1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
namespace Tests\Wikibase\JsonDumpReader; |
4
|
|
|
|
5
|
|
|
use DataValues\Deserializers\DataValueDeserializer; |
6
|
|
|
use Iterator; |
7
|
|
|
use Wikibase\DataModel\DeserializerFactory; |
8
|
|
|
use Wikibase\DataModel\Entity\BasicEntityIdParser; |
9
|
|
|
use Wikibase\JsonDumpReader\Iterator\EntityDumpIterator; |
10
|
|
|
use Wikibase\JsonDumpReader\JsonDumpFactory; |
11
|
|
|
use Wikibase\JsonDumpReader\Reader\ExtractedDumpReader; |
12
|
|
|
|
13
|
|
|
/** |
14
|
|
|
* @covers Wikibase\JsonDumpReader\Iterator\EntityDumpIterator |
15
|
|
|
* @covers Wikibase\JsonDumpReader\JsonDumpFactory |
16
|
|
|
* |
17
|
|
|
* @licence GNU GPL v2+ |
18
|
|
|
* @author Jeroen De Dauw < [email protected] > |
19
|
|
|
*/ |
20
|
|
|
class EntityDumpIteratorTest extends \PHPUnit_Framework_TestCase { |
21
|
|
|
|
22
|
|
|
private function newIteratorForFile( $filePath, callable $onError = null ) { |
23
|
|
|
return ( new JsonDumpFactory() )->newEntityDumpIterator( |
24
|
|
|
new ExtractedDumpReader( $filePath ), |
25
|
|
|
$this->newCurrentEntityDeserializer(), |
26
|
|
|
$onError |
27
|
|
|
); |
28
|
|
|
} |
29
|
|
|
|
30
|
|
|
private function newCurrentEntityDeserializer() { |
31
|
|
|
$factory = new DeserializerFactory( |
32
|
|
|
$this->newDataValueDeserializer(), |
33
|
|
|
new BasicEntityIdParser() |
34
|
|
|
); |
35
|
|
|
|
36
|
|
|
return $factory->newEntityDeserializer(); |
37
|
|
|
} |
38
|
|
|
|
39
|
|
|
private function newDataValueDeserializer() { |
40
|
|
|
$dataValueClasses = [ |
41
|
|
|
'boolean' => 'DataValues\BooleanValue', |
42
|
|
|
'number' => 'DataValues\NumberValue', |
43
|
|
|
'string' => 'DataValues\StringValue', |
44
|
|
|
'unknown' => 'DataValues\UnknownValue', |
45
|
|
|
'globecoordinate' => 'DataValues\Geo\Values\GlobeCoordinateValue', |
46
|
|
|
'monolingualtext' => 'DataValues\MonolingualTextValue', |
47
|
|
|
'multilingualtext' => 'DataValues\MultilingualTextValue', |
48
|
|
|
'quantity' => 'DataValues\QuantityValue', |
49
|
|
|
'time' => 'DataValues\TimeValue', |
50
|
|
|
'wikibase-entityid' => 'Wikibase\DataModel\Entity\EntityIdValue', |
51
|
|
|
]; |
52
|
|
|
|
53
|
|
|
return new DataValueDeserializer( $dataValueClasses ); |
54
|
|
|
} |
55
|
|
|
|
56
|
|
|
private function assertFindsEntities( array $expectedIds, Iterator $dumpIterator, $message = '' ) { |
57
|
|
|
$actualIds = []; |
58
|
|
|
|
59
|
|
|
foreach ( $dumpIterator as $entity ) { |
60
|
|
|
$actualIds[] = $entity->getId()->getSerialization(); |
61
|
|
|
} |
62
|
|
|
|
63
|
|
|
$this->assertEquals( $expectedIds, $actualIds, $message ); |
64
|
|
|
} |
65
|
|
|
|
66
|
|
|
public function testGivenFileWithNoEntities_noEntitiesAreReturned() { |
67
|
|
|
$iterator = $this->newIteratorForFile( ( new \JsonDumpData() )->getEmptyDumpPath() ); |
68
|
|
|
|
69
|
|
|
$this->assertFindsEntities( [], $iterator ); |
70
|
|
|
} |
71
|
|
|
|
72
|
|
|
public function testGivenFileWithOneEntity_oneEntityIsFound() { |
73
|
|
|
$iterator = $this->newIteratorForFile( ( new \JsonDumpData() )->getOneItemDumpPath() ); |
74
|
|
|
|
75
|
|
|
$this->assertFindsEntities( [ 'Q1' ], $iterator ); |
76
|
|
|
} |
77
|
|
|
|
78
|
|
|
public function testGivenFileWithFiveEntities_fiveEntityAreFound() { |
79
|
|
|
$iterator = $this->newIteratorForFile( ( new \JsonDumpData() )->getFiveEntitiesDumpPath() ); |
80
