|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
declare( strict_types = 1 ); |
|
4
|
|
|
|
|
5
|
|
|
namespace Tests\Wikibase\JsonDumpReader; |
|
6
|
|
|
|
|
7
|
|
|
use DataValues\BooleanValue; |
|
8
|
|
|
use DataValues\Deserializers\DataValueDeserializer; |
|
9
|
|
|
use DataValues\Geo\Values\GlobeCoordinateValue; |
|
10
|
|
|
use DataValues\MonolingualTextValue; |
|
11
|
|
|
use DataValues\MultilingualTextValue; |
|
12
|
|
|
use DataValues\NumberValue; |
|
13
|
|
|
use DataValues\QuantityValue; |
|
14
|
|
|
use DataValues\StringValue; |
|
15
|
|
|
use DataValues\TimeValue; |
|
16
|
|
|
use DataValues\UnknownValue; |
|
17
|
|
|
use Iterator; |
|
18
|
|
|
use PHPUnit\Framework\TestCase; |
|
19
|
|
|
use Wikibase\DataModel\DeserializerFactory; |
|
20
|
|
|
use Wikibase\DataModel\Entity\BasicEntityIdParser; |
|
21
|
|
|
use Wikibase\DataModel\Entity\EntityIdValue; |
|
22
|
|
|
use Wikibase\JsonDumpReader\Iterator\EntityDumpIterator; |
|
23
|
|
|
use Wikibase\JsonDumpReader\JsonDumpFactory; |
|
24
|
|
|
use Wikibase\JsonDumpReader\Reader\ExtractedDumpReader; |
|
25
|
|
|
|
|
26
|
|
|
/** |
|
27
|
|
|
* @covers \Wikibase\JsonDumpReader\Iterator\EntityDumpIterator |
|
28
|
|
|
* @covers \Wikibase\JsonDumpReader\JsonDumpFactory |
|
29
|
|
|
* |
|
30
|
|
|
* @licence GNU GPL v2+ |
|
31
|
|
|
* @author Jeroen De Dauw < [email protected] > |
|
32
|
|
|
*/ |
|
33
|
|
|
class EntityDumpIteratorTest extends TestCase { |
|
34
|
|
|
|
|
35
|
|
|
private function newIteratorForFile( $filePath, callable $onError = null ) { |
|
36
|
|
|
return ( new JsonDumpFactory() )->newEntityDumpIterator( |
|
37
|
|
|
new ExtractedDumpReader( $filePath ), |
|
38
|
|
|
$this->newCurrentEntityDeserializer(), |
|
39
|
|
|
$onError |
|
40
|
|
|
); |
|
41
|
|
|
} |
|
42
|
|
|
|
|
43
|
|
|
private function newCurrentEntityDeserializer() { |
|
44
|
|
|
$factory = new DeserializerFactory( |
|
45
|
|
|
$this->newDataValueDeserializer(), |
|
46
|
|
|
new BasicEntityIdParser() |
|
47
|
|
|
); |
|
48
|
|
|
|
|
49
|
|
|
return $factory->newEntityDeserializer(); |
|
50
|
|
|
} |
|
51
|
|
|
|
|
52
|
|
|
private function newDataValueDeserializer() { |
|
53
|
|
|
$dataValueClasses = [ |
|
54
|
|
|
'boolean' => BooleanValue::class, |
|
55
|
|
|
'number' => NumberValue::class,SeekableDumpReader |
|
56
|
|
|
'string' => StringValue::class, |
|
|
|
|
|
|
57
|
|
|
'unknown' => UnknownValue::class, |
|
58
|
|
|
'globecoordinate' => GlobeCoordinateValue::class, |
|
59
|
|
|
'monolingualtext' => MonolingualTextValue::class, |
|
60
|
|
|
'multilingualtext' => MultilingualTextValue::class, |
|
61
|
|
|
'quantity' => QuantityValue::class, |
|
62
|
|
|
'time' => TimeValue::class, |
|
63
|
|
|
'wikibase-entityid' => EntityIdValue::class, |
|
64
|
|
|
]; |
|
65
|
|
|
|
|
66
|
|
|
return new DataValueDeserializer( $dataValueClasses ); |
|
67
|
|
|
} |
|
68
|
|
|
|
|
69
|
|
|
private function assertFindsEntities( array $expectedIds, Iterator $dumpIterator, $message = '' ) { |
|
70
|
|
|
$actualIds = []; |
|
71
|
|
|
|
|
72
|
|
|
foreach ( $dumpIterator as $entity ) { |
|
73
|
|
|
$actualIds[] = $entity->getId()->getSerialization(); |
|
74
|
|
|
} |
|
75
|
|
|
|
|
76
|
|
|
$this->assertEquals( $expectedIds, $actualIds, $message ); |
|
77
|
|
|
} |
|
78
|
|
|
|
|
79
|
|
|
public function testGivenFileWithNoEntities_noEntitiesAreReturned() { |
|
80
|
|
|
$iterator = $this->newIteratorForFile( ( new \JsonDumpData() )->getEmptyDumpPath() ); |
|
81
|
|
|
|
|
82
|
|
|
$this->assertFindsEntities( [], $iterator ); |
|
83
|
|
|
} |
|
84
|
|
|
|
|
85
|
|
|
public function testGivenFileWithOneEntity_oneEntityIsFound() { |
|
86
|
|
|
$iterator = $this->newIteratorForFile( ( new \JsonDumpData() )->getOneItemDumpPath() ); |
|
87
|
|
|
|
|
88
|
|
|
$this->assertFindsEntities( [ 'Q1' ], $iterator ); |
