|
1
|
|
|
<?php |
|
2
|
|
|
/** |
|
3
|
|
|
* This file is part of dispositif/wikibot application |
|
4
|
|
|
* 2019 : Philippe M. <[email protected]> |
|
5
|
|
|
* For the full copyright and MIT license information, please view the LICENSE file. |
|
6
|
|
|
*/ |
|
7
|
|
|
|
|
8
|
|
|
declare(strict_types=1); |
|
9
|
|
|
|
|
10
|
|
|
namespace App\Application; |
|
11
|
|
|
|
|
12
|
|
|
use App\Domain\Utils\TemplateParser; |
|
13
|
|
|
use App\Infrastructure\PageListInterface; |
|
14
|
|
|
use Exception; |
|
15
|
|
|
use Mediawiki\Api\MediawikiFactory; |
|
16
|
|
|
|
|
17
|
|
|
|
|
18
|
|
|
/** |
|
19
|
|
|
* From a titles list, scan the wiki and add the {ouvrage} citations into the database. |
|
20
|
|
|
*/ |
|
21
|
|
|
class ScanWiki2DB |
|
22
|
|
|
{ |
|
23
|
|
|
private $wiki; |
|
24
|
|
|
private $db; |
|
25
|
|
|
private $bot; |
|
26
|
|
|
/** |
|
27
|
|
|
* @var PageListInterface |
|
28
|
|
|
*/ |
|
29
|
|
|
private $pageList; |
|
30
|
|
|
private $priority; |
|
31
|
|
|
|
|
32
|
|
|
public function __construct( |
|
33
|
|
|
MediawikiFactory $wiki, |
|
34
|
|
|
QueueInterface $dbAdapter, |
|
35
|
|
|
WikiBotConfig $bot, |
|
36
|
|
|
PageListInterface $list, |
|
37
|
|
|
?int $priority |
|
38
|
|
|
= 0 |
|
39
|
|
|
) { |
|
40
|
|
|
$this->wiki = $wiki; // ServiceFactory::wikiApi(); |
|
41
|
|
|
$this->db = $dbAdapter; |
|
42
|
|
|
$this->bot = $bot; |
|
43
|
|
|
$this->pageList = $list; |
|
44
|
|
|
$this->priority = $priority; |
|
45
|
|
|
|
|
46
|
|
|
$this->process(); |
|
47
|
|
|
} |
|
48
|
|
|
|
|
49
|
|
|
/** |
|
50
|
|
|
* @throws Exception |
|
51
|
|
|
*/ |
|
52
|
|
|
public function process():void |
|
53
|
|
|
{ |
|
54
|
|
|
$titles = $this->pageList->getPageTitles(); |
|
55
|
|
|
if (empty($titles)) { |
|
56
|
|
|
echo "pageList vide.\n"; |
|
57
|
|
|
|
|
58
|
|
|
return; |
|
59
|
|
|
} |
|
60
|
|
|
foreach ($titles as $title) { |
|
61
|
|
|
$this->pageScan($title); |
|
62
|
|
|
sleep(4); |
|
63
|
|
|
} |
|
64
|
|
|
} |
|
65
|
|
|
|
|
66
|
|
|
/** |
|
67
|
|
|
* @param string $title |
|
68
|
|
|
* |
|
69
|
|
|
* @return bool |
|
70
|
|
|
* @throws Exception |
|
71
|
|
|
*/ |
|
72
|
|
|
public function pageScan(string $title): bool |
|
73
|
|
|
{ |
|
74
|
|
|
sleep(2); |
|
75
|
|
|
echo "\n-------------------------------------\n\n"; |
|
76
|
|
|
echo date("Y-m-d H:i:s")."\n"; |
|
77
|
|
|
echo $title."\n"; |
|
78
|
|
|
|
|
79
|
|
|
$page = new WikiPageAction($this->wiki, $title); // todo injection |
|
80
|
|
|
$ns = $page->getNs(); |
|
81
|
|
|
if ($ns !== 0) { |
|
82
|
|
|
echo "SKIP : namespace $ns"; |
|
83
|
|
|
|
|
84
|
|
|
return false; |
|
85
|
|
|
} |
|
86
|
|
|
$text = $page->getText(); |
|
87
|
|
|
if (empty($text)) { |
|
88
|
|
|
echo "SKIP : texte vide\n"; |
|
89
|
|
|
|
|
90
|
|
|
return false; |
|
91
|
|
|
} |
|
92
|
|
|
|
|
93
|
|
|
try { |
|
94
|
|
|
$parsedTemplates = TemplateParser::parseAllTemplateByName('ouvrage', $text); |
|
95
|
|
|
} catch (Exception $e) { |
|
96
|
|
|
dump($e); |
|
97
|
|
|
|
|
98
|
|
|
return false; |
|
99
|
|
|
} |
|
100
|
|
|
|
|
101
|
|
|
if (empty($parsedTemplates)) { |
|
102
|
|
|
return false; |
|
103
|
|
|
} |
|
104
|
|
|
|
|
105
|
|
|
$result = $this->insertDB($parsedTemplates['ouvrage'], $title); |
|
106
|
|
|
|
|
107
|
|
|
return !empty($result); |
|
108
|
|
|
} |
|
109
|
|
|
|
|
110
|
|
|
/** |
|
111
|
|
|
* @param $ouvrages |
|
112
|
|
|
* @param string $title |
|
113
|
|
|
* |
|
114
|
|
|
* @return mixed |
|
115
|
|
|
*/ |
|
116
|
|
|
protected function insertDB($ouvrages, string $title) |
|
117
|
|
|
{ |
|
118
|
|
|
$data = []; |
|
119
|
|
|
foreach ($ouvrages as $res) { |
|
120
|
|
|
$oneData = [ |
|
121
|
|
|
'page' => $title, |
|
122
|
|
|
'raw' => $res['raw'], |
|
123
|
|
|
'priority' => $this->priority, |
|
124
|
|
|
]; |
|
125
|
|
|
// filtre doublon |
|
126
|
|
|
if (!in_array($oneData, $data)) { |
|
127
|
|
|
$data[] = $oneData; |
|
128
|
|
|
} |
|
129
|
|
|
} |
|
130
|
|
|
|
|
131
|
|
|
$result = $this->db->insertPageOuvrages($data); |
|
|
|
|
|
|
132
|
|
|
dump($result); |
|
133
|
|
|
|
|
134
|
|
|
return $result; |
|
135
|
|
|
} |
|
136
|
|
|
|
|
137
|
|
|
} |
|
138
|
|
|
|
Dispositif /
Wikibot
Loading history...