
|
|
*/ |
|
81
|
|
|
public function getJS() |
|
82
|
|
|
{ |
|
83
|
|
|
if (!isset($this->jsRegexp)) |
|
84
|
|
|
{ |
|
85
|
|
|
$this->jsRegexp = RegexpConvertor::toJS($this->regexp, $this->isGlobal); |
|
86
|
|
|
} |
|
87
|
|
|
|
|
88
|
|
|
return $this->jsRegexp; |
|
89
|
|
|
} |
|
90
|
|
|
|
|
91
|
|
|
/** |
|
92
|
|
|
* Return all the named captures with a standalone regexp that matches them |
|
93
|
|
|
* |
|
94
|
|
|
* @return array Array of [capture name => regexp] |
|
95
|
|
|
*/ |
|
96
|
|
|
public function getNamedCaptures() |
|
97
|
|
|
{ |
|
98
|
|
|
$captures = []; |
|
99
|
|
|
$regexpInfo = RegexpParser::parse($this->regexp); |
|
100
|
|
|
|
|
101
|
|
|
// Prepare the start/end of the regexp and ensure that we use the D modifier |
|
102
|
|
|
$start = $regexpInfo['delimiter'] . '^'; |
|
103
|
|
|
$end = '$' . $regexpInfo['delimiter'] . $regexpInfo['modifiers']; |
|
104
|
|
|
if (strpos($regexpInfo['modifiers'], 'D') === false) |
|
105
|
|
|
{ |
|
106
|
|
|
$end .= 'D'; |
|
107
|
|
|
} |
|
108
|
|
|
|
|
109
|
|
|
foreach ($this->getNamedCapturesExpressions($regexpInfo['tokens']) as $name => $expr) |
|
110
|
|
|
{ |
|
111
|
|
|
$captures[$name] = $start . $expr . $end; |
|
112
|
|
|
} |
|
113
|
|
|
|
|
114
|
|
|
return $captures; |
|
115
|
|
|
} |
|
116
|
|
|
|
|
117
|
|
|
/** |
|
118
|
|
|
* Return the expression used in each named capture |
|
119
|
|
|
* |
|
120
|
|
|
* @param array[] $tokens |
|
121
|
|
|
* @return array |
|
122
|
|
|
*/ |
|
123
|
|
|
protected function getNamedCapturesExpressions(array $tokens) |
|
124
|
|
|
{ |
|
125
|
|
|
$exprs = []; |
|
126
|
|
|
foreach ($tokens as $token) |
|
127
|
|
|
{ |
|
128
|
|
|
if ($token['type'] !== 'capturingSubpatternStart' || !isset($token['name'])) |
|
129
|
|
|
{ |
|
130
|
|
|
continue; |
|
131
|
|
|
} |
|
132
|
|
|
|
|
133
|
|
|
$expr = $token['content']; |
|
134
|
|
|
if (strpos($expr, '|') !== false) |
|
135
|
|
|
{ |
|
136
|
|
|
$expr = '(?:' . $expr . ')'; |
|
137
|
|
|
} |
|
138
|
|
|
|
|
139
|
|
|
$exprs[$token['name']] = $expr; |
|
140
|
|
|
} |
|
141
|
|
|
|
|
142
|
|
|
return $exprs; |
|
143
|
|
|
} |
|
144
|
|
|
|
|
145
|
|
|
/** |
|
146
|
|
|
* Set this regexp's JavaScript representation |
|
147
|
|
|
* |
|
148
|
|
|
* @param string $jsRegexp |
|
149
|
|
|
* @return void |
|
150
|
|
|
*/ |
|
151
|
|
|
public function setJS($jsRegexp) |
|
152
|
|
|
{ |
|
153
|
|
|
$this->jsRegexp = $jsRegexp; |
|
154
|
|
|
} |
|
155
|
|
|
} |
s9e /
TextFormatter
