|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Queryr\Replicator; |
|
4
|
|
|
|
|
5
|
|
|
use DataValues\BooleanValue; |
|
6
|
|
|
use DataValues\Deserializers\DataValueDeserializer; |
|
7
|
|
|
use DataValues\Geo\Values\GlobeCoordinateValue; |
|
8
|
|
|
use DataValues\MonolingualTextValue; |
|
9
|
|
|
use DataValues\MultilingualTextValue; |
|
10
|
|
|
use DataValues\NumberValue; |
|
11
|
|
|
use DataValues\QuantityValue; |
|
12
|
|
|
use DataValues\Serializers\DataValueSerializer; |
|
13
|
|
|
use DataValues\StringValue; |
|
14
|
|
|
use DataValues\TimeValue; |
|
15
|
|
|
use DataValues\UnknownValue; |
|
16
|
|
|
use Doctrine\DBAL\Connection; |
|
17
|
|
|
use Doctrine\DBAL\DBALException; |
|
18
|
|
|
use Doctrine\DBAL\DriverManager; |
|
19
|
|
|
use Monolog\Formatter\JsonFormatter; |
|
20
|
|
|
use Monolog\Formatter\LineFormatter; |
|
21
|
|
|
use Monolog\Handler\BufferHandler; |
|
22
|
|
|
use Monolog\Handler\StreamHandler; |
|
23
|
|
|
use Monolog\Logger; |
|
24
|
|
|
use Psr\Log\LoggerInterface; |
|
25
|
|
|
use Psr\Log\NullLogger; |
|
26
|
|
|
use Queryr\EntityStore\EntityStoreConfig; |
|
27
|
|
|
use Queryr\EntityStore\EntityStoreFactory; |
|
28
|
|
|
use Queryr\EntityStore\EntityStoreInstaller; |
|
29
|
|
|
use Queryr\Replicator\Importer\CompositeReporter; |
|
30
|
|
|
use Queryr\Replicator\Importer\EntityHandlers\EntityStoreEntityHandler; |
|
31
|
|
|
use Queryr\Replicator\Importer\EntityHandlers\QueryEngineEntityHandler; |
|
32
|
|
|
use Queryr\Replicator\Importer\EntityHandlers\TermStoreEntityHandler; |
|
33
|
|
|
use Queryr\Replicator\Importer\LoggingReporter; |
|
34
|
|
|
use Queryr\Replicator\Importer\PageImporter; |
|
35
|
|
|
use Queryr\Replicator\Importer\PageImportReporter; |
|
36
|
|
|
use Queryr\Replicator\Importer\PagesImporter; |
|
37
|
|
|
use Queryr\Replicator\Importer\StatsReporter; |
|
38
|
|
|
use Queryr\Replicator\Model\EntityPage; |
|
39
|
|
|
use Queryr\TermStore\TermStoreConfig; |
|
40
|
|
|
use Queryr\TermStore\TermStoreInstaller; |
|
41
|
|
|
use Queryr\TermStore\TermStoreWriter; |
|
42
|
|
|
use RuntimeException; |
|
43
|
|
|
use Wikibase\DataModel\Entity\BasicEntityIdParser; |
|
44
|
|
|
use Wikibase\DataModel\Entity\EntityIdValue; |
|
45
|
|
|
use Wikibase\DataModel\SerializerFactory; |
|
46
|
|
|
use Wikibase\InternalSerialization\DeserializerFactory; |
|
47
|
|
|
use Wikibase\JsonDumpReader\DumpReader; |
|
48
|
|
|
use Wikibase\JsonDumpReader\JsonDumpFactory; |
|
49
|
|
|
use Wikibase\QueryEngine\SQLStore\DataValueHandlersBuilder; |
|
50
|
|
|
use Wikibase\QueryEngine\SQLStore\SQLStore; |
|
51
|
|
|
use Wikibase\QueryEngine\SQLStore\StoreConfig; |
|
52
|
|
|
use Wikibase\QueryEngine\SQLStore\StoreSchema; |
|
53
|
|
|
|
|
54
|
|
|
/** |
|
55
|
|
|
* @licence GNU GPL v2+ |
|
56
|
|
|
* @author Jeroen De Dauw < [email protected] > |
|
57
|
|
|
* |
|
58
|
|
|
* @SuppressWarnings(PHPMD) |
|
59
|
|
|
*/ |
|
60
|
|
|
class ServiceFactory { |
|
61
|
|
|
|
|
62
|
|
|
const QUERY_ENGINE_PREFIX = 'qe_'; |
|
63
|
|
|
const ENTITY_STORE_PREFIX = 'es_'; |
|
64
|
|
|
const TERMS_STORE_PREFIX = 'ts_'; |
|
65
|
|
|
|
|
66
|
1 |
|
public static function newFromConnection( Connection $connection ) { |
|
67
|
