1
|
|
|
<?php declare(strict_types = 1); |
2
|
|
|
|
3
|
|
|
namespace JSKOS\RDF; |
4
|
|
|
|
5
|
|
|
use JSKOS\Resource; |
6
|
|
|
|
7
|
|
|
/** |
8
|
|
|
* Mapper maps RDF data to JSKOS based on a set of mapping rules. |
9
|
|
|
* |
10
|
|
|
* The rules define which RDF properties will be used to fill which JSKOS |
11
|
|
|
* fields. The rules further include two special keys: |
12
|
|
|
* |
13
|
|
|
* `_ns` provides a list of RDF prefixes and namespace. Namespaces are |
14
|
|
|
* applied *globally* for all mapping instances. |
15
|
|
|
* |
16
|
|
|
* `_defaultLanguage` defines the default language ("und" if not set). |
17
|
|
|
* This field can be also be used at individual mappings. |
18
|
|
|
* |
19
|
|
|
* RDF data must be provided as EasyRdf_Resource. |
20
|
|
|
* |
21
|
|
|
* @license LGPL |
22
|
|
|
* @author Jakob Voß |
23
|
|
|
*/ |
24
|
|
|
class Mapper |
25
|
|
|
{ |
26
|
|
|
/** |
27
|
|
|
* The mapping rules. |
28
|
|
|
*/ |
29
|
|
|
protected $rules; |
30
|
|
|
|
31
|
|
|
/** |
32
|
|
|
* Default language for literals without language tag. |
33
|
|
|
*/ |
34
|
|
|
protected $defaultLanguage; |
35
|
|
|
|
36
|
|
|
/** |
37
|
|
|
* Allowed JSKOS class names |
38
|
|
|
*/ |
39
|
|
|
public static $JSKOSClasses = [ |
40
|
|
|
'Concept', |
41
|
|
|
'ConceptScheme', |
42
|
|
|
'ConceptType', |
43
|
|
|
'ConceptBundle', |
44
|
|
|
'Concordance', |
45
|
|
|
'Mapping' |
46
|
|
|
]; |
47
|
|
|
|
48
|
|
|
/** |
49
|
|
|
* Create a new mapping based on rules. |
50
|
|
|
*/ |
51
|
|
|
public function __construct(array $rules = null) |
52
|
|
|
{ |
53
|
|
|
if ($rules === null) { |
54
|
|
|
$rules = json_decode(file_get_contents(__DIR__."/rdf2jskos.json"), true); |
55
|
|
|
} |
56
|
|
|
|
57
|
|
|
if (isset($rules['_ns'])) { |
58
|
|
|
foreach ($rules['_ns'] as $prefix => $namespace) { |
59
|
|
|
# TODO: warn if prefix is already defined with different namespace! |
60
|
|
|
\EasyRdf_Namespace::set($prefix, $namespace); |
61
|
|
|
} |
62
|
|
|
} |
63
|
|
|
|
64
|
|
|
if (isset($rules['_defaultLanguage'])) { |
65
|
|
|
$this->defaultLanguage = $rules['_defaultLanguage']; |
66
|
|
|
} else { |
67
|
|
|
$this->defaultLanguage = 'und'; |
68
|
|
|
} |
69
|
|
|
|
70
|
|
|
foreach ($rules as $field => $config) { |
71
|
|
|
if (substr($field, 0, 1) != '_') { |
72
|
|
|
$this->rules[$field] = $config; |
73
|
|
|
} |
74
|
|
|
} |
75
|
|
|
} |
76
|
|
|
|
77
|
|
|
/** |
78
|
|
|
* Apply mapping via extraction of data from an RDF resource and add |
79
|
|
|
* resulting data to a JSKOS Resource. |
80
|
|
|
* |
81
|
|
|
* @param EasyRdf_Resource $rdf |
|
|
|
|
82
|
|
|
* @param PrettyJsonSerializable $jskos |
|
|
|
|
83
|
|
|
*/ |
84
|
|
|
public function applyAtResource(\EasyRdf_Resource $rdf, Resource $jskos) |
85
|
|
|
{ |
86
|
|
|
#error_log($rdf->getGraph()->dump('text')); |
87
|
|
|
#error_log(print_r($this->rules,1)); |
88
|
|
|
|
89
|
|
|
foreach ($this->rules as $property => $mapping) { |
90
|
|
|
$type = $mapping['type']; |
91
|
|
|
if (isset($mapping['jskos']) && in_array($mapping['jskos'], static::$JSKOSClasses)) { |
92
|
|
|
$class = '\JSKOS\\'.$mapping['jskos']; |
93
|
|
|
} else { |
94
|
|
|
$class = null; |
95
|
|
|
} |
96
|
|
|
|
97
|
|
|
foreach ($mapping['properties'] as $rdfProperty) { |
98
|
|
|
if ($type == 'URI') { |
99
|
|
|
foreach (static::getURIs($rdf, $rdfProperty) as $uri) { |
100
|
|
|
if (isset($class)) { |
101
|
|
|
$uri = new $class(['uri'=>$uri]); |
102
|
|
|
} |
103
|
|
|
if (isset($mapping['unique'])) { |
104
|
|
|
$jskos->$property = $uri; |
