|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
/* |
|
4
|
|
|
* This file is part of dispositif/wikibot application (@github) |
|
5
|
|
|
* 2019-2023 © Philippe M./Irønie <[email protected]> |
|
6
|
|
|
* For the full copyright and MIT license information, view the license file. |
|
7
|
|
|
*/ |
|
8
|
|
|
declare(strict_types=1); |
|
9
|
|
|
|
|
10
|
|
|
namespace App\Application\ExternLink; |
|
11
|
|
|
|
|
12
|
|
|
use App\Application\AbstractRefBotWorker; |
|
13
|
|
|
use App\Application\Http\ExternHttpClient; |
|
14
|
|
|
use App\Domain\Exceptions\ConfigException; |
|
15
|
|
|
use App\Domain\ExternLink\ExternRefTransformer; |
|
16
|
|
|
use App\Domain\InfrastructurePorts\InternetDomainParserInterface; |
|
17
|
|
|
use App\Domain\Publisher\ExternMapper; |
|
18
|
|
|
use Throwable; |
|
19
|
|
|
|
|
20
|
|
|
/** |
|
21
|
|
|
* Class Ref2ArticleWorker |
|
22
|
|
|
* @package App\Application\Examples |
|
23
|
|
|
*/ |
|
24
|
|
|
class ExternRefWorker extends AbstractRefBotWorker |
|
25
|
|
|
{ |
|
26
|
|
|
public const TASK_BOT_FLAG = true; |
|
27
|
|
|
public const MAX_REFS_PROCESSED_IN_ARTICLE = 30; |
|
28
|
|
|
public const SLEEP_AFTER_EDITION = 15; // sec |
|
29
|
|
|
public const MINUTES_DELAY_AFTER_LAST_HUMAN_EDIT = 10; // minutes |
|
30
|
|
|
public const CHECK_EDIT_CONFLICT = true; |
|
31
|
|
|
public const ARTICLE_ANALYZED_FILENAME = __DIR__ . '/../resources/article_externRef_edited.txt'; |
|
32
|
|
|
public const SKIP_ADQ = false; |
|
33
|
|
|
public const SKIP_LASTEDIT_BY_BOT = false; |
|
34
|
|
|
public const CITATION_NUMBER_ON_FIRE = 15; |
|
35
|
|
|
public const CITATION_NUMBER_NO_BOTFLAG = 20; |
|
36
|
|
|
public const DEAD_LINK_NO_BOTFLAG = 5; |
|
37
|
|
|
public const SKIP_SITE_BLACKLISTED = true; |
|
38
|
|
|
public const SKIP_ROBOT_NOINDEX = true; |
|
39
|
|
|
|
|
40
|
|
|
protected $modeAuto = true; |
|
41
|
|
|
/** |
|
42
|
|
|
* @var ExternRefTransformer |
|
43
|
|
|
*/ |
|
44
|
|
|
protected $transformer; |
|
45
|
|
|
|
|
46
|
|
|
/** |
|
47
|
|
|
* Traite contenu d'une <ref> ou bien lien externe (précédé d'une puce). |
|
48
|
|
|
* |
|
49
|
|
|
* @param $refContent |
|
50
|
|
|
* |
|
51
|
|
|
* @return string |
|
52
|
|
|
*/ |
|
53
|
|
|
public function processRefContent($refContent): string |
|
54
|
|
|
{ |
|
55
|
|
|
// todo // hack Temporary Skip URL |
|
56
|
|
|
if (preg_match('#books\.google#', $refContent)) { |
|
57
|
|
|
return $refContent; |
|
58
|
|
|
} |
|
59
|
|
|
|
|
60
|
|
|
try { |
|
61
|
|
|
$result = $this->transformer->process($refContent, $this->summary); |
|
62
|
|
|
} catch (Throwable $e) { |
|
63
|
|
|
echo "** Problème détecté 234242\n"; |
|
64
|
|
|
$this->log->critical($e->getMessage() . " " . $e->getFile() . ":" . $e->getLine()); |
|
65
|
|
|
// TODO : parse $e->message -> variable process, taskName, botflag... |
|
66
|
|
|
|
|
67
|
|
|
return $refContent; |
|
68
|
|
|
} |
|
69
|
|
|
|
|
70
|
|
|
if (trim($result) === trim($refContent)) { |
|
71
|
|
|
return $refContent; |
|
72
|
|
|
} |
|
73
|
|
|
|
|
74
|
|
|
// Gestion semi-auto : todo CONDITION POURRI FAUSSE $this->transformer->skipUnauthorised |
|
75
|
|
|
|
|
76
|
|
|
$this->printDiff($refContent, $result, 'echo'); |
|
77
|
|
|
