|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Spatie\MixedContentScanner; |
|
4
|
|
|
|
|
5
|
|
|
use Spatie\Crawler\Crawler; |
|
6
|
|
|
use Spatie\Crawler\CrawlProfile; |
|
7
|
|
|
use Spatie\Crawler\CrawlInternalUrls; |
|
8
|
|
|
use Spatie\Crawler\Url; |
|
9
|
|
|
use Spatie\MixedContentScanner\Exceptions\InvalidUrl; |
|
10
|
|
|
|
|
11
|
|
|
class MixedContentScanner |
|
12
|
|
|
{ |
|
13
|
|
|
/** @var \Spatie\MixedContentScanner\MixedContentObserver */ |
|
14
|
|
|
public $mixedContentObserver; |
|
15
|
|
|
|
|
16
|
|
|
/** @var null|\Spatie\Crawler\CrawlProfile */ |
|
17
|
|
|
public $crawlProfile; |
|
18
|
|
|
|
|
19
|
|
|
public function __construct(MixedContentObserver $mixedContentObserver) |
|
20
|
|
|
{ |
|
21
|
|
|
$this->mixedContentObserver = $mixedContentObserver; |
|
22
|
|
|
} |
|
23
|
|
|
|
|
24
|
|
|
public function scan(string $url, array $clientOptions = []) |
|
25
|
|
|
{ |
|
26
|
|
|
$this->guardAgainstInvalidUrl($url); |
|
27
|
|
|
|
|
28
|
|
|
$url = Url::create($url); |
|
29
|
|
|
|
|
30
|
|
|
Crawler::create($clientOptions) |
|
31
|
|
|
->setCrawlProfile($this->crawlProfile ?? new CrawlInternalUrls($url)) |
|
32
|
|
|
->setCrawlObserver($this->mixedContentObserver) |
|
33
|
|
|
->startCrawling($url); |
|
34
|
|
|
} |
|
35
|
|
|
|
|
36
|
|
|
public function setCrawlProfile(CrawlProfile $crawlProfile) |
|
37
|
|
|
{ |
|
38
|
|
|
$this->crawlProfile = $crawlProfile; |
|
39
|
|
|
|
|
40
|
|
|
return $this; |
|
41
|
|
|
} |
|
42
|
|
|
|
|
43
|
|
|
protected function guardAgainstInvalidUrl(string $url) |
|
44
|
|
|
{ |
|
45
|
|
|
if ($url == '') { |
|
46
|
|
|
throw InvalidUrl::urlIsEmpty(); |
|
47
|
|
|
} |
|
48
|
|
|
|
|
49
|
|
|
if (! $this->startsWith($url, ['http://', 'https://'])) { |
|
50
|
|
|
throw InvalidUrl::invalidScheme($url); |
|
51
|
|
|
} |
|
52
|
|
|
} |
|
53
|
|
|
|
|
54
|
|
|
protected function startsWith(string $haystack, array $needles): bool |
|
55
|
|
|
{ |
|
56
|
|
|
foreach ((array) $needles as $needle) { |
|
57
|
|
|
if ($needle != '' && substr($haystack, 0, strlen($needle)) === (string) $needle) { |
|
58
|
|
|
return true; |
|
59
|
|
|
} |
|
60
|
|
|
} |
|
61
|
|
|
|
|
62
|
|
|
return false; |
|
63
|
|
|
} |
|
64
|
|
|
} |
|
65
|
|
|
|
spatie /
mixed-content-scanner
