|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace PubPeerFoundation\PublicationDataExtractor\Resources\Extractors; |
|
4
|
|
|
|
|
5
|
|
|
use Tightenco\Collect\Support\Arr; |
|
6
|
|
|
|
|
7
|
|
|
class EutilsEfetch extends Extractor implements ProvidesPublicationData, ProvidesIdentifiersData, ProvidesAuthorsData, ProvidesJournalData, ProvidesUpdatesData |
|
8
|
|
|
{ |
|
9
|
|
|
protected $updateTypes = [ |
|
10
|
|
|
'ErratumIn' => 'Erratum', |
|
11
|
|
|
'RetractionIn' => 'Retraction', |
|
12
|
|
|
'ExpressionOfConcernIn' => 'Expression of concern', |
|
13
|
|
|
]; |
|
14
|
|
|
|
|
15
|
|
|
/** |
|
16
|
8 |
|
* Create search tree. |
|
17
|
|
|
*/ |
|
18
|
8 |
|
protected function fillSearchTree(): void |
|
19
|
8 |
|
{ |
|
20
|
8 |
|
$this->searchTree = $this->document->{'PubmedArticle'}; |
|
21
|
|
|
} |
|
22
|
4 |
|
|
|
23
|
|
|
/** |
|
24
|
4 |
|
* Extract and format data needed for the Publication Model. |
|
25
|
|
|
*/ |
|
26
|
4 |
|
public function extractPublicationData(): void |
|
27
|
4 |
|
{ |
|
28
|
4 |
|
$this->resourceOutput['publication'] = [ |
|
29
|
4 |
|
'title' => get_string($this->searchTree, 'MedlineCitation.Article.ArticleTitle'), |
|
30
|
|
|
'url' => (string) 'http://www.ncbi.nlm.nih.gov/pubmed/'.get_string($this->searchTree, 'MedlineCitation.PMID'), |
|
31
|
4 |
|
'published_at' => date_from_pub_date(data_get($this->searchTree, 'MedlineCitation.Article.Journal.JournalIssue.PubDate')), |
|
32
|
|
|
'abstract' => get_string($this->searchTree, 'MedlineCitation.Article.Abstract.AbstractText'), |
|
33
|
|
|
]; |
|
34
|
4 |
|
} |
|
35
|
|
|
|
|
36
|
4 |
|
/** |
|
37
|
4 |
|
* Extract and format data needed for the Identifiers Relationship |
|
38
|
|
|
* on the Publication Model. |
|
39
|
|
|
*/ |
|
40
|
|
|
public function extractIdentifiersData(): void |
|
41
|
|
|
{ |
|
42
|
4 |
|
foreach ($this->searchTree->PubmedData->ArticleIdList->ArticleId as $identifier) { |
|
43
|
|
|
$this->resourceOutput['identifiers'][] = [ |
|
44
|
4 |
|
'value' => (string) $identifier, |
|
45
|
4 |
|
'type' => (string) $identifier['IdType'], |
|
46
|
4 |
|
]; |
|
47
|
4 |
|
} |
|
48
|
4 |
|
if ($value = get_string($this->searchTree, 'MedlineCitation.Article.Journal.ISSN')) { |
|
49
|
|
|
$this->resourceOutput['identifiers'][] = [ |
|
50
|
4 |
|
'value' => $value, |
|
51
|
|
|
'type' => 'issn', |
|
52
|
|
|
]; |
|
53
|
|
|
} |
|
54
|
|
|
} |
|
55
|
|
|
|
|
56
|
4 |
|
/** |
|
57
|
|
|
* Extract and format data needed for the Journals Relationship |
|
58
|
4 |
|
* on the Publication Model. |
|
59
|
4 |
|
*/ |
|
60
|
4 |
|
public function extractJournalData(): void |
|
61
|
4 |
|
{ |
|
62
|
|
|
$this->resourceOutput['journal'] = [ |
|
63
|
|
|
'title' => get_string($this->searchTree, 'MedlineCitation.Article.Journal.Title'), |
|
64
|
|
|
'issn' => $this->getIssns(), |
|
65
|
4 |
|
]; |
|
66
|
4 |
|
} |
|
67
|
4 |
|
|
|
68
|
4 |
|
/** |
|
69
|
|
|
* Extract and format data needed for the Authors Relationship |
|
70
|
|
|
* on the Publication Model. |
|
71
|
4 |
|
*/ |
|
72
|
|
|
public function extractAuthorsData(): void |
|
73
|
|
|
{ |
|
74
|
|
|
try { |
|
75
|
|
|
$this->loopOverAuthors(); |
|
76
|
|
|
} catch (\Exception $e) { |
|
77
|
4 |
|
// Empty catch block, don't want anything to happen in case of exception. |
|
78
|
|
|
} |
|
79
|
4 |
|
} |
|
80
|
|
|
|
|
81
|
4 |
|
public function extractUpdatesData(): void |
|
82
|
4 |
|
{ |
|
83
|
|
|
foreach (get_array($this->searchTree, 'MedlineCitation.CommentsCorrectionsList.CommentsCorrections') as $correction) { |
|
84
|
|
|
if (in_array(stringify($correction['RefType']), array_keys($this->updateTypes))) { |
|
85
|
4 |
|
$this->getUpdateFromCorrection($correction); |
|
86
|
4 |
|
} |
|
87
|
|
|
} |
|
88
|
|
|
} |
|
89
|
4 |
|
|
|
90
|
4 |
|
/** |
|
91
|
4 |