|
|
|
|
81
|
|
|
$this->assertFindsEntities( [ 'Q1', 'Q8', 'P16', 'P19', 'P22' ], $iterator ); |
82
|
|
|
} |
83
|
|
|
|
84
|
|
|
public function testGivenFileWithInvalidEntity_noEntityIsFound() { |
85
|
|
|
$iterator = $this->newIteratorForFile( __DIR__ . '/../data/invalid-item.json' ); |
86
|
|
|
$this->assertFindsEntities( [], $iterator ); |
87
|
|
|
} |
88
|
|
|
|
89
|
|
|
public function testGivenFileWithInvalidEntities_validEntitiesAreFound() { |
90
|
|
|
$iterator = $this->newIteratorForFile( __DIR__ . '/../data/3valid-2invalid.json' ); |
91
|
|
|
$this->assertFindsEntities( [ 'Q1', 'P16', 'P22' ], $iterator ); |
92
|
|
|
} |
93
|
|
|
|
94
|
|
|
public function testCanDoMultipleIterations() { |
95
|
|
|
$iterator = $this->newIteratorForFile( ( new \JsonDumpData() )->getFiveEntitiesDumpPath() ); |
96
|
|
|
|
97
|
|
|
$this->assertFindsEntities( [ 'Q1', 'Q8', 'P16', 'P19', 'P22' ], $iterator, 'first iteration' ); |
98
|
|
|
$this->assertFindsEntities( [ 'Q1', 'Q8', 'P16', 'P19', 'P22' ], $iterator, 'second iteration' ); |
99
|
|
|
} |
100
|
|
|
|
101
|
|
|
public function testInitialPosition() { |
102
|
|
|
$reader = new ExtractedDumpReader( ( new \JsonDumpData() )->getFiveEntitiesDumpPath() ); |
103
|
|
|
|
104
|
|
|
$iterator = new EntityDumpIterator( |
105
|
|
|
( new JsonDumpFactory() )->newObjectDumpIterator( $reader ), |
106
|
|
|
$this->newCurrentEntityDeserializer() |
107
|
|
|
); |
108
|
|
|
|
109
|
|
|
$iterator->rewind(); |
110
|
|
|
$this->assertSame( 'Q1', $iterator->current()->getId()->getSerialization() ); |
111
|
|
|
|
112
|
|
|
$iterator->next(); |
113
|
|
|
$this->assertSame( 'Q8', $iterator->current()->getId()->getSerialization() ); |
114
|
|
|
|
115
|
|
|
$newReader = new ExtractedDumpReader( |
116
|
|
|
( new \JsonDumpData() )->getFiveEntitiesDumpPath(), |
117
|
|
|
$reader->getPosition() |
118
|
|
|
); |
119
|
|
|
|
120
|
|
|
$newIterator = new EntityDumpIterator( |
121
|
|
|
( new JsonDumpFactory() )->newObjectDumpIterator( $newReader ), |
122
|
|
|
$this->newCurrentEntityDeserializer() |
123
|
|
|
); |
124
|
|
|
|
125
|
|
|
$this->assertFindsEntities( [ 'P16', 'P19', 'P22' ], $newIterator ); |
126
|
|
|
} |
127
|
|
|
|
128
|
|
|
public function testGivenFileWithInvalidEntities_errorsAreReported() { |
129
|
|
|
$errors = []; |
130
|
|
|
|
131
|
|
|
$iterator = $this->newIteratorForFile( |
132
|
|
|
__DIR__ . '/../data/3valid-2invalid.json', |
133
|
|
|
function( $errorMessage ) use ( &$errors ) { |
134
|
|
|
$errors[] = $errorMessage; |
135
|
|
|
} |
136
|
|
|
); |
137
|
|
|
|
138
|
|
|
$iterator->rewind(); |
139
|
|
|
while ( $iterator->valid() ) { |
140
|
|
|
$iterator->next(); |
141
|
|
|
} |
142
|
|
|
|
143
|
|
|
$this->assertContainsOnly( 'string', $errors ); |
144
|
|
|
$this->assertCount( 2, $errors ); |
145
|
|
|
} |
146
|
|
|
|
147
|
|
|
public function testGivenFileWithInvalidJsonLine_errorIsRecorded() { |
148
|
|
|
$errors = []; |
149
|
|
|
|
150
|
|
|
$iterator = $this->newIteratorForFile( |
151
|
|
|
__DIR__ . '/../data/invalid-json.json', |
152
|
|
|
function( $errorMessage ) use ( &$errors ) { |
153
|
|
|
$errors[] = $errorMessage; |
154
|
|
|
} |
155
|
|
|
); |
156
|
|
|
|
157
|
|
|
iterator_to_array( $iterator ); |
158
|
|
|
|
159
|
|
|
$this->assertContainsOnly( 'string', $errors ); |
160
|
|
|
$this->assertCount( 1, $errors ); |
161
|
|
|
} |
162
|
|
|
|
163
|
|
|
} |