|
89
|
|
|
} |
|
90
|
|
|
|
|
91
|
|
|
public function testGivenFileWithFiveEntities_fiveEntityAreFound() { |
|
92
|
|
|
$iterator = $this->newIteratorForFile( ( new \JsonDumpData() )->getFiveEntitiesDumpPath() ); |
|
93
|
|
|
|
|
94
|
|
|
$this->assertFindsEntities( [ 'Q1', 'Q8', 'P16', 'P19', 'P22' ], $iterator ); |
|
95
|
|
|
} |
|
96
|
|
|
|
|
97
|
|
|
public function testGivenFileWithInvalidEntity_noEntityIsFound() { |
|
98
|
|
|
$iterator = $this->newIteratorForFile( __DIR__ . '/../data/invalid-item.json' ); |
|
99
|
|
|
$this->assertFindsEntities( [], $iterator ); |
|
100
|
|
|
} |
|
101
|
|
|
|
|
102
|
|
|
public function testGivenFileWithInvalidEntities_validEntitiesAreFound() { |
|
103
|
|
|
$iterator = $this->newIteratorForFile( __DIR__ . '/../data/3valid-2invalid.json' ); |
|
104
|
|
|
$this->assertFindsEntities( [ 'Q1', 'P16', 'P22' ], $iterator ); |
|
105
|
|
|
} |
|
106
|
|
|
|
|
107
|
|
|
public function testCanDoMultipleIterations() { |
|
108
|
|
|
$iterator = $this->newIteratorForFile( ( new \JsonDumpData() )->getFiveEntitiesDumpPath() ); |
|
109
|
|
|
|
|
110
|
|
|
$this->assertFindsEntities( [ 'Q1', 'Q8', 'P16', 'P19', 'P22' ], $iterator, 'first iteration' ); |
|
111
|
|
|
$this->assertFindsEntities( [ 'Q1', 'Q8', 'P16', 'P19', 'P22' ], $iterator, 'second iteration' ); |
|
112
|
|
|
} |
|
113
|
|
|
|
|
114
|
|
|
public function testInitialPosition() { |
|
115
|
|
|
$reader = new ExtractedDumpReader( ( new \JsonDumpData() )->getFiveEntitiesDumpPath() ); |
|
116
|
|
|
|
|
117
|
|
|
$iterator = new EntityDumpIterator( |
|
118
|
|
|
( new JsonDumpFactory() )->newObjectDumpIterator( $reader ), |
|
119
|
|
|
$this->newCurrentEntityDeserializer() |
|
120
|
|
|
); |
|
121
|
|
|
|
|
122
|
|
|
$iterator->rewind(); |
|
123
|
|
|
$this->assertSame( 'Q1', $iterator->current()->getId()->getSerialization() ); |
|
124
|
|
|
|
|
125
|
|
|
$iterator->next(); |
|
126
|
|
|
$this->assertSame( 'Q8', $iterator->current()->getId()->getSerialization() ); |
|
127
|
|
|
|
|
128
|
|
|
$newReader = new ExtractedDumpReader( |
|
129
|
|
|
( new \JsonDumpData() )->getFiveEntitiesDumpPath(), |
|
130
|
|
|
$reader->getPosition() |
|
131
|
|
|
); |
|
132
|
|
|
|
|
133
|
|
|
$newIterator = new EntityDumpIterator( |
|
134
|
|
|
( new JsonDumpFactory() )->newObjectDumpIterator( $newReader ), |
|
135
|
|
|
$this->newCurrentEntityDeserializer() |
|
136
|
|
|
); |
|
137
|
|
|
|
|
138
|
|
|
$this->assertFindsEntities( [ 'P16', 'P19', 'P22' ], $newIterator ); |
|
139
|
|
|
} |
|
140
|
|
|
|
|
141
|
|
|
public function testGivenFileWithInvalidEntities_errorsAreReported() { |
|
142
|
|
|
$errors = []; |
|
143
|
|
|
|
|
144
|
|
|
$iterator = $this->newIteratorForFile( |
|
145
|
|
|
__DIR__ . '/../data/3valid-2invalid.json', |
|
146
|
|
|
function( $errorMessage ) use ( &$errors ) { |
|
147
|
|
|
$errors[] = $errorMessage; |
|
148
|
|
|
} |
|
149
|
|
|
); |
|
150
|
|
|
|
|
151
|
|
|
$iterator->rewind(); |
|
152
|
|
|
while ( $iterator->valid() ) { |
|
153
|
|
|
$iterator->next(); |
|
154
|
|
|
} |
|
155
|
|
|
|
|
156
|
|
|
$this->assertContainsOnly( 'string', $errors ); |
|
157
|
|
|
$this->assertCount( 2, $errors ); |
|
158
|
|
|
} |
|
159
|
|
|
|
|
160
|
|
|
public function testGivenFileWithInvalidJsonLine_errorIsRecorded() { |
|
161
|
|
|
$errors = []; |
|
162
|
|
|
|
|
163
|
|
|
$iterator = $this->newIteratorForFile( |
|
164
|
|
|
__DIR__ . '/../data/invalid-json.json', |
|
165
|
|
|
function( $errorMessage ) use ( &$errors ) { |
|
166
|
|
|
$errors[] = $errorMessage; |
|
167
|
|
|
} |
|
168
|
|
|
); |
|
169
|
|
|
|
|
170
|
|
|
iterator_to_array( $iterator ); |
|
171
|
|
|
|
|
172
|
|
|
$this->assertContainsOnly( 'string', $errors ); |
|
173
|
|
|
$this->assertCount( 1, $errors ); |
|
174
|
|
|
} |
|
175
|
|
|
|
|
176
|
|
|
} |