|
* @param $correction |
|
92
|
|
|
*/ |
|
93
|
4 |
|
protected function getUpdateFromCorrection($correction): void |
|
94
|
|
|
{ |
|
95
|
|
|
try { |
|
96
|
|
|
$this->resourceOutput['updates'][] = [ |
|
97
|
|
|
'timestamp' => $this->getUpdateTimestamp(stringify($correction->RefSource)), |
|
98
|
|
|
'identifier' => [ |
|
99
|
8 |
|
'pubmed' => get_string($correction, 'PMID'), |
|
100
|
|
|
], |
|
101
|
|
|
'type' => $this->getReadableUpdateType(stringify($correction['RefType'])), |
|
102
|
8 |
|
]; |
|
103
|
7 |
|
} catch (\Exception $e) { |
|
104
|
7 |
|
// Don't stop in case of unreadable date format |
|
105
|
7 |
|
} |
|
106
|
7 |
|
} |
|
107
|
|
|
|
|
108
|
|
|
protected function getReadableUpdateType($refType) |
|
109
|
1 |
|
{ |
|
110
|
|
|
return $this->updateTypes[$refType]; |
|
111
|
8 |
|
} |
|
112
|
|
|
|
|
113
|
|
|
/** |
|
114
|
|
|
* Get all available ISSNs values from the tree. |
|
115
|
|
|
* |
|
116
|
|
|
* @return array |
|
117
|
|
|
*/ |
|
118
|
|
|
protected function getIssns() |
|
119
|
|
|
{ |
|
120
|
|
|
$issn = []; |
|
121
|
|
|
|
|
122
|
|
|
if ($number = get_string($this->searchTree, 'MedlineCitation.Article.Journal.ISSN')) { |
|
123
|
|
|
$issn[] = $number; |
|
124
|
|
|
} |
|
125
|
|
|
|
|
126
|
|
|
if ($number = get_string($this->searchTree, 'MedlineCitation.MedlineJournalInfo.ISSNLinking')) { |
|
127
|
|
|
$issn[] = $number; |
|
128
|
|
|
} |
|
129
|
|
|
|
|
130
|
|
|
return $issn; |
|
131
|
|
|
} |
|
132
|
|
|
|
|
133
|
|
|
/** |
|
134
|
|
|
* Loop over authors array. |
|
135
|
|
|
*/ |
|
136
|
|
|
protected function loopOverAuthors(): void |
|
137
|
|
|
{ |
|
138
|
|
|
foreach ($this->searchTree->MedlineCitation->Article->AuthorList->Author as $author) { |
|
139
|
|
|
$this->createAuthorEntry($author); |
|
140
|
|
|
} |
|
141
|
|
|
} |
|
142
|
|
|
|
|
143
|
|
|
/** |
|
144
|
|
|
* Create an author entry in output. |
|
145
|
|
|
* |
|
146
|
|
|
* @param $author |
|
147
|
|
|
*/ |
|
148
|
|
|
protected function createAuthorEntry($author): void |
|
149
|
|
|
{ |
|
150
|
|
|
if (! empty($lastName = get_string($author, 'LastName'))) { |
|
151
|
|
|
$affiliations = $this->loopOverAffiliations($author); |
|
152
|
|
|
|
|
153
|
|
|
$this->resourceOutput['authors'][] = [ |
|
154
|
|
|
'first_name' => get_string($author, 'ForeName'), |
|
155
|
|
|
'last_name' => $lastName, |
|
156
|
|
|
'email' => $this->getEmailsFromAffiliations($affiliations), |
|
157
|
|
|
'affiliation' => $affiliations, |
|
158
|
|
|
]; |
|
159
|
|
|
} |
|
160
|
|
|
} |
|
161
|
|
|
|
|
162
|
|
|
/** |
|
163
|
|
|
* Loop over affiliations. |
|
164
|
|
|
* |
|
165
|
|
|
* @param array $author |
|
166
|
|
|
* @return array |
|
167
|
|
|
*/ |
|
168
|
|
|
protected function loopOverAffiliations($author): array |
|
169
|
|
|
{ |
|
170
|
|
|
$affiliations = []; |
|
171
|
|
|
foreach ($author->AffiliationInfo as $affiliation) { |
|
172
|
|
|
$affiliations[]['name'] = get_string($affiliation, 'Affiliation'); |
|
173
|
|
|
} |
|
174
|
|
|
|
|
175
|
|
|
return $affiliations; |
|
176
|
|
|
} |
|
177
|
|
|
|
|
178
|
|
|
/** |
|
179
|
|
|
* Get emails from affiliations array. |
|
180
|
|
|
* |
|
181
|
|
|
* @param array $affiliations |
|
182
|
|
|
* @return string |
|
183
|
|
|
*/ |
|
184
|
|
|
protected function getEmailsFromAffiliations($affiliations): string |
|
185
|
|
|
{ |
|
186
|
|
|
return get_string(find_emails_in_array(Arr::pluck($affiliations, 'name')), 0); |
|
187
|
|
|
} |
|
188
|
|
|
|
|
189
|
|
|
protected function getUpdateTimestamp($refSource) |
|
190
|
|
|
{ |
|
191
|
|
|
preg_match('/\s(\d{4}\s\w{3}(\s\d{1,2})?);/', $refSource, $matches); |
|
192
|
|
|
|
|
193
|
|
|
return date_from_human_readable($matches[1])->timestamp; |
|
194
|
|
|
} |
|
195
|
|
|
} |
|
196
|
|
|
|
PubPeerFoundation /
PublicationDataExtractor
