|
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\Domain\InfrastructurePorts\DeadlinkArchiverInterface; |
|
13
|
|
|
use App\Domain\InfrastructurePorts\ExternHttpClientInterface; |
|
14
|
|
|
use App\Domain\Models\WebarchiveDTO; |
|
15
|
|
|
use DateTimeImmutable; |
|
16
|
|
|
use Psr\Http\Message\ResponseInterface; |
|
17
|
|
|
use Psr\Log\LoggerInterface; |
|
18
|
|
|
use Psr\Log\NullLogger; |
|
19
|
|
|
|
|
20
|
|
|
/** |
|
21
|
|
|
* Get archive url from Wikiwix webarchiver API. |
|
22
|
|
|
* |
|
23
|
|
|
* JSON response example: |
|
24
|
|
|
* "status": 200, |
|
25
|
|
|
* "contenttype": "text\/html; charset=utf-8", |
|
26
|
|
|
* "timestamp": 912380400, |
|
27
|
|
|
* "datetime": "19981130000000", |
|
28
|
|
|
* "longformurl": "https:\/\/archive.wikiwix.com\/cache\/19981130000000\/http:\/\/casamaures.org\/keskispas.php?lng=fr&pg=1064" |
|
29
|
|
|
*/ |
|
30
|
|
|
class WikiwixAdapter implements DeadlinkArchiverInterface |
|
31
|
|
|
{ |
|
32
|
|
|
public const ARCHIVER_NAME = '[[Wikiwix]]'; |
|
33
|
|
|
private const API_URL = 'https://archive.wikiwix.com/cache/index2.php?apiresponse=1&url='; |
|
34
|
|
|
|
|
35
|
|
|
public function __construct( |
|
36
|
|
|
protected readonly ExternHttpClientInterface $externHttpClient, |
|
37
|
|
|
protected readonly LoggerInterface $log = new NullLogger() |
|
38
|
|
|
) |
|
39
|
|
|
{ |
|
40
|
|
|
} |
|
41
|
|
|
|
|
42
|
|
|
public function searchWebarchive(string $url): ?WebarchiveDTO |
|
43
|
|
|
{ |
|
44
|
|
|
$archiveData = $this->requestWikiwixApi($url); |
|
45
|
|
|
if (empty($archiveData['longformurl'])) { |
|
46
|
|
|
return null; |
|
47
|
|
|
} |
|
48
|
|
|
|
|
49
|
|
|
return new WebarchiveDTO( |
|
50
|
|
|
self::ARCHIVER_NAME, |
|
51
|
|
|
$url, |
|
52
|
|
|
(string) $archiveData['longformurl'], |
|
53
|
|
|
$archiveData['timestamp'] |
|
54
|
|
|
? DateTimeImmutable::createFromFormat('U', (string)$archiveData['timestamp']) |
|
55
|
|
|
: null |
|
56
|
|
|
); // todo factory ? |
|
57
|
|
|
} |
|
58
|
|
|
|
|
59
|
|
|
protected function requestWikiwixApi(string $url): array |
|
60
|
|
|
{ |
|
61
|
|
|
$response = $this->externHttpClient->getClient()->request( |
|
|
|
|
|
|
62
|
|
|
'GET', |
|
63
|
|
|
self::API_URL . urlencode($url) |
|
64
|
|
|
); |
|
65
|
|
|
|
|
66
|
|
|
if (!$response instanceof ResponseInterface || $response->getStatusCode() !== 200) { |
|
67
|
|
|
return []; |
|
68
|
|
|
} |
|
69
|
|
|
$jsonString = $response->getBody()->getContents(); |
|
70
|
|
|
$data = json_decode($jsonString, true) ?? []; |
|
71
|
|
|
|
|
72
|
|
|
// check wikiwix archive status |
|
73
|
|
|
if (empty($data['status']) || (int) $data['status'] !== 200) { |
|
74
|
|
|
$this->log->debug('WikiwixAdapter response: ' . $jsonString); |
|
75
|
|
|
|
|
76
|
|
|
return []; |
|
77
|
|
|
} |
|
78
|
|
|
|
|
79
|
|
|
return $data; |
|
80
|
|
|
} |
|
81
|
|
|
} |
Dispositif /
Wikibot
Loading history...