|
1
|
|
|
<?php |
|
2
|
|
|
/* |
|
3
|
|
|
* This file is part of dispositif/wikibot application (@github) |
|
4
|
|
|
* 2019-2023 © Philippe M./Irønie <[email protected]> |
|
5
|
|
|
* For the full copyright and MIT license information, view the license file. |
|
6
|
|
|
*/ |
|
7
|
|
|
|
|
8
|
|
|
declare(strict_types=1); |
|
9
|
|
|
|
|
10
|
|
|
namespace App\Infrastructure; |
|
11
|
|
|
|
|
12
|
|
|
use App\Application\Utils\HttpUtil; |
|
13
|
|
|
use App\Domain\InfrastructurePorts\InternetDomainParserInterface; |
|
14
|
|
|
use Exception; |
|
15
|
|
|
use Pdp\Domain; |
|
16
|
|
|
use Pdp\ResolvedDomainName; |
|
17
|
|
|
use Pdp\Rules; |
|
18
|
|
|
|
|
19
|
|
|
/** |
|
20
|
|
|
* Doc https://packagist.org/packages/jeremykendall/php-domain-parser |
|
21
|
|
|
*/ |
|
22
|
|
|
class InternetDomainParser implements InternetDomainParserInterface |
|
23
|
|
|
{ |
|
24
|
|
|
private const PATH_CACHE_PUBLIC_SUFFIX_LIST = __DIR__ . '/resources/public_suffix_list.dat'; |
|
25
|
|
|
|
|
26
|
|
|
private readonly Rules $rules; |
|
27
|
|
|
|
|
28
|
|
|
public function __construct() |
|
29
|
|
|
{ |
|
30
|
|
|
if (!file_exists(self::PATH_CACHE_PUBLIC_SUFFIX_LIST)) { |
|
31
|
|
|
throw new Exception('Public suffix list not found'); |
|
32
|
|
|
} |
|
33
|
|
|
$this->rules = Rules::fromPath(self::PATH_CACHE_PUBLIC_SUFFIX_LIST); |
|
|
|
|
|
|
34
|
|
|
} |
|
35
|
|
|
|
|
36
|
|
|
|
|
37
|
|
|
/** |
|
38
|
|
|
* https://www.google.fr => google.fr |
|
39
|
|
|
* http://fu.co.uk => fu.co.uk |
|
40
|
|
|
* @throws Exception |
|
41
|
|
|
*/ |
|
42
|
|
|
public function getRegistrableDomainFromURL(string $httpURL): string |
|
43
|
|
|
{ |
|
44
|
|
|
$result = $this->getResolvedDomainName($httpURL); |
|
45
|
|
|
|
|
46
|
|
|
return $result->registrableDomain()->toString(); |
|
47
|
|
|
} |
|
48
|
|
|
|
|
49
|
|
|
/** |
|
50
|
|
|
* todo move ? HttpUtil |
|
51
|
|
|
* Ok static method (only native php parsing). |
|
52
|
|
|
* https://www.google.fr => google.fr |
|
53
|
|
|
* http://fu.co.uk => fu.co.uk |
|
54
|
|
|
*/ |
|
55
|
|
|
public static function extractSubdomainString(string $httpURL): string |
|
56
|
|
|
{ |
|
57
|
|
|
if (!HttpUtil::isHttpURL($httpURL)) { |
|
58
|
|
|
throw new Exception('string is not an URL ' . $httpURL); |
|
59
|
|
|
} |
|
60
|
|
|
|
|
61
|
|
|
return parse_url($httpURL, PHP_URL_HOST); |
|
62
|
|
|
} |
|
63
|
|
|
|
|
64
|
|
|
protected function getResolvedDomainName(string $httpURL): ResolvedDomainName |
|
65
|
|
|
{ |
|
66
|
|
|
$domain = Domain::fromIDNA2008(parse_url($httpURL, PHP_URL_HOST)); |
|
67
|
|
|
|
|
68
|
|
|
return $this->rules->resolve($domain); |
|
69
|
|
|
} |
|
70
|
|
|
} |
Dispositif /
Wikibot
Loading history...