|
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 Psr\Log\LoggerInterface; |
|
13
|
|
|
use Psr\Log\NullLogger; |
|
14
|
|
|
|
|
15
|
|
|
class ExternHttpErrorLogic |
|
16
|
|
|
{ |
|
17
|
|
|
public const LOG_REQUEST_ERROR = __DIR__ . '/../../Application/resources/external_request_error.log'; |
|
18
|
|
|
|
|
19
|
|
|
/** |
|
20
|
|
|
* @var LoggerInterface |
|
21
|
|
|
*/ |
|
22
|
|
|
private $log; |
|
23
|
|
|
|
|
24
|
|
|
public function __construct(?LoggerInterface $log = null) |
|
25
|
|
|
{ |
|
26
|
|
|
$this->log = $log ?? new NullLogger(); |
|
27
|
|
|
} |
|
28
|
|
|
|
|
29
|
|
|
public function manageHttpErrors(string $errorMessage, string $url): string |
|
30
|
|
|
{ |
|
31
|
|
|
// "410 gone" => {lien brisé} |
|
32
|
|
|
if (preg_match('#410 Gone#i', $errorMessage)) { |
|
33
|
|
|
$this->log->notice('410 Gone'); |
|
34
|
|
|
|
|
35
|
|
|
return $this->formatLienBrise($url); |
|
36
|
|
|
} // 403 |
|
37
|
|
|
elseif (preg_match('#403 Forbidden#i', $errorMessage)) { |
|
38
|
|
|
$this->log403($url); |
|
39
|
|
|
|
|
40
|
|
|
return $url; |
|
41
|
|
|
} elseif (preg_match('#404 Not Found#i', $errorMessage)) { |
|
42
|
|
|
$this->log->notice('404 Not Found'); |
|
43
|
|
|
|
|
44
|
|
|
if (ExternRefTransformer::REPLACE_404) { |
|
45
|
|
|
return $this->formatLienBrise($url); |
|
46
|
|
|
} |
|
47
|
|
|
return $url; |
|
48
|
|
|
} elseif (preg_match('#401 Unauthorized#i', $errorMessage)) { |
|
49
|
|
|
$this->log->notice('401 Unauthorized : skip ' . $url); |
|
50
|
|
|
|
|
51
|
|
|
return $url; |
|
52
|
|
|
} else { |
|
53
|
|
|
// autre : ne pas générer de {lien brisé}, car peut-être 404 temporaire |
|
54
|
|
|
$this->log->warning('erreur sur extractWebData ' . $errorMessage); |
|
55
|
|
|
|
|
56
|
|
|
//file_put_contents(self::LOG_REQUEST_ERROR, $this->domain."\n", FILE_APPEND); |
|
57
|
|
|
|
|
58
|
|
|
return $url; |
|
59
|
|
|
} |
|
60
|
|
|
} |
|
61
|
|
|
|
|
62
|
|
|
protected function formatLienBrise(string $url): string |
|
63
|
|
|
{ |
|
64
|
|
|
return sprintf( |
|
65
|
|
|
'{{Lien brisé |url= %s |titre=%s |brisé le=%s}}', |
|
66
|
|
|
$url, |
|
67
|
|
|
$this->generateTitleFromURLText($url), |
|
68
|
|
|
date('d-m-Y') |
|
69
|
|
|
); |
|
70
|
|
|
} |
|
71
|
|
|
|
|
72
|
|
|
// todo move template |
|
73
|
|
|
|
|
74
|
|
|
/** |
|
75
|
|
|
* URL => "parismatch.com/People/bla…" |
|
76
|
|
|
*/ |
|
77
|
|
|
protected function generateTitleFromURLText(string $url): string |
|
78
|
|
|
{ |
|
79
|
|
|
$text = str_replace(['https://', 'http://', 'www.'], '', $url); |
|
80
|
|
|
if (strlen($text) > 30) { |
|
81
|
|
|
$text = substr($text, 0, 30) . '…'; |
|
82
|
|
|
} |
|
83
|
|
|
|
|
84
|
|
|
return $text; |
|
85
|
|
|
} |
|
86
|
|
|
|
|
87
|
|
|
protected function log403(string $url): void |
|
88
|
|
|
{ |
|
89
|
|
|
$this->log->warning('403 Forbidden : ' . $url); |
|
90
|
|
|
file_put_contents(self::LOG_REQUEST_ERROR, '403 Forbidden : ' . $url . "\n", FILE_APPEND); |
|
91
|
|
|
} |
|
92
|
|
|
} |
Dispositif /
Wikibot
