|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace PubPeerFoundation\PublicationDataExtractor\Resources\Extractors; |
|
4
|
|
|
|
|
5
|
|
|
use PubPeerFoundation\PublicationDataExtractor\Support\UpdateTypesStandardiser; |
|
6
|
|
|
use PubPeerFoundation\PublicationDataExtractor\Exceptions\UnparseableApiException; |
|
7
|
|
|
use PubPeerFoundation\PublicationDataExtractor\Exceptions\JournalTitleNotFoundException; |
|
8
|
|
|
|
|
9
|
|
|
class Crossref implements Extractor, ProvidesPublicationData, ProvidesIdentifiersData, ProvidesJournalData, ProvidesAuthorsData, ProvidesUpdatesData |
|
10
|
|
|
{ |
|
11
|
|
|
protected $document; |
|
12
|
|
|
|
|
13
|
|
|
protected $searchTree; |
|
14
|
|
|
|
|
15
|
|
|
protected $output = []; |
|
16
|
|
|
|
|
17
|
24 |
|
public function __construct($document) |
|
18
|
|
|
{ |
|
19
|
24 |
|
$this->document = $document; |
|
20
|
24 |
|
} |
|
21
|
|
|
|
|
22
|
11 |
|
public function extract(): array |
|
23
|
|
|
{ |
|
24
|
11 |
|
$this->getDataFromDocument(); |
|
25
|
|
|
|
|
26
|
9 |
|
$this->extractAuthorsData(); |
|
27
|
9 |
|
$this->extractIdentifiersData(); |
|
28
|
9 |
|
try { |
|
29
|
9 |
|
$this->extractJournalData(); |
|
30
|
9 |
|
} catch (JournalTitleNotFoundException $e) { |
|
31
|
9 |
|
$this->searchTree['container-title'] = $this->searchTree['publisher']; |
|
32
|
9 |
|
} |
|
33
|
|
|
$this->extractPublicationData(); |
|
34
|
9 |
|
$this->extractTagsData(); |
|
35
|
|
|
$this->extractTypesData(); |
|
36
|
|
|
$this->extractUpdatesData(); |
|
37
|
7 |
|
|
|
38
|
|
|
return $this->output; |
|
39
|
7 |
|
} |
|
40
|
2 |
|
|
|
41
|
|
|
protected function getDataFromDocument() |
|
42
|
|
|
{ |
|
43
|
5 |
|
if ('ok' !== $this->document['status']) { |
|
44
|
5 |
|
throw new UnparseableApiException(); |
|
45
|
|
|
} |
|
46
|
|
|
|
|
47
|
|
|
$this->searchTree = $this->document['message']; |
|
48
|
|
|
} |
|
49
|
11 |
|
|
|
50
|
|
|
/** |
|
51
|
11 |
|
* Extract and format data needed for the Publication Model. |
|
52
|
|
|
*/ |
|
53
|
11 |
|
public function extractPublicationData() |
|
54
|
11 |
|
{ |
|
55
|
11 |
|
$date = $this->extractDateFrom(['published-print', 'published-online', 'issued']); |
|
56
|
11 |
|
|
|
57
|
11 |
|
$this->output['publication'] = [ |
|
58
|
|
|
'title' => get_string($this->searchTree, 'title'), |
|
59
|
11 |
|
'abstract' => get_string($this->searchTree, 'abstract'), |
|
60
|
|
|
'url' => $this->searchTree['URL'] ?? null, |
|
61
|
|
|
'published_at' => $date, |
|
62
|
|
|
]; |
|
63
|
|
|
} |
|
64
|
11 |
|
|
|
65
|
|
|
/** |
|
66
|
11 |
|
* Extract and format data needed for the Identifiers Relationship |
|
67
|
|
|
* on the Publication Model. |
|
68
|
|
|
*/ |
|
69
|
|
|
public function extractIdentifiersData() |
|
70
|
|
|
{ |
|
71
|
|
|
if (! empty($this->searchTree['DOI'])) { |
|
72
|
|
|
$this->output['identifiers'][] = [ |
|
73
|
17 |
|
'value' => $this->searchTree['DOI'], |
|
74
|
|
|
'type' => 'doi', |
|
75
|
17 |
|
]; |
|
76
|
9 |
|
} |
|
77
|
9 |
|
|
|
78
|
9 |
|
foreach (get_array($this->searchTree, 'ISSN') as $issn) { |
|
79
|
|
|
$this->output['identifiers'][] = [ |
|
80
|
|
|
'value' => $issn, |
|
81
|
|
|
'type' => 'issn', |
|
82
|
17 |
|
]; |
|
83
|
12 |
|
} |
|
84
|
11 |
|
} |
|
85
|
11 |
|
|
|
86
|
11 |
|
/** |
|
87
|
11 |
|
* Extract and format data needed for the Journal Relationship |
|
88
|
|
|
* on the Publication Model. |
|
89
|
|
|
* |
|
90
|
|
|
* @throws JournalTitleNotFoundException |
|
91
|
1 |
|
*/ |
|
92
|
1 |
|
public function extractJournalData() |
|
93
|
1 |
|
{ |
|
94
|
|
|
if (empty($this->searchTree['container-title']) && empty($this->searchTree['ISSN'])) { |
|
95
|
|
|
throw new JournalTitleNotFoundException(); |
|
96
|
|
|
} |
|
97
|
17 |
|
|
|
98
|
|
|
$this->output['journal'] = [ |
|
99
|
|
|
'title' => get_string($this->searchTree, 'container-title'), |
|
100
|
|
|
'issn' => get_array($this->searchTree, 'ISSN'), |
|
101
|
|
|
'publisher' => get_string($this->searchTree, 'publisher'), |
|
102
|
|
|
]; |
|
103
|
12 |
|
} |
|
104
|
|
|
|
|
105
|
12 |
|
/** |
|
106
|
1 |
|
* Extract and format data needed for the Authors Relationship |
|
107
|
|
|
* on the Publication Model. |
|
108
|
|
|
*/ |
|
109
|
12 |
|
public function extractAuthorsData() |
|
110
|
12 |
|
{ |
|
111
|
12 |
|
foreach (get_array($this->searchTree, 'author') as $author) { |
|
112
|
12 |
|
if (isset($author['family'])) { |
|
113
|
|
|
$this->output['authors'][] = [ |
|
114
|
12 |
|
'first_name' => get_string($author, 'given'), |
|
115
|
|
|
'last_name' => get_string($author, 'family'), |
|
116
|
|
|
'orcid' => get_string($author, 'ORCID'), |
|
117
|
|
|
'affiliation' => get_array($author, 'affiliation'), |
|
118
|
|
|
]; |
|
119
|
|
|
} |
|
120
|
9 |
|
} |
|
121
|
|
|
} |
|
122
|
9 |
|
|
|
123
|
9 |
|
/** |
|
124
|
9 |
|
* Extract and format data needed for the Types Relationship |
|
125
|
9 |
|
* on the Publication Model. |
|
126
|
9 |
|
*/ |
|
127
|
9 |
|
public function extractTypesData() |
|
128
|
9 |
|
{ |
|
129
|
9 |
|
$this->output['types'][] = [ |
|
130
|
|
|
'name' => get_string($this->searchTree, 'type'), |
|
131
|
|
|
]; |
|
132
|
|
|
} |
|
133
|
|
|
|
|
134
|
9 |
|
/** |
|
135
|
|
|
* Extract and format data needed for the tags Relationship |
|
136
|
|
|
* on the Publication Model. |
|
137
|
|
|
*/ |
|
138
|
|
|
public function extractTagsData() |
|
139
|
|
|
{ |
|
140
|
9 |
|
foreach (get_array($this->searchTree, 'subject') as $tag) { |
|
141
|
|
|
$this->output['tags'][] = [ |
|
142
|
9 |
|
'name' => $tag, |
|
143
|
9 |
|
]; |
|
144
|
|
|
} |
|
145
|
9 |
|
} |
|
146
|
|
|
|
|
147
|
|
|
protected function extractDateFrom($array) |
|
148
|
|
|
{ |
|
149
|
|
|
$datePartsContainer = array_values(array_filter($array, function ($string) { |
|
150
|
|
|
return isset($this->searchTree[$string]); |
|
151
|
9 |
|
}))[0]; |
|
152
|
|
|
|
|
153
|
9 |
|
return date_from_parts($this->searchTree[$datePartsContainer]['date-parts'][0]); |
|
154
|
9 |
|
} |
|
155
|
9 |
|
|
|
156
|
9 |
|
public function extractUpdatesData() |
|
157
|
|
|
{ |
|
158
|
|
|
foreach (get_array($this->searchTree, 'update-to') as $update) { |
|
159
|
|
|
$this->output['updates'][] = [ |
|
160
|
9 |
|
'timestamp' => $update['updated']['timestamp'], |
|
161
|
|
|
'identifier' => [ |
|
162
|
|
|
'doi' => get_string($update, 'DOI'), |
|
163
|
|
|
], |
|
164
|
11 |
|
'type' => UpdateTypesStandardiser::getType($update['type']), |
|
165
|
11 |
|
]; |
|
166
|
11 |
|
} |
|
167
|
|
|
} |
|
168
|
|
|
} |
|
169
|
|
|
|
PubPeerFoundation /
PublicationDataExtractor
