|
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\Examples; |
|
11
|
|
|
|
|
12
|
|
|
use App\Application\Bot; |
|
13
|
|
|
use App\Application\WikiPageAction; |
|
14
|
|
|
use App\Domain\Utils\TemplateParser; |
|
15
|
|
|
use App\Infrastructure\DbAdapter; |
|
16
|
|
|
use App\Infrastructure\ServiceFactory; |
|
17
|
|
|
|
|
18
|
|
|
//use App\Application\CLI; |
|
19
|
|
|
|
|
20
|
|
|
include __DIR__.'/../myBootstrap.php'; |
|
21
|
|
|
|
|
22
|
|
|
/** |
|
23
|
|
|
* From json list of articles => add to SQL TempRawOpti |
|
24
|
|
|
*/ |
|
25
|
|
|
$process = new ArticleScan(); |
|
26
|
|
|
//$json = file_get_contents(__DIR__.'/../resources/titles.json'); |
|
27
|
|
|
//$articles = json_decode($json, true); |
|
28
|
|
|
|
|
29
|
|
|
$articles = file(__DIR__.'/../resources/importISBN_nov.txt'); |
|
30
|
|
|
|
|
31
|
|
|
foreach ($articles as $article) { |
|
32
|
|
|
$article = trim($article); |
|
33
|
|
|
if (empty($article)) { |
|
34
|
|
|
continue; |
|
35
|
|
|
} |
|
36
|
|
|
$process->pageScan($article, 0); |
|
37
|
|
|
sleep(4); |
|
38
|
|
|
} |
|
39
|
|
|
|
|
40
|
|
|
class ArticleScan |
|
41
|
|
|
{ |
|
42
|
|
|
private $wiki; |
|
43
|
|
|
private $db; |
|
44
|
|
|
private $bot; |
|
45
|
|
|
|
|
46
|
|
|
public function __construct() |
|
47
|
|
|
{ |
|
48
|
|
|
$this->wiki = ServiceFactory::wikiApi(); |
|
49
|
|
|
$this->db = new DbAdapter(); |
|
50
|
|
|
$this->bot = new Bot(); |
|
51
|
|
|
} |
|
52
|
|
|
|
|
53
|
|
|
public function pageScan(string $title, ?int $priority = 0): bool |
|
54
|
|
|
{ |
|
55
|
|
|
sleep(2); |
|
56
|
|
|
echo "\n-------------------------------------\n\n"; |
|
57
|
|
|
echo date("Y-m-d H:i:s")."\n"; |
|
58
|
|
|
echo $title."\n"; |
|
59
|
|
|
|
|
60
|
|
|
$page = new WikiPageAction($this->wiki, $title); |
|
61
|
|
|
$text = $page->getText(); |
|
62
|
|
|
|
|
63
|
|
|
// Skip AdQ |
|
64
|
|
|
if (preg_match('#{{ ?En-tête label#i', $text) > 0) { |
|
65
|
|
|
echo "SKIP : AdQ ou BA.\n"; |
|
66
|
|
|
|
|
67
|
|
|
//$this->db->skipArticle($title); |
|
68
|
|
|
|
|
69
|
|
|
return false; |
|
70
|
|
|
} |
|
71
|
|
|
|
|
72
|
|
|
try { |
|
73
|
|
|
$parse = TemplateParser::parseAllTemplateByName('ouvrage', $text); |
|
|
|
|
|
|
74
|
|
|
} catch (\Exception $e) { |
|
75
|
|
|
dump($e); |
|
76
|
|
|
|
|
77
|
|
|
return false; |
|
78
|
|
|
} |
|
79
|
|
|
|
|
80
|
|
|
if (empty($parse)) { |
|
81
|
|
|
return false; |
|
82
|
|
|
} |
|
83
|
|
|
$data = []; |
|
84
|
|
|
foreach ($parse['ouvrage'] as $res) { |
|
85
|
|
|
$thisdata = [ |
|
86
|
|
|
'page' => $title, |
|
87
|
|
|
'raw' => $res['raw'], |
|
88
|
|
|
'priority' => $priority, |
|
89
|
|
|
]; |
|
90
|
|
|
// filtre doublon |
|
91
|
|
|
if (!in_array($thisdata, $data)) { |
|
92
|
|
|
$data[] = $thisdata; |
|
93
|
|
|
} |
|
94
|
|
|
} |
|
95
|
|
|
|
|
96
|
|
|
$result = $this->db->insertTempRawOpti($data); |
|
97
|
|
|
dump($result); |
|
98
|
|
|
return !empty($result); |
|
99
|
|
|
} |
|
100
|
|
|
|
|
101
|
|
|
} |
|
102
|
|
|
|
Dispositif /
Wikibot
Loading history...