|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
declare(strict_types=1); |
|
4
|
|
|
|
|
5
|
|
|
namespace ReliqArts\StyleImporter\HTML\Extractor; |
|
6
|
|
|
|
|
7
|
|
|
use ReliqArts\StyleImporter\HTML\Extractor; |
|
8
|
|
|
|
|
9
|
|
|
final class ElementExtractor implements Extractor |
|
10
|
|
|
{ |
|
11
|
|
|
private const PATTERN_CLASSES = '#\<\w[^<>]*\sclass=[\"\\\']([\w\s-]+)[\"\\\'][\s\>\/]#'; |
|
12
|
|
|
private const PATTERN_IDS = '#\<\w[^<>]*\sid=[\"\\\']([\w-]+)[\"\\\'][\s\>\/]\>?#'; |
|
13
|
|
|
private const PATTERN_TAGS = '#\<([\w-]+)[\s\>]#'; |
|
14
|
|
|
|
|
15
|
|
|
/** |
|
16
|
|
|
* @param string $html |
|
17
|
|
|
* |
|
18
|
|
|
* @return string[] |
|
19
|
|
|
*/ |
|
20
|
|
|
public function extractClasses(string $html): array |
|
21
|
|
|
{ |
|
22
|
|
|
$extractedClasses = $this->extractItemsByPattern($html, self::PATTERN_CLASSES); |
|
23
|
|
|
$classNames = []; |
|
24
|
|
|
|
|
25
|
|
|
foreach ($extractedClasses as $combinedClass) { |
|
26
|
|
|
$splitClasses = array_map( |
|
27
|
|
|
function (string $class): string { |
|
28
|
|
|
return sprintf('.%s', $class); |
|
29
|
|
|
}, |
|
30
|
|
|
preg_split('/\s+/', $combinedClass) |
|
|
|
|
|
|
31
|
|
|
); |
|
32
|
|
|
|
|
33
|
|
|
array_push($classNames, ...$splitClasses); |
|
34
|
|
|
} |
|
35
|
|
|
|
|
36
|
|
|
return array_unique($classNames); |
|
37
|
|
|
} |
|
38
|
|
|
|
|
39
|
|
|
/** |
|
40
|
|
|
* @param string $html |
|
41
|
|
|
* |
|
42
|
|
|
* @return string[] |
|
43
|
|
|
*/ |
|
44
|
|
|
public function extractIds(string $html): array |
|
45
|
|
|
{ |
|
46
|
|
|
$idNames = $this->extractItemsByPattern($html, self::PATTERN_IDS); |
|
47
|
|
|
|
|
48
|
|
|
return array_map( |
|
49
|
|
|
function (string $id): string { |
|
50
|
|
|
return sprintf('#%s', $id); |
|
51
|
|
|
}, |
|
52
|
|
|
array_unique($idNames) |
|
53
|
|
|
); |
|
54
|
|
|
} |
|
55
|
|
|
|
|
56
|
|
|
/** |
|
57
|
|
|
* @param string $html |
|
58
|
|
|
* |
|
59
|
|
|
* @return string[] |
|
60
|
|
|
*/ |
|
61
|
|
|
public function extractTags(string $html): array |
|
62
|
|
|
{ |
|
63
|
|
|
return array_unique($this->extractItemsByPattern($html, self::PATTERN_TAGS)); |
|
64
|
|
|
} |
|
65
|
|
|
|
|
66
|
|
|
/** |
|
67
|
|
|
* {@inheritdoc} |
|
68
|
|
|
* |
|
69
|
|
|
* @param string $html |
|
70
|
|
|
* |
|
71
|
|
|
* @return array |
|
72
|
|
|
*/ |
|
73
|
|
|
public function extract(string $html): array |
|
74
|
|
|
{ |
|
75
|
|
|
return array_merge( |
|
76
|
|
|
$this->extractTags($html), |
|
77
|
|
|
$this->extractIds($html), |
|
78
|
|
|
$this->extractClasses($html) |
|
79
|
|
|
); |
|
80
|
|
|
} |
|
81
|
|
|
|
|
82
|
|
|
/** |
|
83
|
|
|
* @param string $subject |
|
84
|
|
|
* @param string $pattern |
|
85
|
|
|
* |
|
86
|
|
|
* @return array |
|
87
|
|
|
*/ |
|
88
|
|
|
private function extractItemsByPattern(string $subject, string $pattern): array |
|
89
|
|
|
{ |
|
90
|
|
|
$matchCount = preg_match_all($pattern, $subject, $matches); |
|
91
|
|
|
|
|
92
|
|
|
if (empty($matchCount)) { |
|
93
|
|
|
return []; |
|
94
|
|
|
} |
|
95
|
|
|
|
|
96
|
|
|
return $matches[1]; |
|
97
|
|
|
} |
|
98
|
|
|
} |
|
99
|
|
|
|
reliqarts /
laravel-style-importer
Loading history...