1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
namespace PPP\Wikidata\ValueParsers; |
4
|
|
|
|
5
|
|
|
use InvalidArgumentException; |
6
|
|
|
use ValueParsers\ParserOptions; |
7
|
|
|
use ValueParsers\StringValueParser; |
8
|
|
|
use ValueParsers\ValueParser; |
9
|
|
|
use Wikibase\DataModel\Entity\EntityIdValue; |
10
|
|
|
use Wikibase\DataModel\Entity\Item; |
11
|
|
|
use Wikibase\DataModel\Entity\ItemId; |
12
|
|
|
use Wikibase\DataModel\Entity\PropertyId; |
13
|
|
|
use Wikibase\DataModel\Snak\PropertyValueSnak; |
14
|
|
|
use Wikibase\DataModel\Statement\Statement; |
15
|
|
|
use Wikibase\DataModel\Term\Term; |
16
|
|
|
use Wikibase\EntityStore\EntityStore; |
17
|
|
|
|
18
|
|
|
/** |
19
|
|
|
* Try to find a Wikibase entities from a given string. |
20
|
|
|
* |
21
|
|
|
* @licence AGPLv3+ |
22
|
|
|
* @author Thomas Pellissier Tanon |
23
|
|
|
*/ |
24
|
|
|
class WikibaseEntityIdParser extends StringValueParser { |
25
|
|
|
|
26
|
|
|
const FORMAT_NAME = 'wikibase-entity'; |
27
|
|
|
|
28
|
|
|
private static $INSTANCES_TO_FILTER = array('Q4167410', 'Q17362920', 'Q4167836', 'Q13406463', 'Q11266439', 'Q14204246', 'Q21286738'); |
29
|
|
|
|
30
|
|
|
const INSTANCEOF_PID = 'P31'; |
31
|
|
|
|
32
|
|
|
/** |
33
|
|
|
* Identifier for the option that holds the type of entity the parser should looks for. |
34
|
|
|
*/ |
35
|
|
|
const OPT_ENTITY_TYPE = 'type'; |
36
|
|
|
|
37
|
|
|
/** |
38
|
|
|
* @var EntityStore |
39
|
|
|
*/ |
40
|
|
|
private $entityStore; |
41
|
|
|
|
42
|
|
|
/** |
43
|
|
|
* @param EntityStore $entityStore |
44
|
|
|
* @param ParserOptions|null $options |
45
|
|
|
*/ |
46
|
9 |
|
public function __construct(EntityStore $entityStore, ParserOptions $options = null) { |
47
|
9 |
|
$options->requireOption(self::OPT_ENTITY_TYPE); |
|
|
|
|
48
|
|
|
|
49
|
9 |
|
$this->entityStore = $entityStore; |
50
|
9 |
|
parent::__construct($options); |
51
|
9 |
|
} |
52
|
|
|
|
53
|
10 |
|
protected function stringParse($value) { |
54
|
10 |
|
if($value === '') { |
55
|
1 |
|
return array(); |
56
|
|
|
} |
57
|
|
|
|
58
|
9 |
|
return $this->entityIdsToEntityIdValues($this->getEntityIdsForValue($value)); |
59
|
|
|
} |
60
|
|
|
|
61
|
9 |
|
private function getEntityIdsForValue($value) { |
62
|
9 |
|
$languageCode = $this->getOption(ValueParser::OPT_LANG); |
63
|
9 |
|
$term = new Term($languageCode, $value); |
64
|
|
|
|
65
|
9 |
|
$entityType = $this->getOption(self::OPT_ENTITY_TYPE); |
66
|
|
|
switch($entityType) { |
67
|
9 |
|
case 'item': |
68
|
7 |
|
return $this->getItemIdsForTerm($term); |
69
|
2 |
|
case 'property': |
70
|
2 |
|
return $this->getPropertyIdsForTerm($term); |
71
|
|
|
default: |
72
|
|
|
throw new InvalidArgumentException('Unknown entity type ' . $entityType); |
73
|
|
|
} |
74
|
|
|
} |
75
|
|
|
|
76
|
7 |
|
private function getItemIdsForTerm(Term $term) { |
