|
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 App\Domain\InfrastructurePorts\DeadlinkArchiverInterface; |
|
13
|
|
|
use App\Domain\InfrastructurePorts\InternetDomainParserInterface; |
|
14
|
|
|
use App\Domain\Models\Summary; |
|
15
|
|
|
use App\Domain\Models\WebarchiveDTO; |
|
16
|
|
|
use App\Domain\Publisher\ExternMapper; |
|
17
|
|
|
use App\Infrastructure\InternetDomainParser; |
|
18
|
|
|
use App\Infrastructure\Monitor\NullLogger; |
|
19
|
|
|
use App\Infrastructure\ServiceFactory; |
|
20
|
|
|
use DateTimeImmutable; |
|
21
|
|
|
use DateTimeInterface; |
|
22
|
|
|
use Psr\Log\LoggerInterface; |
|
23
|
|
|
|
|
24
|
|
|
/** |
|
25
|
|
|
* Transform dead link url in {lien brisé} or import web archive URL |
|
26
|
|
|
*/ |
|
27
|
|
|
class DeadLinkTransformer |
|
28
|
|
|
{ |
|
29
|
|
|
private const USE_TOR_FOR_ARCHIVE = false; |
|
30
|
|
|
private const DELAY_PARSE_ARCHIVE = 3; |
|
31
|
|
|
private const REPLACE_RAW_WIKIWIX_BY_LIENWEB = false; |
|
32
|
|
|
|
|
33
|
|
|
/** |
|
34
|
|
|
* @param DeadlinkArchiverInterface[] $archivers |
|
35
|
|
|
*/ |
|
36
|
|
|
public function __construct( |
|
37
|
|
|
protected array $archivers = [], |
|
38
|
|
|
protected ?InternetDomainParserInterface $domainParser = null, |
|
39
|
|
|
protected ?ExternRefTransformerInterface $externRefTransformer = null, |
|
40
|
|
|
protected LoggerInterface $log = new NullLogger() |
|
41
|
|
|
) |
|
42
|
|
|
{ |
|
43
|
|
|
} |
|
44
|
|
|
|
|
45
|
|
|
public function formatFromUrl(string $url, DateTimeInterface $now = new DateTimeImmutable()): string |
|
46
|
|
|
{ |
|
47
|
|
|
// choose randomly one archiver |
|
48
|
|
|
// TODO choose IA sur url IA, Wikiwix sur url Wikiwix, etc |
|
49
|
|
|
$oneArchiver = !empty($this->archivers) ? $this->archivers[array_rand($this->archivers)] : null; |
|
50
|
|
|
|
|
51
|
|
|
// HACK : Temporary skip transform on archiver URL (éviter archive IA sur url Wikiwix) |
|
52
|
|
|
if ($this->isWebArchiveUrl($url)) { |
|
53
|
|
|
$this->log->notice('Skip {lien brisé} on web archive url', ['stats' => 'externref.skip.lienBriseOnwebarchiveurl']); |
|
54
|
|
|
return $url; |
|
55
|
|
|
} |
|
56
|
|
|
|
|
57
|
|
|
if ($oneArchiver instanceof DeadlinkArchiverInterface) { |
|
58
|
|
|
$webarchiveDTO = $oneArchiver->searchWebarchive($url); |
|
59
|
|
|
if ($webarchiveDTO instanceof WebarchiveDTO) { |
|
60
|
|
|
if ($webarchiveDTO->getArchiver() === '[[Wikiwix]]') { |
|
61
|
|
|
$this->log->notice('🥝 Wikiwix found'); |
|
62
|
|
|
} |
|
63
|
|
|
if ($webarchiveDTO->getArchiver() === '[[Internet Archive]]') { |
|
64
|
|
|
$this->log->notice('🏛️ InternetArchive found'); |
|
65
|
|
|
} |
|
66
|
|
|
$this->log->debug('archive url: ' . $webarchiveDTO->getArchiveUrl()); |
|
67
|
|
|
|
|
68
|
|
|
return $this->generateLienWebFromArchive($webarchiveDTO); |
|
69
|
|
|
} |
|
70
|
|
|
$this->log->notice('web archive not found'); |
|
71
|
|
|
} |
|
72
|
|
|
|
|
73
|
|
|
return $this->generateLienBrise($url, $now); |
|
74
|
|
|
} |
|
75
|
|
|
|
|
76
|
|
|
private function generateLienWebFromArchive(WebarchiveDTO $dto): string |
|
77
|
|
|
{ |
|
78
|
|
|
sleep(self::DELAY_PARSE_ARCHIVE); |
|
79
|
|
|
|
|
80
|
|
|
$externRefProcessOnArchive = $this->externRefProcessOnArchive($dto); |
|
81
|
|
|
|
|
82
|
|
|
// Wikiwix : "Sorry, this system is overloaded. Please come back in a minute." |
|
83
|
|
|
// manage content-type 'application/pdf' which is not parsed by ExternRefTransformer |
|
84
|
|
|
if ( |
|
85
|
|
|
self::REPLACE_RAW_WIKIWIX_BY_LIENWEB |
|
86
|
|
|
&& str_starts_with($externRefProcessOnArchive, 'https://archive.wikiwix.com/cache/') |
|
87
|
|
|
) { |
|
88
|
|
|
$this->log->notice('Replace raw wikiwix by lien web'); |
|
89
|
|
|
|
|
90
|
|
|
return sprintf( |
|
91
|
|
|
'{{Lien web |url= %s |titre=%s |site= %s |consulté le=%s |archive-date=%s}}', |
|
92
|
|
|
$dto->getArchiveUrl(), |
|
93
|
|
|
'Archive ' . $this->generateTitleFromURLText($dto->getOriginalUrl()) . '<!-- titre à compléter -->', |
|
94
|
|
|
'via ' . $dto->getArchiver(), |
|
95
|
|
|
date('d-m-Y'), |
|
96
|
|
|
$dto->getArchiveDate() instanceof DateTimeInterface ? $dto->getArchiveDate()->format('d-m-Y') : '' |
|
97
|
|
|
); |
|
98
|
|
|
} |
|
99
|
|
|
|
|
100
|
|
|
return $externRefProcessOnArchive; |
|
101
|
|
|
} |
|
102
|
|
|
|
|
103
|
|
|
/** |
|
104
|
|
|
* To extract the title+author+lang+… from the webarchive page. |
|
105
|
|
|
*/ |
|
106
|
|
|
private function externRefProcessOnArchive(WebarchiveDTO $dto): string |
|
107
|
|
|
{ |
|
108
|
|
|
$summary = new Summary('test'); |
|
109
|
|
|
if (!$this->externRefTransformer instanceof ExternRefTransformerInterface) { |
|
110
|
|
|
$this->externRefTransformer = new ExternRefTransformer( |
|
111
|
|
|
new ExternMapper($this->log), |
|
112
|
|
|
ServiceFactory::getHttpClient(self::USE_TOR_FOR_ARCHIVE), |
|
113
|
|
|
new InternetDomainParser(), |
|
114
|
|
|
$this->log, |
|
115
|
|
|
); // todo inverse dependency |
|
116
|
|
|
} |
|
117
|
|
|
|
|
118
|
|
|
$options = $this->domainParser instanceof InternetDomainParserInterface |
|
119
|
|
|
? ['originalRegistrableDomain' => $this->domainParser->getRegistrableDomainFromURL($dto->getOriginalUrl())] |
|
120
|
|
|
: []; |
|
121
|
|
|
|
|
122
|
|
|
return $this->externRefTransformer->process($dto->getArchiveUrl(), $summary, $options); |
|
|
|
|
|
|
123
|
|
|
} |
|
124
|
|
|
|
|
125
|
|
|
protected function generateTitleFromURLText(string $url): string |
|
126
|
|
|
{ |
|
127
|
|
|
$text = str_replace(['https://', 'http://', 'www.'], '', $url); |
|
128
|
|
|
if (strlen($text) > 30) { |
|
129
|
|
|
$text = substr($text, 0, 30) . '…'; |
|
130
|
|
|
} |
|
131
|
|
|
|
|
132
|
|
|
return $text; |
|
133
|
|
|
} |
|
134
|
|
|
|
|
135
|
|
|
protected function generateLienBrise(string $url, DateTimeInterface $now): string |
|
136
|
|
|
{ |
|
137
|
|
|
if ($this->isWebArchiveUrl($url)) { |
|
138
|
|
|
$this->notice('Skip {lien brisé} on web archive url', ['stats' => 'externref.skip.lienBriseOnwebarchiveurl']); |
|
|
|
|
|
|
139
|
|
|
|
|
140
|
|
|
return $url; |
|
141
|
|
|
} |
|
142
|
|
|
|
|
143
|
|
|
return sprintf( |
|
144
|
|
|
'{{Lien brisé |url= %s |titre=%s |brisé le=%s}}', |
|
145
|
|
|
$this->stripWebArchivePrefix($url), |
|
146
|
|
|
$this->generateTitleFromURLText($url), |
|
147
|
|
|
$now->format('d-m-Y') |
|
148
|
|
|
); |
|
149
|
|
|
} |
|
150
|
|
|
|
|
151
|
|
|
/** |
|
152
|
|
|
* Bug https://w.wiki/7kUm |
|
153
|
|
|
*/ |
|
154
|
|
|
private function stripWebArchivePrefix(string $url): string |
|
155
|
|
|
{ |
|
156
|
|
|
$url = preg_replace('#^https?://web\.archive\.org/web/\d+/#', '', $url); |
|
157
|
|
|
$url = preg_replace('#^https?://archive\.is/\d+/#', '', $url); |
|
158
|
|
|
|
|
159
|
|
|
return preg_replace('#^https?://archive\.wikiwix\.com/cache/\d+/#', '', $url); |
|
160
|
|
|
} |
|
161
|
|
|
|
|
162
|
|
|
/** |
|
163
|
|
|
* todo move |
|
164
|
|
|
*/ |
|
165
|
|
|
protected function isWebArchiveUrl(string $url): bool |
|
166
|
|
|
{ |
|
167
|
|
|
return str_starts_with($url, 'http://web.archive.org/web/') |
|
168
|
|
|
|| str_starts_with($url, 'https://web.archive.org/web/') |
|
169
|
|
|
|| str_starts_with($url, 'https://archive.is/') |
|
170
|
|
|
|| str_starts_with($url, 'https://archive.wikiwix.com/cache/'); |
|
171
|
|
|
} |
|
172
|
|
|
} |
Dispositif /
Wikibot
Loading history...
This check looks for calls to methods that do not seem to exist on a given type. It looks for the method on the type itself as well as in inherited classes or implemented interfaces.
This is most likely a typographical error or the method has been renamed.