1
|
|
|
<?php |
2
|
|
|
/* |
3
|
|
|
* This file is part of dispositif/wikibot application (@github) |
4
|
|
|
* 2019/2020 © Philippe/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
|
|
|
|
11
|
|
|
namespace App\Domain\Publisher; |
12
|
|
|
|
13
|
|
|
|
14
|
|
|
use Exception; |
15
|
|
|
|
16
|
|
|
class OpenGraphMapper implements MapperInterface |
17
|
|
|
{ |
18
|
|
|
use ExternConverterTrait; |
19
|
|
|
|
20
|
|
|
/** |
21
|
|
|
* Tenir compte du <title>bla</title> pour générer un {lien web} ? |
22
|
|
|
*/ |
23
|
|
|
protected $htmlTitleAllowed = true; |
24
|
|
|
|
25
|
|
|
protected $titleFromHtmlState = false; |
26
|
|
|
|
27
|
|
|
/** |
28
|
|
|
* @param array|null $options |
29
|
|
|
*/ |
30
|
|
|
public function __construct(?array $options = []) |
31
|
5 |
|
{ |
32
|
|
|
if (!empty($options['htmlTitleAllowed']) && is_bool($options['htmlTitleAllowed'])) { |
33
|
|
|
$this->htmlTitleAllowed = $options['htmlTitleAllowed']; |
34
|
5 |
|
} |
35
|
5 |
|
} |
36
|
5 |
|
|
37
|
5 |
|
/** |
38
|
5 |
|
* todo pretty ALLCAP UGLY TITLE |
39
|
5 |
|
* Mapping from Open Graph and Dublin Core meta tags |
40
|
5 |
|
* https://ogp.me/ |
41
|
|
|
* https://www.dublincore.org/schemas/ |
42
|
5 |
|
* todo extract DC ? |
43
|
5 |
|
* |
44
|
5 |
|
* @param array $meta |
45
|
|
|
* |
46
|
5 |
|
* @return array |
47
|
5 |
|
* @throws Exception |
48
|
5 |
|
*/ |
49
|
5 |
|
public function process($meta): array |
50
|
|
|
{ |
51
|
5 |
|
// Mode "pas de titre html" |
52
|
|
|
if (!$this->htmlTitleAllowed && !empty($meta['html-title'])) { |
53
|
|
|
unset($meta['html-title']); |
54
|
5 |
|
} |
55
|
5 |
|
|
56
|
5 |
|
// si usage <title> HTML |
57
|
5 |
|
if ($this->htmlTitleAllowed |
58
|
|
|
&& empty($meta['og:title']) |
59
|
5 |
|
&& empty($meta['twitter:title']) |
60
|
5 |
|
&& empty($meta['DC.title']) |
61
|
5 |
|
&& !empty($meta['html-title']) |
62
|
5 |
|
) { |
63
|
|
|
$this->titleFromHtmlState = true; |
64
|
|
|
} |
65
|
|
|
|
66
|
5 |
|
return [ |
67
|
5 |
|
'DATA-TYPE' => 'Open Graph/Dublin Core', |
68
|
5 |
|
'DATA-ARTICLE' => $this->isAnArticle($meta['og:type'] ?? ''), |
69
|
5 |
|
'site' => $this->clean($meta['og:site_name'] ?? null), |
70
|
5 |
|
'titre' => $this->clean( |
71
|
5 |
|
$meta['og:title'] ?? $meta['twitter:title'] ?? $meta['DC.title'] ?? $meta['html-title'] ?? null |
72
|
5 |
|
), |
73
|
5 |
|
'url' => $meta['og:url'] ?? $meta['URL'] ?? $meta['html-url'] ?? null, |
74
|
|
|
'langue' => $this->convertLangue( |
75
|
|
|
$meta['og:locale'] ?? $meta['DC.language'] ?? $meta['citation_language'] ?? $meta['lang'] ?? |
76
|
|
|
$meta['language'] ?? |
77
|
|
|
$meta['content-language'] ?? $meta['Content-Language'] ?? $meta['html-lang'] ?? null |
78
|
|
|
), |
79
|
|
|
'consulté le' => date('d-m-Y'), |
80
|
|
|
'auteur' => $this->cleanAuthor( |
81
|
|
|
$meta['og:article:author'] ?? |
82
|
|
|
$meta['article:author'] ?? $meta['citation_author'] ?? $meta['article:author_name'] ?? null |
83
|
|
|
), |
84
|
|
|
'format' => $this->convertOGtype2format($meta['og:type'] ?? null), |
85
|
|
|
'date' => $this->convertDate( |
86
|
|
|
$meta['og:article:published_time'] ?? $meta['article:published_time'] ?? |
87
|
|
|
$meta['DC.date'] ?? $meta['citation_date'] ?? $meta['citation_publication_date'] ?? null |
88
|
|
|
), |
89
|
|
|
'accès url' => $this->convertURLaccess($meta), |
90
|
|
|
|
91
|
|
|
// DUBLIN CORE ONLY |
92
|
|
|
'périodique' => $this->clean($meta['DC.isPartOf'] ?? $meta['citation_journal_title'] ?? null), |
93
|
|
|
'et al.' => $this->authorsEtAl( |
94
|
|
|
$meta['citation_authors'] ?? $meta['DC.Contributor'] ?? null, |
95
|
|
|
true |
96
|
|
|
), |
97
|
|
|
'auteur1' => $this->wikifyPressAgency( |
98
|
|
|
$this->cleanAuthor( |
99
|
|
|
$this->authorsEtAl($meta['citation_authors'] ?? $meta['DC.Contributor'] ?? $meta['Author'] ?? null) |
100
|
|
|
) |
101
|
|
|
), |
102
|
|
|
'volume' => $meta['citation_volume'] ?? null, |
103
|
|
|
'numéro' => $meta['citation_issue'] ?? null, |
104
|
|
|
'page' => $this->convertDCpage($meta), |
105
|
|
|
'doi' => $meta['citation_doi'] ?? $meta['DOI'] ?? null, |
106
|
|
|
'éditeur' => $meta['DC.publisher'] ?? $meta['dc.publisher'] ?? null, // Persée dégeulasse todo? |
107
|
|
|
'pmid' => $meta['citation_pmid'] ?? null, |
108
|
|
|
'issn' => $meta["citation_issn"] ?? null, |
109
|
|
|
'isbn' => $meta["citation_isbn"] ?? null, |
110
|
|
|
// "prism.eIssn" => "2262-7197" |
111
|
|
|
|
112
|
|
|
// WIKI |
113
|
|
|
//'note' => $this->addNote(), |
114
|
|
|
]; |
115
|
|
|
} |
116
|
|
|
} |
117
|
|
|
|