|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
declare(strict_types=1); |
|
4
|
|
|
|
|
5
|
|
|
namespace PHPHtmlParser\Selector; |
|
6
|
|
|
|
|
7
|
|
|
/** |
|
8
|
|
|
* This is the parser for the selector. |
|
9
|
|
|
* |
|
10
|
|
|
* |
|
11
|
|
|
*/ |
|
12
|
|
|
class Parser implements ParserInterface |
|
13
|
|
|
{ |
|
14
|
|
|
|
|
15
|
|
|
private static $instance = null; |
|
16
|
|
|
|
|
17
|
315 |
|
public static function getInstance(): Parser { |
|
18
|
315 |
|
if (is_null(self::$instance)) { |
|
19
|
3 |
|
self::$instance = new Parser(); |
|
20
|
|
|
} |
|
21
|
315 |
|
return self::$instance; |
|
22
|
|
|
} |
|
23
|
|
|
|
|
24
|
3 |
|
private function __construct() { |
|
25
|
|
|
// use getInstance() |
|
26
|
3 |
|
} |
|
27
|
|
|
|
|
28
|
|
|
/** |
|
29
|
|
|
* Pattern of CSS selectors, modified from 'mootools' |
|
30
|
|
|
* |
|
31
|
|
|
* @var string |
|
32
|
|
|
*/ |
|
33
|
|
|
protected $pattern = "/([\w\-:\*>]*)(?:\#([\w\-]+)|\.([\w\.\-]+))?(?:\[@?(!?[\w\-:]+)(?:([!*^$]?=)[\"']?(.*?)[\"']?)?\])?([\/, ]+)/is"; |
|
34
|
|
|
|
|
35
|
|
|
/** |
|
36
|
|
|
* Parses the selector string |
|
37
|
|
|
* |
|
38
|
|
|
* @param string $selector |
|
39
|
|
|
* |
|
40
|
|
|
* @return array |
|
41
|
|
|
*/ |
|
42
|
315 |
|
public function parseSelectorString(string $selector): array |
|
43
|
|
|
{ |
|
44
|
315 |
|
$selectors = []; |
|
45
|
|
|
|
|
46
|
315 |
|
$matches = []; |
|
47
|
315 |
|
preg_match_all($this->pattern, trim($selector).' ', $matches, PREG_SET_ORDER); |
|
48
|
|
|
|
|
49
|
|
|
// skip tbody |
|
50
|
315 |
|
$result = []; |
|
51
|
315 |
|
foreach ($matches as $match) { |
|
52
|
|
|
// default values |
|
53
|
315 |
|
$tag = strtolower(trim($match[1])); |
|
54
|
315 |
|
$operator = '='; |
|
55
|
315 |
|
$key = null; |
|
56
|
315 |
|
$value = null; |
|
57
|
315 |
|
$noKey = false; |
|
58
|
315 |
|
$alterNext = false; |
|
59
|
|
|
|
|
60
|
|
|
// check for elements that alter the behavior of the next element |
|
61
|
315 |
|
if ($tag == '>') { |
|
62
|
3 |
|
$alterNext = true; |
|
63
|
|
|
} |
|
64
|
|
|
|
|
65
|
|
|
// check for id selector |
|
66
|
315 |
|
if ( ! empty($match[2])) { |
|
67
|
21 |
|
$key = 'id'; |
|
68
|
21 |
|
$value = $match[2]; |
|
69
|
|
|
} |
|
70
|
|
|
|
|
71
|
|
|
// check for class selector |
|
72
|
315 |
|
if ( ! empty($match[3])) { |
|
73
|
54 |
|
$key = 'class'; |
|
74
|
54 |
|
$value = explode('.', $match[3]); |
|
75
|
|
|
} |
|
76
|
|
|
|
|
77
|
|
|
// and final attribute selector |
|
78
|
315 |
|
if ( ! empty($match[4])) { |
|
79
|
255 |
|
$key = strtolower($match[4]); |
|
80
|
|
|
} |
|
81
|
315 |
|
if ( ! empty($match[5])) { |
|
82
|
246 |
|
$operator = $match[5]; |
|
83
|
|
|
} |
|
84
|
315 |
|
if ( ! empty($match[6])) { |
|
85
|
246 |
|
$value = $match[6]; |
|
86
|
246 |
|
if (strpos($value, '][') !== false) { |
|
87
|
|
|
// we have multiple type selectors |
|
88
|
3 |
|
$keys = []; |
|
89
|
3 |
|
$keys[] = $key; |
|
90
|
3 |
|
$key = $keys; |
|
91
|
3 |
|
$parts = explode('][', $value); |
|
92
|
3 |
|
$value = []; |
|
93
|
3 |
|
foreach ($parts as $part) { |
|
94
|
3 |
|
if (strpos($part, '=') !== false) { |
|
95
|
3 |
|
list($first, $second) = explode('=', $part); |
|
96
|
3 |
|
$key[] = $first; |
|
97
|
3 |
|
$value[] = $second; |
|
98
|
|
|
} else { |
|
99
|
3 |
|
$value[] = $part; |
|
100
|
|
|
} |
|
101
|
|
|
} |
|
102
|
|
|
} |
|
103
|
|
|
} |
|
104
|
|
|
|
|
105
|
|
|
// check for elements that do not have a specified attribute |
|
106
|
315 |
|
if (is_string($key) && isset($key[0]) && $key[0] == '!') { |
|
107
|
3 |
|
$key = substr($key, 1); |
|
108
|
3 |
|
$noKey = true; |
|
109
|
|
|
} |
|
110
|
|
|
|
|
111
|
315 |
|
$result[] = [ |
|
112
|
315 |
|
'tag' => $tag, |
|
113
|
315 |
|
'key' => $key, |
|
114
|
315 |
|
'value' => $value, |
|
115
|
315 |
|
'operator' => $operator, |
|
116
|
315 |
|
'noKey' => $noKey, |
|
117
|
315 |
|
'alterNext' => $alterNext, |
|
118
|
|
|
]; |
|
119
|
315 |
|
if (isset($match[7]) && is_string($match[7]) && trim($match[7]) == ',') { |
|
120
|
3 |
|
$selectors[] = $result; |
|
121
|
107 |
|
$result = []; |
|
122
|
|
|
} |
|
123
|
|
|
} |
|
124
|
|
|
|
|
125
|
|
|
// save last results |
|
126
|
315 |
|
if (count($result) > 0) { |
|
127
|
315 |
|
$selectors[] = $result; |
|
128
|
|
|
} |
|
129
|
|
|
|
|
130
|
315 |
|
return $selectors; |
|
131
|
|
|
} |
|
132
|
|
|
} |
|
133
|
|
|
|
paquettg /
php-html-parser