77
|
7 |
|
$itemIds = $this->filterItemIds( |
78
|
7 |
|
$this->entityStore->getItemIdForTermLookup()->getItemIdsForTerm($term) |
79
|
7 |
|
); |
80
|
|
|
|
81
|
|
|
try { |
82
|
7 |
|
$itemIds[] = new ItemId($term->getText()); |
83
|
7 |
|
} catch(InvalidArgumentException $e) { |
84
|
|
|
//The term is not a valid QID |
85
|
|
|
} |
86
|
|
|
|
87
|
7 |
|
return $this->sortItemIds(array_unique($itemIds)); |
88
|
|
|
} |
89
|
|
|
|
90
|
2 |
|
private function getPropertyIdsForTerm(Term $term) { |
91
|
2 |
|
$propertyIds = $this->entityStore->getPropertyIdForTermLookup()->getPropertyIdsForTerm($term); |
92
|
|
|
|
93
|
|
|
try { |
94
|
2 |
|
$propertyIds[] = new PropertyId($term->getText()); |
95
|
2 |
|
} catch(InvalidArgumentException $e) { |
96
|
|
|
//The term is not a valid PID |
97
|
|
|
} |
98
|
|
|
|
99
|
2 |
|
return array_unique($propertyIds); |
100
|
|
|
} |
101
|
|
|
|
102
|
7 |
|
private function filterItemIds(array $itemIds) { |
103
|
7 |
|
$entities = $this->entityStore->getEntityDocumentLookup()->getEntityDocumentsForIds($itemIds); |
104
|
7 |
|
$filtered = array(); |
105
|
|
|
|
106
|
7 |
|
foreach($entities as $entity) { |
107
|
6 |
|
if($entity instanceof Item && !$this->shouldItemBeIgnored($entity)) { |
108
|
6 |
|
$filtered[] = $entity->getId(); |
109
|
6 |
|
} |
110
|
7 |
|
} |
111
|
|
|
|
112
|
7 |
|
return $filtered; |
113
|
|
|
} |
114
|
|
|
|
115
|
6 |
|
private function shouldItemBeIgnored(Item $item) { |
116
|
|
|
/** @var Statement $statement */ |
117
|
6 |
|
foreach($item->getStatements()->getByPropertyId(new PropertyId(self::INSTANCEOF_PID)) as $statement) { |
118
|
6 |
|
$mainSnak = $statement->getMainSnak(); |
119
|
|
|
if( |
120
|
6 |
|
$mainSnak instanceof PropertyValueSnak && |
121
|
6 |
|
$mainSnak->getDataValue() instanceof EntityIdValue && |
122
|
6 |
|
in_array($mainSnak->getDataValue()->getEntityId()->getSerialization(), self::$INSTANCES_TO_FILTER) |
123
|
6 |
|
) { |
124
|
5 |
|
return true; |
125
|
|
|
} |
126
|
6 |
|
} |
127
|
|
|
|
128
|
6 |
|
return false; |
129
|
|
|
} |
130
|
|
|
|
131
|
9 |
|
private function entityIdsToEntityIdValues(array $entityIds) { |
132
|
9 |
|
$values = array(); |
133
|
9 |
|
foreach($entityIds as $entityId) { |
134
|
8 |
|
$values[] = new EntityIdValue($entityId); |
135
|
9 |
|
} |
136
|
|
|
|
137
|
9 |
|
return $values; |
138
|
|
|
} |
139
|
|
|
|
140
|
|
|
|
141
|
7 |
|
private function sortItemIds(array $itemIds) { |
142
|
7 |
|
usort( |
143
|
7 |
|
$itemIds, |
144
|
3 |
|
function(ItemId $a, ItemId $b) { |
145
|
3 |
|
return $a->getNumericId() - $b->getNumericId(); |
146
|
|
|
} |
147
|
7 |
|
); |
148
|
|
|
|
149
|
7 |
|
return $itemIds; |
150
|
|
|
} |
151
|
|
|
} |
152
|
|
|
|
If a variable is not always an object, we recommend to add an additional type check to ensure your method call is safe: