|
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\Validators; |
|
11
|
|
|
|
|
12
|
|
|
use App\Domain\ValidatorInterface; |
|
13
|
|
|
use Psr\Log\LoggerInterface; |
|
14
|
|
|
use Psr\Log\NullLogger; |
|
15
|
|
|
|
|
16
|
|
|
/** |
|
17
|
|
|
* Detect if robots noindex. |
|
18
|
|
|
* https://developers.google.com/search/docs/crawling-indexing/robots-meta-tag?hl=fr |
|
19
|
|
|
*/ |
|
20
|
|
|
class RobotNoIndexValidator implements ValidatorInterface |
|
21
|
|
|
{ |
|
22
|
|
|
public $noindexWhitelist = ['test.com']; // move to config |
|
23
|
|
|
|
|
24
|
|
|
public function __construct( |
|
25
|
|
|
private readonly array $pageData, |
|
26
|
|
|
private readonly string $url, |
|
27
|
|
|
private readonly LoggerInterface $log = new NullLogger() |
|
28
|
|
|
) |
|
29
|
|
|
{ |
|
30
|
|
|
} |
|
31
|
|
|
|
|
32
|
|
|
// "NOINDEX" => true |
|
33
|
|
|
public function validate(): bool |
|
34
|
|
|
{ |
|
35
|
|
|
$robots = $this->pageData['meta']['robots'] ?? null; |
|
36
|
|
|
if ( |
|
37
|
|
|
!empty($robots) |
|
38
|
|
|
&& ( |
|
39
|
|
|
stripos((string)$robots, 'noindex') !== false |
|
40
|
|
|
|| stripos((string)$robots, 'none') !== false |
|
41
|
|
|
) |
|
42
|
|
|
) { |
|
43
|
|
|
$this->log->notice('robots NOINDEX : ' . $this->url); |
|
44
|
|
|
|
|
45
|
|
|
if (empty($this->pageData['meta']['prettyDomainName'])) { |
|
46
|
|
|
$this->log->warning('No prettyDomainName for ' . $this->url); |
|
47
|
|
|
|
|
48
|
|
|
return true; |
|
49
|
|
|
} |
|
50
|
|
|
|
|
51
|
|
|
return !$this->isNoIndexDomainWhitelisted($this->pageData['meta']['prettyDomainName']); |
|
52
|
|
|
} |
|
53
|
|
|
|
|
54
|
|
|
return false; |
|
55
|
|
|
} |
|
56
|
|
|
|
|
57
|
|
|
protected function isNoIndexDomainWhitelisted(?string $prettyDomain): bool |
|
58
|
|
|
{ |
|
59
|
|
|
if (in_array($prettyDomain ?? '', $this->noindexWhitelist)) { |
|
60
|
|
|
$this->log->notice('ROBOT_NOINDEX_WHITELIST ' . $prettyDomain); |
|
61
|
|
|
|
|
62
|
|
|
return true; |
|
63
|
|
|
} |
|
64
|
|
|
|
|
65
|
|
|
return false; |
|
66
|
|
|
} |
|
67
|
|
|
} |
Dispositif /
Wikibot
