1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
declare(strict_types=1); |
4
|
|
|
|
5
|
|
|
namespace PHPHtmlParser\Selector; |
6
|
|
|
|
7
|
|
|
use PHPHtmlParser\Contracts\Selector\ParserInterface; |
8
|
|
|
use PHPHtmlParser\DTO\Selector\ParsedSelectorCollectionDTO; |
9
|
|
|
use PHPHtmlParser\DTO\Selector\ParsedSelectorDTO; |
10
|
|
|
use PHPHtmlParser\DTO\Selector\RuleDTO; |
11
|
|
|
|
12
|
|
|
/** |
13
|
|
|
* This is the default parser for the selector. |
14
|
|
|
*/ |
15
|
|
|
class Parser implements ParserInterface |
16
|
|
|
{ |
17
|
|
|
/** |
18
|
|
|
* Pattern of CSS selectors, modified from 'mootools'. |
19
|
|
|
* |
20
|
|
|
* @var string |
21
|
|
|
*/ |
22
|
|
|
private $pattern = "/([\w\-:\*>]*)(?:\#([\w\-]+)|\.([\w\.\-]+))?(?:\[@?(!?[\w\-:]+)(?:([!*^$]?=)[\"']?(.*?)[\"']?)?\])?([\/, ]+)/is"; |
23
|
|
|
|
24
|
|
|
/** |
25
|
|
|
* Parses the selector string. |
26
|
|
|
*/ |
27
|
351 |
|
public function parseSelectorString(string $selector): ParsedSelectorCollectionDTO |
28
|
|
|
{ |
29
|
351 |
|
$selectors = []; |
30
|
351 |
|
$matches = []; |
31
|
351 |
|
$rules = []; |
32
|
351 |
|
\preg_match_all($this->pattern, \trim($selector) . ' ', $matches, PREG_SET_ORDER); |
33
|
|
|
|
34
|
|
|
// skip tbody |
35
|
351 |
|
foreach ($matches as $match) { |
36
|
|
|
// default values |
37
|
351 |
|
$tag = \strtolower(\trim($match[1])); |
38
|
351 |
|
$operator = '='; |
39
|
351 |
|
$key = null; |
40
|
351 |
|
$value = null; |
41
|
351 |
|
$noKey = false; |
42
|
351 |
|
$alterNext = false; |
43
|
|
|
|
44
|
|
|
// check for elements that alter the behavior of the next element |
45
|
351 |
|
if ($tag == '>') { |
46
|
3 |
|
$alterNext = true; |
47
|
|
|
} |
48
|
|
|
|
49
|
|
|
// check for id selector |
50
|
351 |
|
if (!empty($match[2])) { |
51
|
21 |
|
$key = 'id'; |
52
|
21 |
|
$value = $match[2]; |
53
|
|
|
} |
54
|
|
|
|
55
|
|
|
// check for class selector |
56
|
351 |
|
if (!empty($match[3])) { |
57
|
57 |
|
$key = 'class'; |
58
|
57 |
|
$value = \explode('.', $match[3]); |
59
|
|
|
} |
60
|
|
|
|
61
|
|
|
// and final attribute selector |
62
|
351 |
|
if (!empty($match[4])) { |
63
|
300 |
|
$key = \strtolower($match[4]); |
64
|
|
|
} |
65
|
351 |
|
if (!empty($match[5])) { |
66
|
291 |
|
$operator = $match[5]; |
67
|
|
|
} |
68
|
351 |
|
if (!empty($match[6])) { |
69
|
291 |
|
$value = $match[6]; |
70
|
291 |
|
if (\strpos($value, '][') !== false) { |
71
|
|
|
// we have multiple type selectors |
72
|
3 |
|
$keys = []; |
73
|
3 |
|
$keys[] = $key; |
74
|
3 |
|
$key = $keys; |
75
|
3 |
|
$parts = \explode('][', $value); |
76
|
3 |
|
$value = []; |
77
|
3 |
|
foreach ($parts as $part) { |
78
|
3 |
|
if (\strpos($part, '=') !== false) { |
79
|
3 |
|
list($first, $second) = \explode('=', $part); |
80
|
3 |
|
$key[] = $first; |
81
|
3 |
|
$value[] = $second; |
82
|
|
|
} else { |
83
|
3 |
|
$value[] = $part; |
84
|
|
|
} |
85
|
|
|
} |
86
|
|
|
} |
87
|
|
|
} |
88
|
|
|
|
89
|
|
|
// check for elements that do not have a specified attribute |
90
|
351 |
|
if (\is_string($key) && isset($key[0]) && $key[0] == '!') { |
91
|
3 |
|
$key = \substr($key, 1); |
92
|
3 |
|
$noKey = true; |
93
|
|
|
} |
94
|
|
|
|
95
|
351 |
|
$rules[] = RuleDTO::makeFromPrimitives( |
96
|
351 |
|
$tag, |
97
|
234 |
|
$operator, |
98
|
234 |
|
$key, |
99
|
234 |
|
$value, |
100
|
234 |
|
$noKey, |
101
|
234 |
|
$alterNext |
102
|
|
|
); |
103
|
351 |
|
if (isset($match[7]) && \is_string($match[7]) && \trim($match[7]) == ',') { |
104
|
3 |
|
$selectors[] = ParsedSelectorDTO::makeFromRules($rules); |
105
|
3 |
|
$rules = []; |
106
|
|
|
} |
107
|
|
|
} |
108
|
|
|
|
109
|
|
|
// save last results |
110
|
351 |
|
if (\count($rules) > 0) { |
111
|
351 |
|
$selectors[] = ParsedSelectorDTO::makeFromRules($rules); |
112
|
|
|
} |
113
|
|
|
|
114
|
351 |
|
return ParsedSelectorCollectionDTO::makeCollection($selectors); |
115
|
|
|
} |
116
|
|
|
} |
117
|
|
|
|