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\Domain\Publisher; |
11
|
|
|
// |
12
|
|
|
//use App\Application\PublisherAction; |
13
|
|
|
//use App\Domain\Models\Wiki\ArticleOrLienBriseInterface; |
14
|
|
|
//use App\Domain\WikiTemplateFactory; |
15
|
|
|
//use Exception; |
16
|
|
|
//use Throwable; |
17
|
|
|
// |
18
|
|
|
///** |
19
|
|
|
// * news URL to {{Article}}. |
20
|
|
|
// * Class ArticleFromURL |
21
|
|
|
// * |
22
|
|
|
// * @package App\Application\Examples |
23
|
|
|
// */ |
24
|
|
|
//class ArticleFromURL |
25
|
|
|
//{ |
26
|
|
|
// private $publisherAction; |
27
|
|
|
// /** |
28
|
|
|
// * @var ArticleOrLienBriseInterface|null |
29
|
|
|
// */ |
30
|
|
|
// private $article; |
31
|
|
|
// /** |
32
|
|
|
// * @var string |
33
|
|
|
// */ |
34
|
|
|
// private $url; |
35
|
|
|
// |
36
|
|
|
// public function __construct(PublisherAction $publisher) |
37
|
|
|
// { |
38
|
|
|
// $this->publisherAction = $publisher; |
39
|
|
|
// $this->url = $publisher->getUrl(); |
40
|
|
|
// try { |
41
|
|
|
// $this->article = $this->process(); |
42
|
|
|
// } catch (Exception $e) { |
43
|
|
|
// dump($e); |
44
|
|
|
// die; |
45
|
|
|
// } |
46
|
|
|
// } |
47
|
|
|
// |
48
|
|
|
// /** |
49
|
|
|
// * @throws Exception |
50
|
|
|
// */ |
51
|
|
|
// private function process(): ?ArticleOrLienBriseInterface |
52
|
|
|
// { |
53
|
|
|
// $mapper = PublisherMapperFactory::fromURL($this->url); |
54
|
|
|
// if (!$mapper) { |
55
|
|
|
// return null; |
56
|
|
|
// } |
57
|
|
|
// sleep(10); |
58
|
|
|
// $arrayLD = []; |
59
|
|
|
// try { |
60
|
|
|
// $html = $this->publisherAction->getHTMLSource(); |
61
|
|
|
// $htmlData = $this->publisherAction->extractWebData($html); |
62
|
|
|
// } catch (Throwable $e) { |
63
|
|
|
// if (strpos($e->getMessage(), '404 Not Found') !== false |
64
|
|
|
// || strpos($e->getMessage(), '410 Gone') !== false |
65
|
|
|
// ) { |
66
|
|
|
// dump('****** lien brisé !!!!'); |
67
|
|
|
// $lienBrise = WikiTemplateFactory::create('lien brisé'); |
68
|
|
|
// $lienBrise->hydrate( |
69
|
|
|
// [ |
70
|
|
|
// 'url' => $this->url, |
71
|
|
|
// 'titre' => 'Article de presse', |
72
|
|
|
// 'brisé le' => date('d-m-Y'), |
73
|
|
|
// ], |
74
|
|
|
// true |
75
|
|
|
// ); |
76
|
|
|
// |
77
|
|
|
// return $lienBrise; // ok |
78
|
|
|
// } |
79
|
|
|
// echo "*** Erreur ".$e->getMessage()."\n"; |
80
|
|
|
// |
81
|
|
|
// return null; |
82
|
|
|
// } |
83
|
|
|
// |
84
|
|
|
// if (empty($htmlData)) { |
85
|
|
|
// echo "*** Pas de data Json-LD ou meta\n"; |
86
|
|
|
// |
87
|
|
|
// return null; |
88
|
|
|
// } |
89
|
|
|
// $htmlData['url'] = $this->url; |
90
|
|
|
// |
91
|
|
|
// try { |
92
|
|
|
// $articleData = $mapper->process($htmlData); |
93
|
|
|
// } catch (Throwable $e) { |
94
|
|
|
// echo sprintf( |
95
|
|
|
// "SKIP : %s %s:%s \n", |
96
|
|
|
// $e->getMessage(), |
97
|
|
|
// $e->getFile(), |
98
|
|
|
// $e->getLine() |
99
|
|
|
// ); |
100
|
|
|
// |
101
|
|
|
// return null; |
102
|
|
|
// } |
103
|
|
|
// |
104
|
|
|
// if (!empty($articleData) && !empty($articleData['titre'])) { |
105
|
|
|
// $article = WikiTemplateFactory::create('article'); |
106
|
|
|
// $article->hydrate($articleData, true); |
107
|
|
|
// if (!$article->hasParamValue('lire en ligne')) { |
108
|
|
|
// $article->setParam('lire en ligne', $this->url); |
109
|
|
|
// } |
110
|
|
|
// |
111
|
|
|
// return $article; // ok |
112
|
|
|
// } |
113
|
|
|
// |
114
|
|
|
// return null; |
115
|
|
|
// } |
116
|
|
|
// |
117
|
|
|
// public function getResult(): ?ArticleOrLienBriseInterface |
118
|
|
|
// { |
119
|
|
|
// return $this->article; |
120
|
|
|
// } |
121
|
|
|
// |
122
|
|
|
//} |
123
|
|
|
|