|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace SCI\DataValues; |
|
4
|
|
|
|
|
5
|
|
|
/** |
|
6
|
|
|
* @license GNU GPL v2+ |
|
7
|
|
|
* @since 1.0 |
|
8
|
|
|
* |
|
9
|
|
|
* @author mwjames |
|
10
|
|
|
*/ |
|
11
|
|
|
class ResourceIdentifierStringValueParser { |
|
12
|
|
|
|
|
13
|
|
|
/** |
|
14
|
|
|
* @var string |
|
15
|
|
|
*/ |
|
16
|
|
|
private $typeid; |
|
17
|
|
|
|
|
18
|
|
|
/** |
|
19
|
|
|
* @param string $typeid |
|
20
|
|
|
*/ |
|
21
|
52 |
|
public function __construct( $typeid ) { |
|
22
|
52 |
|
$this->typeid = $typeid; |
|
23
|
52 |
|
} |
|
24
|
|
|
|
|
25
|
|
|
/** |
|
26
|
|
|
* @since 1.0 |
|
27
|
|
|
* |
|
28
|
|
|
* @return true |
|
29
|
|
|
*/ |
|
30
|
40 |
|
public function parse( &$value ) { |
|
31
|
|
|
|
|
32
|
|
|
// Remove a possible url-prefix |
|
33
|
|
|
|
|
34
|
40 |
|
if ( $this->typeid === '_sci_doi' ) { |
|
35
|
|
|
// Remove anything before 10 as in http://dx.doi.org/10.1000/123 |
|
36
|
14 |
|
$value = substr( $value, strcspn( $value, '10' ) ); |
|
37
|
27 |
|
} elseif ( strrpos( $value, '://' ) !== false ) { |
|
38
|
|
|
// Remove anything before the last / as in http://foo/bar123 |
|
39
|
4 |
|
$value = substr( $value, strrpos( $value, '/' ) + 1 ); |
|
40
|
|
|
} |
|
41
|
|
|
|
|
42
|
|
|
// Remove other possible prefixes |
|
43
|
|
|
|
|
44
|
|
|
// http://www.doi.org/doi_handbook/2_Numbering.html#2.4 |
|
45
|
|
|
// All DOI names are converted to upper case upon registration |
|
46
|
|
|
// but since it system is case insensitive, we store it as lower |
|
47
|
|
|
// case as it "looks" better |
|
48
|
40 |
|
if ( $this->typeid === '_sci_doi' ) { |
|
49
|
14 |
|
$value = strtolower( $value ); |
|
50
|
|
|
} else { |
|
51
|
27 |
|
$value = str_replace( [ 'VIAF', 'OCLC', 'PMID' ], '', strtoupper( $value ) ); |
|
52
|
|
|
} |
|
53
|
|
|
|
|
54
|
40 |
|
return $this->canMatchValueToPattern( $value ); |
|
|
|
|
|
|
55
|
|
|
} |
|
56
|
|
|
|
|
57
|
|
|
/** |
|
58
|
|
|
* @return string |
|
59
|
|
|
*/ |
|
60
|
1 |
|
public function getCanonicalName() { |
|
61
|
|
|
|
|
62
|
1 |
|
switch ( $this->typeid ) { |
|
63
|
1 |
|
case '_sci_viaf': |
|
64
|
|
|
return 'VIAF'; |
|
65
|
1 |
|
case '_sci_oclc': |
|
66
|
|
|
return 'OLCL'; |
|
67
|
1 |
|
case '_sci_pmid': |
|
68
|
|
|
return 'PMID'; |
|
69
|
1 |
|
case '_sci_pmcid': |
|
70
|
|
|
return 'PMCID'; |
|
71
|
1 |
|
case '_sci_olid': |
|
72
|
|
|
return 'OLID'; |
|
73
|
1 |
|
case '_sci_doi': |
|
74
|
1 |
|
return 'DOI'; |
|
75
|
|
|
} |
|
76
|
|
|
|
|
77
|
|
|
return null; |
|
78
|
|
|
} |
|
79
|
|
|
|
|
80
|
|
|
/** |
|
81
|
|
|
* @return string |
|
82
|
|
|
*/ |
|
83
|
7 |
|
public function getResourceTargetUri() { |
|
84
|
|
|
|
|
85
|
7 |
|
switch ( $this->typeid ) { |
|
86
|
7 |
|
case '_sci_viaf': |
|
87
|
|
|
// http://www.oclc.org/research/activities/viaf.html |
|
88
|
|
|
// http://id.loc.gov/vocabulary/identifiers/viaf.html |
|
89
|
1 |
|
return "https://viaf.org/viaf/"; |
|
90
|
6 |
|
case '_sci_oclc': |
|
91
|
|
|
// http://www.oclc.org/support/documentation/glossary/oclc.en.html#OCLCControlNumber |
|
92
|
1 |
|
return "https://www.worldcat.org/oclc/"; |
|
93
|
5 |
|
case '_sci_pmcid': |
|
94
|
|
|
// https://www.nlm.nih.gov/pubs/techbull/nd09/nd09_pmc_urls.html |
|
95
|
1 |
|
return "https://www.ncbi.nlm.nih.gov/pmc/"; |
|
96
|
4 |
|
case '_sci_pmid': |
|
97
|
1 |
|
return "https://www.ncbi.nlm.nih.gov/pubmed/"; |
|
98
|
3 |
|
case '_sci_olid': |
|
99
|
1 |
|
return "https://openlibrary.org/books/"; |
|
100
|
2 |
|
case '_sci_doi': |
|
101
|
|
|
// https://en.wikipedia.org/wiki/Digital_object_identifier |
|
102
|
2 |
|
return "https://doi.org/"; |
|
103
|
|
|
} |
|
104
|
|
|
} |
|
105
|
|
|
|
|
106
|
40 |
|
private function canMatchValueToPattern( &$value ) { |
|
107
|
|
|
|
|
108
|
40 |
|
switch ( $this->typeid ) { |
|
109
|
40 |
|
case '_sci_viaf': |
|
110
|
35 |
|
case '_sci_oclc': |
|
111
|
29 |
|
case '_sci_pmid': |
|
112
|
16 |
|
return preg_match( "/^[0-9]*$/", $value ); |
|
113
|
25 |
|
case '_sci_pmcid': |
|
114
|
6 |
|
return preg_match( "/PMC[\d]+/", $value ); |
|
115
|
20 |
|
case '_sci_olid': |
|
116
|
7 |
|
return preg_match( "/OL[A-Z0-9]+/", $value ); |
|
117
|
14 |
|
case '_sci_doi': |
|
118
|
|
|
// http://stackoverflow.com/questions/27910/finding-a-doi-in-a-document-or-page# |
|
119
|
14 |
|
return preg_match( "/\b(10[.][0-9]{4,}(?:[.][0-9]+)*\/(?:(?![\"&\'])\S)+)\b/", $value ); |
|
120
|
|
|
} |
|
121
|
|
|
|
|
122
|
|
|
return false; |
|
123
|
|
|
} |
|
124
|
|
|
|
|
125
|
|
|
} |
|
126
|
|
|
|
SemanticMediaWiki /
SemanticCite
Loading history...