105
|
|
|
} else { |
106
|
|
|
if (empty($jskos->$property)) { |
107
|
|
|
$jskos->$property = []; |
108
|
|
|
} |
109
|
|
|
$jskos->$property[] = $uri; |
110
|
|
|
} |
111
|
|
|
} |
112
|
|
|
} elseif ($type == 'literal') { |
113
|
|
|
foreach ($rdf->allLiterals($rdfProperty) as $literal) { |
114
|
|
|
$value = static::cleanString($literal); |
115
|
|
|
if (!isset($value)) { |
116
|
|
|
continue; |
117
|
|
|
} |
118
|
|
|
|
119
|
|
|
if ($literal->getLang()) { |
120
|
|
|
$language = $literal->getLang(); |
121
|
|
|
} elseif (isset($mapping['_defaultLanguage'])) { |
122
|
|
|
$language = $mapping['_defaultLanguage']; |
123
|
|
|
} else { |
124
|
|
|
$language = $this->defaultLanguage; |
125
|
|
|
} |
126
|
|
|
|
127
|
|
|
$languageMap = isset($jskos->$property) ? $jskos->$property : []; |
128
|
|
|
|
129
|
|
|
if (isset($mapping['unique'])) { |
130
|
|
|
$languageMap[$language] = $value; |
131
|
|
|
} else { |
132
|
|
|
$list = $languageMap[$language] ?? []; |
133
|
|
|
$list[] = $value; |
134
|
|
|
$languageMap[$language] = $list; |
135
|
|
|
} |
136
|
|
|
|
137
|
|
|
$jskos->$property = $languageMap; |
138
|
|
|
} |
139
|
|
|
} elseif ($type == 'plain') { |
140
|
|
|
foreach ($rdf->allLiterals($rdfProperty) as $literal) { |
141
|
|
|
$value = static::cleanString($literal); |
142
|
|
|
if (!isset($value)) { |
143
|
|
|
continue; |
144
|
|
|
} |
145
|
|
|
|
146
|
|
|
if (isset($mapping['pattern']) && !preg_match($mapping['pattern'], $value)) { |
147
|
|
|
continue; |
148
|
|
|
} |
149
|
|
|
if (isset($mapping['unique'])) { |
150
|
|
|
$jskos->$property = $value; |
151
|
|
|
} else { |
152
|
|
|
if (empty($jskos->$property)) { |
153
|
|
|
$jskos->$property = [$value]; |
154
|
|
|
} else { |
155
|
|
|
$jskos->$property[] = $value; |
156
|
|
|
} |
157
|
|
|
} |
158
|
|
|
} |
159
|
|
|
} |
160
|
|
|
} |
161
|
|
|
} |
162
|
|
|
} |
163
|
|
|
|
164
|
|
|
/** |
165
|
|
|
* Clean up a string value by trimming whitespace and mapping empty strings to null. |
166
|
|
|
*/ |
167
|
|
|
public static function cleanString($string) |
168
|
|
|
{ |
169
|
|
|
$string = trim((string)$string); |
170
|
|
|
return $string !== "" ? $string : null; |
171
|
|
|
} |
172
|
|
|
|
173
|
|
|
/** |
174
|
|
|
* Silently try to load RDF from an URL. |
175
|
|
|
* @return EasyRdf_Resource|null |
176
|
|
|
*/ |
177
|
|
|
public static function loadRDF($url, $uri = null, $format = null, $forceHttp = false) |
178
|
|
|
{ |
179
|
|
|
try { |
180
|
|
|
$rdf = \EasyRdf_Graph::newAndLoad($url, $format); |
181
|
|
|
if ($forceHttp) { |
182
|
|
|
$turtle = $rdf->serialise('turtle'); |
183
|
|
|
$turtle = str_replace("<https://","<http://", $turtle); |
184
|
|
|
$graph = new \EasyRdf_Graph(); |
185
|
|
|
$graph->parse($turtle, 'turtle'); |
186
|
|
|
$rdf = $graph; |
187
|
|
|
} |
188
|
|
|
return $rdf->resource($uri ? $uri : $url); |
189
|
|
|
} catch (\Exception $e) { |
190
|
|
|
return; |
191
|
|
|
} |
192
|
|
|
} |
193
|
|
|
|
194
|
|
|
/** |
195
|
|
|
* Get a list of RDF object URIs. |
196
|
|
|
*/ |
197
|
|
|
public static function getURIs(\EasyRDF_Resource $subject, $predicate, $pattern = null) |
198
|
|
|
{ |
199
|
|
|
$uris = []; |
200
|
|
|
foreach ($subject->allResources($predicate) as $object) { |
201
|
|
|
if (!$object->isBNode()) { |
202
|
|
|
$object = $object->getUri(); |
203
|
|
|
if (!$pattern or preg_match($pattern, $object)) { |
204
|
|
|
$uris[] = $object; |
205
|
|
|
} |
206
|
|
|
} |
207
|
|
|
} |
208
|
|
|
return $uris; |
209
|
|
|
} |
210
|
|
|
} |
211
|
|
|
|