|
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\Utils\DateUtil; |
|
14
|
|
|
use App\Infrastructure\Monitor\NullLogger; |
|
15
|
|
|
use Psr\Log\LoggerInterface; |
|
16
|
|
|
|
|
17
|
|
|
/** |
|
18
|
|
|
* Find the last addition edit (and date) of a string in a wiki article. |
|
19
|
|
|
* Request on http://wikipedia.ramselehof.de/wikiblame.php |
|
20
|
|
|
* See also https://github.com/FlominatorTM/wikiblame |
|
21
|
|
|
*/ |
|
22
|
|
|
class WikiBlameAdapter |
|
23
|
|
|
{ |
|
24
|
|
|
protected HttpClientInterface $client; |
|
25
|
|
|
|
|
26
|
|
|
public function __construct(private readonly LoggerInterface|NullLogger $log = new NullLogger()) |
|
27
|
|
|
{ |
|
28
|
|
|
$this->client = ServiceFactory::getHttpClient(); |
|
29
|
|
|
} |
|
30
|
|
|
|
|
31
|
|
|
/** |
|
32
|
|
|
* return [ |
|
33
|
|
|
* "versionid" => (int) 103659068 |
|
34
|
|
|
* "dateraw" => "09 mai 2014" |
|
35
|
|
|
* "datetime" => DateTime |
|
36
|
|
|
*/ |
|
37
|
|
|
public function searchDiff(string $article, string $string, bool $hasWikicode = false): ?array |
|
38
|
|
|
{ |
|
39
|
|
|
$url = 'http://wikipedia.ramselehof.de/wikiblame.php?project=wikipedia&article=' |
|
40
|
|
|
. str_replace(' ', '+', $article) |
|
41
|
|
|
. '&needle=' . urlencode($string) |
|
42
|
|
|
. '&lang=fr&limit=2000&offjahr=2024&offmon=1&offtag=1&offhour=23&offmin=55' |
|
43
|
|
|
. '&searchmethod=int&order=desc' |
|
44
|
|
|
. '&force_wikitags=' . ($hasWikicode ? 'on' : 'off') |
|
45
|
|
|
. '&user_lang=fr&ignorefirst=0'; |
|
46
|
|
|
|
|
47
|
|
|
$response = $this->client->get($url); |
|
48
|
|
|
|
|
49
|
|
|
if ($response->getStatusCode() !== 200) { |
|
50
|
|
|
$this->log->warning('WikiBlame: ' . $response->getStatusCode() . ' ' . $response->getReasonPhrase()); |
|
51
|
|
|
return null; |
|
52
|
|
|
} |
|
53
|
|
|
|
|
54
|
|
|
$rawHtml = $response->getBody()->getContents(); |
|
55
|
|
|
if (empty($rawHtml)) { |
|
56
|
|
|
$this->log->warning('WikiBlame: empty response'); |
|
57
|
|
|
return null; |
|
58
|
|
|
} |
|
59
|
|
|
|
|
60
|
|
|
return $this->parseRawHtml($rawHtml); |
|
61
|
|
|
} |
|
62
|
|
|
|
|
63
|
|
|
private function parseRawHtml(string $rawHtml): ?array |
|
64
|
|
|
{ |
|
65
|
|
|
if (!preg_match('#Insertion détectée entre le <a href="https:[^\"]+">[^<]+<\/a> et le <a href="https:[^\"]+&oldid=(\d+)">[\d:]+, ([^<]+)<\/a>: <a#', $rawHtml, $matches)) { |
|
66
|
|
|
$this->log->warning('WikiBlame: no result from HTML'); |
|
67
|
|
|
return null; |
|
68
|
|
|
} |
|
69
|
|
|
|
|
70
|
|
|
return [ |
|
71
|
|
|
'versionid' => (int)$matches[1], |
|
72
|
|
|
'dateraw' => $matches[2], |
|
73
|
|
|
'datetime' => DateUtil::simpleFrench2object($matches[2]), |
|
74
|
|
|
]; |
|
75
|
|
|
} |
|
76
|
|
|
} |
Dispositif /
Wikibot
