|
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\Application\OuvrageScan; |
|
11
|
|
|
|
|
12
|
|
|
use App\Application\InfrastructurePorts\DbAdapterInterface; |
|
13
|
|
|
use App\Application\InfrastructurePorts\PageListForAppInterface as PageListInterface; |
|
14
|
|
|
use App\Application\WikiBotConfig; |
|
15
|
|
|
use App\Application\WikiPageAction; |
|
16
|
|
|
use App\Domain\Utils\TemplateParser; |
|
17
|
|
|
use Exception; |
|
18
|
|
|
use Mediawiki\Api\MediawikiFactory; |
|
19
|
|
|
use Psr\Log\LoggerInterface; |
|
20
|
|
|
|
|
21
|
|
|
|
|
22
|
|
|
/** |
|
23
|
|
|
* From a titles list, scan the wiki and add the {ouvrage} citations into the database. |
|
24
|
|
|
*/ |
|
25
|
|
|
class ScanWiki2DB |
|
26
|
|
|
{ |
|
27
|
|
|
protected LoggerInterface $logger; |
|
28
|
|
|
|
|
29
|
|
|
public function __construct( |
|
30
|
|
|
private readonly MediawikiFactory $wiki, |
|
31
|
|
|
private readonly DbAdapterInterface $db, |
|
32
|
|
|
private readonly WikiBotConfig $bot, |
|
33
|
|
|
private readonly PageListInterface $pageList, |
|
34
|
|
|
private readonly int $priority = 0 |
|
35
|
|
|
) |
|
36
|
|
|
{ |
|
37
|
|
|
$this->logger = $this->bot->getLogger(); |
|
38
|
|
|
$this->process(); |
|
39
|
|
|
} |
|
40
|
|
|
|
|
41
|
|
|
/** |
|
42
|
|
|
* @throws Exception |
|
43
|
|
|
*/ |
|
44
|
|
|
public function process(): void |
|
45
|
|
|
{ |
|
46
|
|
|
$titles = $this->pageList->getPageTitles(); |
|
47
|
|
|
if ($titles === []) { |
|
48
|
|
|
$this->logger->info("pageList vide."); |
|
49
|
|
|
|
|
50
|
|
|
return; |
|
51
|
|
|
} |
|
52
|
|
|
foreach ($titles as $title) { |
|
53
|
|
|
$this->pageScan($title); |
|
54
|
|
|
sleep(4); |
|
55
|
|
|
} |
|
56
|
|
|
} |
|
57
|
|
|
|
|
58
|
|
|
/** |
|
59
|
|
|
* @throws Exception |
|
60
|
|
|
*/ |
|
61
|
|
|
public function pageScan(string $title): bool |
|
62
|
|
|
{ |
|
63
|
|
|
sleep(2); |
|
64
|
|
|
$this->logger->notice(sprintf('%s - %s', date("Y-m-d H:i:s"), $title)); |
|
65
|
|
|
|
|
66
|
|
|
$page = new WikiPageAction($this->wiki, $title); // todo injection |
|
67
|
|
|
$ns = $page->getNs(); |
|
68
|
|
|
if ($ns !== 0) { |
|
69
|
|
|
$this->logger->debug("SKIP : namespace $ns"); |
|
70
|
|
|
|
|
71
|
|
|
return false; |
|
72
|
|
|
} |
|
73
|
|
|
$text = $page->getText(); |
|
74
|
|
|
if (empty($text)) { |
|
75
|
|
|
$this->logger->debug("SKIP : empty text"); |
|
76
|
|
|
|
|
77
|
|
|
return false; |
|
78
|
|
|
} |
|
79
|
|
|
|
|
80
|
|
|
try { |
|
81
|
|
|
$parsedTemplates = TemplateParser::parseAllTemplateByName('ouvrage', $text); |
|
82
|
|
|
} catch (Exception $e) { |
|
83
|
|
|
$this->logger->error("SKIP : parse error " . $e->getMessage()); |
|
84
|
|
|
|
|
85
|
|
|
return false; |
|
86
|
|
|
} |
|
87
|
|
|
|
|
88
|
|
|
if (empty($parsedTemplates)) { |
|
89
|
|
|
return false; |
|
90
|
|
|
} |
|
91
|
|
|
|
|
92
|
|
|
$result = $this->insertDB($parsedTemplates['ouvrage'], $title); |
|
93
|
|
|
|
|
94
|
|
|
return !empty($result); |
|
95
|
|
|
} |
|
96
|
|
|
|
|
97
|
|
|
protected function insertDB(array $ouvrages, string $title): bool|array |
|
98
|
|
|
{ |
|
99
|
|
|
$data = []; |
|
100
|
|
|
foreach ($ouvrages as $res) { |
|
101
|
|
|
$oneData = [ |
|
102
|
|
|
'page' => $title, |
|
103
|
|
|
'raw' => $res['raw'], |
|
104
|
|
|
'priority' => $this->priority, |
|
105
|
|
|
]; |
|
106
|
|
|
|
|
107
|
|
|
if ((strlen($title) > 250) || empty($oneData['raw']) || strlen($oneData['raw']) > 2500) { |
|
108
|
|
|
$this->logger->warning("Skip : string to long : ", $oneData); |
|
109
|
|
|
continue; |
|
110
|
|
|
} |
|
111
|
|
|
|
|
112
|
|
|
// filter duplicates |
|
113
|
|
|
if (!in_array($oneData, $data)) { |
|
114
|
|
|
$data[] = $oneData; |
|
115
|
|
|
} |
|
116
|
|
|
} |
|
117
|
|
|
|
|
118
|
|
|
if (empty($data)) { |
|
119
|
|
|
$this->logger->notice("Skip : empty data"); |
|
120
|
|
|
return false; |
|
121
|
|
|
} |
|
122
|
|
|
|
|
123
|
|
|
$result = $this->db->insertPageOuvrages($data); |
|
124
|
|
|
|
|
125
|
|
|
if ($result === false) { |
|
126
|
|
|
$this->logger->error("Insert DB failed"); |
|
127
|
|
|
} else { |
|
128
|
|
|
$this->logger->notice("Insert DB : ", $result); |
|
129
|
|
|
} |
|
130
|
|
|
|
|
131
|
|
|
return $result; |
|
132
|
|
|
} |
|
133
|
|
|
} |
|
134
|
|
|
|
Dispositif /
Wikibot
