|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace PubPeerFoundation\PublicationDataExtractor\Resources\Extractors; |
|
4
|
|
|
|
|
5
|
|
|
class Arxiv extends Extractor implements ProvidesPublicationData, ProvidesIdentifiersData, ProvidesAuthorsData, ProvidesJournalData, ProvidesTypesData |
|
6
|
|
|
{ |
|
7
|
|
|
/** |
|
8
|
|
|
* Create search tree. |
|
9
|
|
|
*/ |
|
10
|
|
|
protected function fillSearchTree(): void |
|
11
|
|
|
{ |
|
12
|
|
|
$this->searchTree = $this->document->entry; |
|
13
|
|
|
} |
|
14
|
|
|
|
|
15
|
4 |
|
/** |
|
16
|
|
|
* Extract and format data needed for the Publication Model. |
|
17
|
4 |
|
*/ |
|
18
|
4 |
|
public function extractPublicationData(): void |
|
19
|
|
|
{ |
|
20
|
4 |
|
$this->resourceOutput['publication'] = [ |
|
21
|
|
|
'title' => get_string($this->searchTree, 'title'), |
|
22
|
4 |
|
'abstract' => get_string($this->searchTree, 'summary'), |
|
23
|
|
|
'url' => get_string($this->searchTree, 'id'), |
|
24
|
4 |
|
'published_at' => date_from_parseable_format(get_string($this->searchTree, 'published')), |
|
25
|
4 |
|
]; |
|
26
|
4 |
|
} |
|
27
|
4 |
|
|
|
28
|
4 |
|
/** |
|
29
|
|
|
* Extract and format data needed for the Identifiers Relationship |
|
30
|
4 |
|
* on the Publication Model. |
|
31
|
|
|
*/ |
|
32
|
|
|
public function extractIdentifiersData(): void |
|
33
|
4 |
|
{ |
|
34
|
|
|
$this->resourceOutput['identifiers'][] = [ |
|
35
|
4 |
|
'value' => (string) $this->getIdentifier(), |
|
36
|
4 |
|
'type' => 'arxiv', |
|
37
|
|
|
]; |
|
38
|
|
|
|
|
39
|
|
|
$this->resourceOutput['identifiers'][] = [ |
|
40
|
|
|
'value' => '2331-8422', |
|
41
|
4 |
|
'type' => 'issn', |
|
42
|
|
|
]; |
|
43
|
4 |
|
|
|
44
|
4 |
|
$namespace = $this->searchTree->getNamespaces(true); |
|
45
|
4 |
|
foreach ($this->searchTree->children($namespace['arxiv']) as $key => $child) { |
|
46
|
4 |
|
if ('doi' === $key) { |
|
47
|
4 |
|
$this->resourceOutput['identifiers'][] = [ |
|
48
|
|
|
'value' => (string) $child, |
|
49
|
4 |
|
'type' => 'doi', |
|
50
|
|
|
]; |
|
51
|
|
|
} |
|
52
|
|
|
} |
|
53
|
|
|
} |
|
54
|
|
|
|
|
55
|
4 |
|
/** |
|
56
|
|
|
* Extract and format data needed for the Journals Relationship |
|
57
|
4 |
|
* on the Publication Model. |
|
58
|
4 |
|
*/ |
|
59
|
4 |
|
public function extractJournalData(): void |
|
60
|
|
|
{ |
|
61
|
|
|
$this->resourceOutput['journal'] = [ |
|
62
|
4 |
|
'title' => 'arXiv', |
|
63
|
|
|
'issn' => ['2331-8422'], |
|
64
|
|
|
]; |
|
65
|
|
|
} |
|
66
|
4 |
|
|
|
67
|
|
|
/** |
|
68
|
|
|
* Extract and format data needed for the Authors Relationship |
|
69
|
|
|
* on the Publication Model. |
|
70
|
|
|
*/ |
|
71
|
|
|
public function extractAuthorsData(): void |
|
72
|
4 |
|
{ |
|
73
|
|
|
foreach ($this->searchTree->author as $author) { |
|
74
|
4 |
|
$name = explode(' ', $author->name, 2); |
|
75
|
|
|
$this->resourceOutput['authors'][] = [ |
|
76
|
|
|
'first_name' => $name[0] ?? null, |
|
77
|
|
|
'last_name' => $name[1] ?? null, |
|
78
|
4 |
|
]; |
|
79
|
|
|
} |
|
80
|
|
|
} |
|
81
|
|
|
|
|
82
|
|
|
/** |
|
83
|
|
|
* Extract and format data needed for the Types Relationship |
|
84
|
4 |
|
* on the Publication Model. |
|
85
|
|
|
*/ |
|
86
|
4 |
|
public function extractTypesData(): void |
|
87
|
4 |
|
{ |
|
88
|
4 |
|
$this->resourceOutput['types'][] = [ |
|
89
|
4 |
|
'name' => 'arxiv', |
|
90
|
4 |
|
]; |
|
91
|
|
|
} |
|
92
|
|
|
|
|
93
|
4 |
|
/** |
|
94
|
|
|
* Get the Identifier from the data. |
|
95
|
|
|
* |
|
96
|
|
|
* @return string |
|
97
|
|
|
*/ |
|
98
|
|
|
protected function getIdentifier(): string |
|
99
|
4 |
|
{ |
|
100
|
|
|
preg_match('/(\d{2}(0|1)[0-9]\.\d{4,5})(v|V)?(\d)?/', $this->searchTree->id[0], $matches); |
|
101
|
4 |
|
|
|
102
|
|
|
return $matches[1]; |
|
103
|
|
|
} |
|
104
|
|
|
} |
|
105
|
|
|
|
PubPeerFoundation /
PublicationDataExtractor
