|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Spatie\MixedContentScanner; |
|
4
|
|
|
|
|
5
|
|
|
use Exception; |
|
6
|
|
|
use GuzzleHttp\Psr7\Uri; |
|
7
|
|
|
use Illuminate\Support\Collection; |
|
8
|
|
|
use Symfony\Component\DomCrawler\Crawler as DomCrawler; |
|
9
|
|
|
|
|
10
|
|
|
class MixedContentExtractor |
|
11
|
|
|
{ |
|
12
|
|
|
public static function extract(string $html, string $currentUri): array |
|
13
|
|
|
{ |
|
14
|
|
|
return static::getSearchNodes() |
|
15
|
|
|
->mapSpread(function ($tagName, $attribute) use ($html, $currentUri) { |
|
16
|
|
|
return (new DomCrawler($html, $currentUri)) |
|
17
|
|
|
->filterXPath("//{$tagName}[@{$attribute}]") |
|
18
|
|
|
->reduce(function (DomCrawler $node) { |
|
19
|
|
|
return ! self::isShortLink($node); |
|
20
|
|
|
}) |
|
21
|
|
|
->each(function (DomCrawler $node) use ($tagName, $attribute) { |
|
22
|
|
|
try { |
|
23
|
|
|
$url = new Uri($node->attr($attribute)); |
|
24
|
|
|
|
|
25
|
|
|
if ($tagName === 'link' && $attribute === 'href') { |
|
26
|
|
|
if ($node->attr('rel') !== 'stylesheet') { |
|
27
|
|
|
return; |
|
28
|
|
|
} |
|
29
|
|
|
} |
|
30
|
|
|
|
|
31
|
|
|
return $url->getScheme() === 'http' ? [$tagName, $url] : null; |
|
32
|
|
|
} catch (Exception $e) { |
|
33
|
|
|
// ignore invalid links |
|
34
|
|
|
} |
|
35
|
|
|
}); |
|
36
|
|
|
}) |
|
37
|
|
|
->flatten(1) |
|
38
|
|
|
->filter() |
|
39
|
|
|
->mapSpread(function ($tagName, $mixedContentUrl) use ($currentUri) { |
|
40
|
|
|
return new MixedContent($tagName, $mixedContentUrl, new Uri($currentUri)); |
|
41
|
|
|
}) |
|
42
|
|
|
->toArray(); |
|
43
|
|
|
} |
|
44
|
|
|
|
|
45
|
|
|
protected static function getSearchNodes(): Collection |
|
46
|
|
|
{ |
|
47
|
|
|
return collect([ |
|
48
|
|
|
['audio', 'src'], |
|
49
|
|
|
['embed', 'src'], |
|
50
|
|
|
['form', 'action'], |
|
51
|
|
|
['link', 'href'], |
|
52
|
|
|
['iframe', 'src'], |
|
53
|
|
|
['img', 'src'], |
|
54
|
|
|
['img', 'srcset'], |
|
55
|
|
|
['object', 'data'], |
|
56
|
|
|
['param', 'value'], |
|
57
|
|
|
['script', 'src'], |
|
58
|
|
|
['source', 'src'], |
|
59
|
|
|
['source', 'srcset'], |
|
60
|
|
|
['video', 'src'], |
|
61
|
|
|
]); |
|
62
|
|
|
} |
|
63
|
|
|
|
|
64
|
|
|
protected static function isShortLink(DomCrawler $node): bool |
|
65
|
|
|
{ |
|
66
|
|
|
$relAttribute = $node->getNode(0)->attributes->getNamedItem('rel'); |
|
67
|
|
|
|
|
68
|
|
|
if (is_null($relAttribute)) { |
|
69
|
|
|
return false; |
|
70
|
|
|
} |
|
71
|
|
|
|
|
72
|
|
|
return strtolower($relAttribute->nodeValue) === 'shortlink'; |
|
73
|
|
|
} |
|
74
|
|
|
} |
|
75
|
|
|
|
spatie /
mixed-content-scanner
