|
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\Domain\ExternLink; |
|
11
|
|
|
|
|
12
|
|
|
use App\Application\Http\ExternHttpClient; |
|
13
|
|
|
use App\Infrastructure\InternetDomainParser; |
|
14
|
|
|
use Exception; |
|
15
|
|
|
use Psr\Log\LoggerInterface; |
|
16
|
|
|
use Psr\Log\NullLogger; |
|
17
|
|
|
|
|
18
|
|
|
/** |
|
19
|
|
|
* Todo move infra ? |
|
20
|
|
|
*/ |
|
21
|
|
|
class CheckURL |
|
22
|
|
|
{ |
|
23
|
|
|
/** |
|
24
|
|
|
* @var LoggerInterface |
|
25
|
|
|
*/ |
|
26
|
|
|
protected $log; |
|
27
|
|
|
|
|
28
|
|
|
/** |
|
29
|
|
|
* @var string |
|
30
|
|
|
*/ |
|
31
|
|
|
protected $registrableDomain; |
|
32
|
|
|
/** |
|
33
|
|
|
* @var string |
|
34
|
|
|
*/ |
|
35
|
|
|
protected $url; |
|
36
|
|
|
|
|
37
|
|
|
public function __construct(?LoggerInterface $logger = null) |
|
38
|
|
|
{ |
|
39
|
|
|
$this->log = $logger ?? new NullLogger(); |
|
40
|
|
|
} |
|
41
|
|
|
|
|
42
|
|
|
public function isURLAuthorized(string $url): bool |
|
43
|
|
|
{ |
|
44
|
|
|
$this->url = $url; |
|
45
|
|
|
$this->registrableDomain = null; |
|
46
|
|
|
if (!ExternHttpClient::isHttpURL($url)) { |
|
47
|
|
|
$this->log->debug('Skip : not a valid URL : ' . $url); |
|
48
|
|
|
return false; |
|
49
|
|
|
} |
|
50
|
|
|
|
|
51
|
|
|
if ($this->hasForbiddenFilenameExtension()) { |
|
52
|
|
|
return false; |
|
53
|
|
|
} |
|
54
|
|
|
if (!ExternHttpClient::isHttpURL($url)) { |
|
55
|
|
|
throw new Exception('string is not an URL ' . $url); |
|
56
|
|
|
} |
|
57
|
|
|
|
|
58
|
|
|
$this->findRegistrableDomain(); |
|
59
|
|
|
|
|
60
|
|
|
return true; |
|
61
|
|
|
} |
|
62
|
|
|
|
|
63
|
|
|
public function getRegistrableDomain($url): ?string |
|
64
|
|
|
{ |
|
65
|
|
|
if ($url === $this->url && $this->registrableDomain) { |
|
66
|
|
|
return $this->registrableDomain; |
|
67
|
|
|
} |
|
68
|
|
|
$this->url = $url; |
|
69
|
|
|
return $this->findRegistrableDomain(); |
|
70
|
|
|
} |
|
71
|
|
|
|
|
72
|
|
|
protected function findRegistrableDomain(): ?string |
|
73
|
|
|
{ |
|
74
|
|
|
try { |
|
75
|
|
|
$this->registrableDomain = (new InternetDomainParser())->getRegistrableDomainFromURL($this->url); |
|
76
|
|
|
} catch (Exception $e) { |
|
77
|
|
|
$this->log->warning('Skip : not a valid URL : ' . $this->url); |
|
78
|
|
|
return null; |
|
79
|
|
|
} |
|
80
|
|
|
return $this->registrableDomain; |
|
81
|
|
|
} |
|
82
|
|
|
|
|
83
|
|
|
/** |
|
84
|
|
|
* todo move URL parsing |
|
85
|
|
|
* Skip PDF GIF etc |
|
86
|
|
|
* https://fr.wikipedia.org/wiki/Liste_d%27extensions_de_fichiers |
|
87
|
|
|
* |
|
88
|
|
|
* @param string $url |
|
89
|
|
|
* |
|
90
|
|
|
* @return bool |
|
91
|
|
|
*/ |
|
92
|
|
|
protected function hasForbiddenFilenameExtension(): bool |
|
93
|
|
|
{ |
|
94
|
|
|
return (bool)preg_match( |
|
95
|
|
|
'#\.(pdf|jpg|jpeg|gif|png|xls|xlsx|xlr|xml|xlt|txt|csv|js|docx|exe|gz|zip|ini|movie|mp3|mp4|ogg|raw|rss|tar|tgz|wma)$#i', |
|
96
|
|
|
$this->url |
|
97
|
|
|
); |
|
98
|
|
|
} |
|
99
|
|
|
} |