if (!$this->autoOrYesConfirmation('Conserver cette modif ?')) { |
|
78
|
|
|
return $refContent; |
|
79
|
|
|
} |
|
80
|
|
|
|
|
81
|
|
|
if (preg_match('#{{lien brisé#i', $result)) { |
|
82
|
|
|
$this->summary->memo['count lien brisé'] = 1 + ($this->summary->memo['count lien brisé'] ?? 0); |
|
83
|
|
|
if ($this->summary->memo['count lien brisé'] >= self::DEAD_LINK_NO_BOTFLAG) { |
|
84
|
|
|
$this->summary->setBotFlag(false); |
|
85
|
|
|
} |
|
86
|
|
|
} |
|
87
|
|
|
if ($this->summary->citationNumber >= self::CITATION_NUMBER_NO_BOTFLAG) { |
|
88
|
|
|
$this->summary->setBotFlag(false); |
|
89
|
|
|
} |
|
90
|
|
|
|
|
91
|
|
|
$this->summary->memo['count URL'] = 1 + ($this->summary->memo['count URL'] ?? 0); |
|
92
|
|
|
|
|
93
|
|
|
return $result; |
|
94
|
|
|
} |
|
95
|
|
|
|
|
96
|
|
|
protected function setUpInConstructor(): void |
|
97
|
|
|
{ |
|
98
|
|
|
if (!$this->domainParser instanceof InternetDomainParserInterface) { |
|
99
|
|
|
dump($this->domainParser); |
|
100
|
|
|
throw new ConfigException('DomainParser not set'); |
|
101
|
|
|
} |
|
102
|
|
|
// TODO extract ExternRefTransformerInterface |
|
103
|
|
|
$this->transformer = new ExternRefTransformer( |
|
104
|
|
|
new ExternMapper($this->log), |
|
105
|
|
|
new ExternHttpClient($this->log), |
|
106
|
|
|
$this->domainParser, |
|
107
|
|
|
$this->log |
|
108
|
|
|
); |
|
109
|
|
|
$this->transformer->skipSiteBlacklisted = self::SKIP_SITE_BLACKLISTED; |
|
110
|
|
|
$this->transformer->skipRobotNoIndex = self::SKIP_ROBOT_NOINDEX; |
|
111
|
|
|
//todo? move in __constructor + parent::__constructor() |
|
112
|
|
|
} |
|
113
|
|
|
|
|
114
|
|
|
/** |
|
115
|
|
|
* todo move to a Summary child ? |
|
116
|
|
|
* Rewriting default Summary::serialize() |
|
117
|
|
|
* @return string |
|
118
|
|
|
*/ |
|
119
|
|
|
protected function generateSummaryText(): string |
|
120
|
|
|
{ |
|
121
|
|
|
$prefixSummary = ($this->summary->isBotFlag()) ? 'bot ' : ''; |
|
122
|
|
|
$suffix = ''; |
|
123
|
|
|
if (isset($this->summary->memo['count article'])) { |
|
124
|
|
|
$suffix .= ' ' . $this->summary->memo['count article'] . 'x {article}'; |
|
125
|
|
|
} |
|
126
|
|
|
if (isset($this->summary->memo['count lien web'])) { |
|
127
|
|
|
$suffix .= ' ' . $this->summary->memo['count lien web'] . 'x {lien web}'; |
|
128
|
|
|
} |
|
129
|
|
|
if (isset($this->summary->memo['presse'])) { |
|
130
|
|
|
$suffix .= ' 🗞️'; // 🗞️ 📰 |
|
131
|
|
|
} |
|
132
|
|
|
if (isset($this->summary->memo['science'])) { |
|
133
|
|
|
$suffix .= ' 🧪'; // 🧪 🔬 |
|
134
|
|
|
} |
|
135
|
|
|
if (isset($this->summary->memo['count lien brisé'])) { |
|
136
|
|
|
$suffix .= ', ⚠️️️lien brisé'; //⚠️💩 |
|
137
|
|
|
$suffix .= ($this->summary->memo['count lien brisé'] > 1) ? ' x' . $this->summary->memo['count lien brisé'] : |
|
138
|
|
|
''; |
|
139
|
|
|
} |
|
140
|
|
|
if (isset($this->summary->memo['accès url non libre'])) { |
|
141
|
|
|
$suffix .= ' 🔒'; |
|
142
|
|
|
} |
|
143
|
|
|
|
|
144
|
|
|
if ($this->summary->citationNumber >= self::CITATION_NUMBER_ON_FIRE) { |
|
145
|
|
|
$suffix .= ' 🔥'; |
|
146
|
|
|
} |
|
147
|
|
|
|
|
148
|
|
|
return $prefixSummary . $this->summary->taskName . $suffix; |
|
149
|
|
|
} |
|
150
|
|
|
|
|
151
|
|
|
} |
|
152
|
|
|
|
Dispositif /
Wikibot
