|
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\InfrastructurePorts\HttpClientInterface; |
|
13
|
|
|
use App\Domain\InfrastructurePorts\DeadlinkArchiverInterface; |
|
14
|
|
|
use App\Domain\Models\WebarchiveDTO; |
|
15
|
|
|
use DateTime; |
|
16
|
|
|
use DateTimeInterface; |
|
17
|
|
|
use Psr\Log\LoggerInterface; |
|
18
|
|
|
use Psr\Log\NullLogger; |
|
19
|
|
|
|
|
20
|
|
|
/** |
|
21
|
|
|
* https://archive.org/help/wayback_api.php |
|
22
|
|
|
* todo closest by date |
|
23
|
|
|
*/ |
|
24
|
|
|
class InternetArchiveAdapter implements DeadlinkArchiverInterface |
|
25
|
|
|
{ |
|
26
|
|
|
final public const ARCHIVER_NAME = '[[Internet Archive]]'; // [[Wayback Machine]] ? |
|
27
|
|
|
private const SEARCH_CLOSEST_TIMESTAMP = '20220101'; |
|
28
|
|
|
|
|
29
|
|
|
public function __construct( |
|
30
|
|
|
protected readonly HttpClientInterface $client, |
|
31
|
|
|
protected readonly LoggerInterface $log = new NullLogger() |
|
32
|
|
|
) |
|
33
|
|
|
{ |
|
34
|
|
|
} |
|
35
|
|
|
|
|
36
|
|
|
public function searchWebarchive(string $url, ?DateTimeInterface $date = null): ?WebarchiveDTO |
|
37
|
|
|
{ |
|
38
|
|
|
$archiveData = $this->requestInternetArchiveApi($url, $date); |
|
39
|
|
|
if (empty($archiveData)) { |
|
40
|
|
|
return null; |
|
41
|
|
|
} |
|
42
|
|
|
|
|
43
|
|
|
$iaDateOrNull = $this->convertIATimestampToDateTime($archiveData['timestamp'] ?? null); |
|
44
|
|
|
|
|
45
|
|
|
return new WebarchiveDTO( |
|
46
|
|
|
self::ARCHIVER_NAME, |
|
47
|
|
|
$url, |
|
48
|
|
|
(string)$archiveData['url'], |
|
49
|
|
|
$iaDateOrNull |
|
50
|
|
|
); |
|
51
|
|
|
} |
|
52
|
|
|
|
|
53
|
|
|
protected function requestInternetArchiveApi(string $url, ?DateTimeInterface $date = null): array |
|
54
|
|
|
{ |
|
55
|
|
|
$response = $this->client->get( |
|
56
|
|
|
'https://archive.org/wayback/available?timestamp=' . self::SEARCH_CLOSEST_TIMESTAMP . '&url=' . urlencode($url), |
|
57
|
|
|
[ |
|
58
|
|
|
'timeout' => 20, |
|
59
|
|
|
'allow_redirects' => true, |
|
60
|
|
|
'headers' => ['User-Agent' => getenv('USER_AGENT')], |
|
61
|
|
|
'http_errors' => false, // no Exception on 4xx 5xx |
|
62
|
|
|
'verify' => false, |
|
63
|
|
|
] |
|
64
|
|
|
); |
|
65
|
|
|
|
|
66
|
|
|
if ($response->getStatusCode() !== 200) { |
|
67
|
|
|
$this->log->debug('InternetArchive: incorrect response', [ |
|
68
|
|
|
'status' => $response->getStatusCode(), |
|
69
|
|
|
'content-type' => $response->getHeader('Content-Type'), |
|
70
|
|
|
]); |
|
71
|
|
|
return []; |
|
72
|
|
|
} |
|
73
|
|
|
$jsonString = $response->getBody()->getContents(); |
|
74
|
|
|
$data = json_decode($jsonString, true, 512, JSON_THROW_ON_ERROR) ?? []; |
|
75
|
|
|
|
|
76
|
|
|
if (!isset($data['archived_snapshots']['closest'])) { |
|
77
|
|
|
$this->log->info('InternetArchive: no closest snapshot', [ |
|
78
|
|
|
'url' => $url, |
|
79
|
|
|
'date' => $date, |
|
80
|
|
|
'json' => $jsonString, |
|
81
|
|
|
]); |
|
82
|
|
|
|
|
83
|
|
|
return []; |
|
84
|
|
|
} |
|
85
|
|
|
|
|
86
|
|
|
// validate snapshot data |
|
87
|
|
|
$closest = $data['archived_snapshots']['closest']; |
|
88
|
|
|
if ($closest['status'] !== "200" || $closest['available'] !== true || empty($closest['url'])) { |
|
89
|
|
|
$this->log->debug('InternetArchive: snapshot invalid', $closest); |
|
90
|
|
|
return []; |
|
91
|
|
|
} |
|
92
|
|
|
|
|
93
|
|
|
return $closest; |
|
94
|
|
|
} |
|
95
|
|
|
|
|
96
|
|
|
/** |
|
97
|
|
|
* "YYYYMMDDhhmmss" |
|
98
|
|
|
*/ |
|
99
|
|
|
protected function convertIATimestampToDateTime(?string $iaTimestamp): ?DateTimeInterface |
|
100
|
|
|
{ |
|
101
|
|
|
if (empty($iaTimestamp)) { |
|
102
|
|
|
return null; |
|
103
|
|
|
} |
|
104
|
|
|
$iaDateTime = new DateTime(); |
|
105
|
|
|
$iaDateTime->setDate( |
|
106
|
|
|
(int)substr($iaTimestamp, 0, 4), |
|
107
|
|
|
(int)substr($iaTimestamp, 4, 2), |
|
108
|
|
|
(int)substr($iaTimestamp, 6, 2) |
|
109
|
|
|
); |
|
110
|
|
|
$iaDateTime->setTime( |
|
111
|
|
|
(int)substr($iaTimestamp, 8, 2), |
|
112
|
|
|
(int)substr($iaTimestamp, 10, 2), |
|
113
|
|
|
(int)substr($iaTimestamp, 12, 2) |
|
114
|
|
|
); |
|
115
|
|
|
|
|
116
|
|
|
return $iaDateTime; |
|
117
|
|
|
} |
|
118
|
|
|
} |
Dispositif /
Wikibot
