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 |
8
|
|
|
{ |
9
|
|
|
/** |
10
|
|
|
* Create search tree. |
11
|
|
|
*/ |
12
|
|
|
protected function getDataFromDocument() |
13
|
|
|
{ |
14
|
|
|
$this->searchTree = $this->document->{'PubmedArticle'}; |
15
|
|
|
} |
16
|
8 |
|
|
17
|
|
|
/** |
18
|
8 |
|
* Extract and format data needed for the Publication Model. |
19
|
8 |
|
*/ |
20
|
8 |
|
public function extractPublicationData() |
21
|
|
|
{ |
22
|
4 |
|
$this->output['publication'] = [ |
23
|
|
|
'title' => get_string($this->searchTree, 'MedlineCitation.Article.ArticleTitle'), |
24
|
4 |
|
'url' => (string) 'http://www.ncbi.nlm.nih.gov/pubmed/'.get_string($this->searchTree, 'MedlineCitation.PMID'), |
25
|
|
|
'published_at' => date_from_pub_date(data_get($this->searchTree, 'MedlineCitation.Article.Journal.JournalIssue.PubDate')), |
26
|
4 |
|
'abstract' => get_string($this->searchTree, 'MedlineCitation.Article.Abstract.AbstractText'), |
27
|
4 |
|
]; |
28
|
4 |
|
} |
29
|
4 |
|
|
30
|
|
|
/** |
31
|
4 |
|
* Extract and format data needed for the Identifiers Relationship |
32
|
|
|
* on the Publication Model. |
33
|
|
|
*/ |
34
|
4 |
|
public function extractIdentifiersData() |
35
|
|
|
{ |
36
|
4 |
|
foreach ($this->searchTree->PubmedData->ArticleIdList->ArticleId as $identifier) { |
37
|
4 |
|
$this->output['identifiers'][] = [ |
38
|
|
|
'value' => (string) $identifier, |
39
|
|
|
'type' => (string) $identifier['IdType'], |
40
|
|
|
]; |
41
|
|
|
} |
42
|
4 |
|
if ($value = get_string($this->searchTree, 'MedlineCitation.Article.Journal.ISSN')) { |
43
|
|
|
$this->output['identifiers'][] = [ |
44
|
4 |
|
'value' => $value, |
45
|
4 |
|
'type' => 'issn', |
46
|
4 |
|
]; |
47
|
4 |
|
} |
48
|
4 |
|
} |
49
|
|
|
|
50
|
4 |
|
/** |
51
|
|
|
* Extract and format data needed for the Journals Relationship |
52
|
|
|
* on the Publication Model. |
53
|
|
|
*/ |
54
|
|
|
public function extractJournalData() |
55
|
|
|
{ |
56
|
4 |
|
$this->output['journal'] = [ |
57
|
|
|
'title' => get_string($this->searchTree, 'MedlineCitation.Article.Journal.Title'), |
58
|
4 |
|
'issn' => $this->getIssns(), |
59
|
4 |
|
]; |
60
|
4 |
|
} |
61
|
4 |
|
|
62
|
|
|
/** |
63
|
|
|
* Extract and format data needed for the Authors Relationship |
64
|
|
|
* on the Publication Model. |
65
|
4 |
|
*/ |
66
|
4 |
|
public function extractAuthorsData() |
67
|
4 |
|
{ |
68
|
4 |
|
try { |
69
|
|
|
foreach ($this->searchTree->MedlineCitation->Article->AuthorList->Author as $author) { |
70
|
|
|
if (! empty($lastName = get_string($author, 'LastName'))) { |
71
|
4 |
|
$affiliations = []; |
72
|
|
|
foreach ($author->AffiliationInfo as $affiliation) { |
73
|
|
|
$affiliations[]['name'] = get_string($affiliation, 'Affiliation'); |
74
|
|
|
} |
75
|
|
|
|
76
|
|
|
$this->output['authors'][] = [ |
77
|
4 |
|
'first_name' => get_string($author, 'ForeName'), |
78
|
|
|
'last_name' => $lastName, |
79
|
4 |
|
'email' => get_string(find_emails_in_array(Arr::pluck($affiliations, 'name')), 0), |
80
|
|
|
'affiliation' => $affiliations, |
81
|
4 |
|
]; |
82
|
4 |
|
} |
83
|
|
|
} |
84
|
|
|
} catch (\Exception $e) { |
85
|
4 |
|
// Empty catch block, don't want anything to happen in case of exception. |
86
|
4 |
|
} |
87
|
|
|
} |
88
|
|
|
|
89
|
4 |
|
/** |
90
|
4 |
|
* Get all available ISSNs values from the tree. |
91
|
4 |
|
* |
92
|
|
|
* @return array |
93
|
4 |
|
*/ |
94
|
|
|
protected function getIssns() |
95
|
|
|
{ |
96
|
|
|
$issn = []; |
97
|
|
|
|
98
|
|
|
if ($number = get_string($this->searchTree, 'MedlineCitation.Article.Journal.ISSN')) { |
99
|
8 |
|
$issn[] = $number; |
100
|
|
|
} |
101
|
|
|
|
102
|
8 |
|
if ($number = get_string($this->searchTree, 'MedlineCitation.MedlineJournalInfo.ISSNLinking')) { |
103
|
7 |
|
$issn[] = $number; |
104
|
7 |
|
} |
105
|
7 |
|
|
106
|
7 |
|
return $issn; |
107
|
|
|
} |
108
|
|
|
} |
109
|
|
|
|