1 |
|
return new self( $connection ); |
|
68
|
|
|
} |
|
69
|
|
|
|
|
70
|
|
|
/** |
|
71
|
|
|
* @throws RuntimeException |
|
72
|
|
|
*/ |
|
73
|
|
|
public static function newFromConfig(): self { |
|
74
|
|
|
$config = DatabaseConfigFile::newInstance()->read(); |
|
75
|
|
|
|
|
76
|
|
|
try { |
|
77
|
|
|
$connection = DriverManager::getConnection( $config ); |
|
78
|
|
|
} |
|
79
|
|
|
catch ( DBALException $ex ) { |
|
80
|
|
|
throw new RuntimeException( |
|
81
|
|
|
'Could not establish database connection: ' . $ex->getMessage() |
|
82
|
|
|
); |
|
83
|
|
|
} |
|
84
|
|
|
|
|
85
|
|
|
$factory = new self( $connection ); |
|
86
|
|
|
$factory->setLogger( $factory->newProductionLogger() ); |
|
87
|
|
|
|
|
88
|
|
|
return $factory; |
|
89
|
|
|
} |
|
90
|
|
|
|
|
91
|
|
|
/** |
|
92
|
|
|
* @var Connection |
|
93
|
|
|
*/ |
|
94
|
|
|
private $connection; |
|
95
|
|
|
|
|
96
|
|
|
/** |
|
97
|
|
|
* @var LoggerInterface |
|
98
|
|
|
*/ |
|
99
|
|
|
private $logger; |
|
100
|
|
|
|
|
101
|
1 |
|
private function __construct( Connection $connection ) { |
|
102
|
1 |
|
$this->connection = $connection; |
|
103
|
1 |
|
$this->logger = new NullLogger(); |
|
104
|
1 |
|
} |
|
105
|
|
|
|
|
106
|
|
|
public function newProductionLogger() { |
|
107
|
|
|
$logger = new Logger( 'Replicator logger' ); |
|
108
|
|
|
|
|
109
|
|
|
$streamHandler = new StreamHandler( $this->newLoggerPath( ( new \DateTime() )->format( 'Y-m-d\TH:i:s\Z' ) ) ); |
|
110
|
|
|
$bufferHandler = new BufferHandler( $streamHandler, 500, Logger::INFO, true, true ); |
|
111
|
|
|
$streamHandler->setFormatter( new LineFormatter( "%message%\n" ) ); |
|
112
|
|
|
$logger->pushHandler( $bufferHandler ); |
|
113
|
|
|
|
|
114
|
|
|
$errorHandler = new StreamHandler( $this->newLoggerPath( 'error' ), Logger::ERROR ); |
|
115
|
|
|
$errorHandler->setFormatter( new JsonFormatter() ); |
|
116
|
|
|
$logger->pushHandler( $errorHandler ); |
|
117
|
|
|
|
|
118
|
|
|
return $logger; |
|
119
|
|
|
} |
|
120
|
|
|
|
|
121
|
|
|
private function newLoggerPath( $fileName ) { |
|
122
|
|
|
return __DIR__ . '/../var/log/' . $fileName . '.log'; |
|
123
|
|
|
} |
|
124
|
|
|
|
|
125
|
|
|
public function setLogger( LoggerInterface $logger ) { |
|
126
|
|
|
$this->logger = $logger; |
|
127
|
|
|
} |
|
128
|
|
|
|
|
129
|
|
|
public function newQueryEngineInstaller() { |
|
130
|
|
|
$sqlStore = $this->newSqlStore(); |
|
131
|
|
|
return $sqlStore->newInstaller( $this->connection->getSchemaManager() ); |
|
132
|
|
|
} |
|
133
|
|
|
|
|
134
|
|
|
public function newQueryEngineUninstaller() { |
|
135
|
|
|
$sqlStore = $this->newSqlStore(); |
|
136
|
|
|
return $sqlStore->newUninstaller( $this->connection->getSchemaManager() ); |
|
137
|
|
|
} |
|
138
|
|
|
|
|
139
|
|
|
private function newSqlStore() { |
|
140
|
|
|
$handlers = new DataValueHandlersBuilder(); |
|
141
|
|
|
|
|
142
|
|
|
$schema = new StoreSchema( |
|
143
|
|
|
self::QUERY_ENGINE_PREFIX, |
|
144
|
|
|
$handlers->withSimpleMainSnakHandlers()->getHandlers() |
|
145
|
|
|
); |
|
146
|
|
|
|
|
147
|
|
|
$config = new StoreConfig( 'QueryR Replicator QueryEngine' ); |
|
148
|
|
|
|
|
149
|
|
|
return new SQLStore( $schema, $config ); |
|
150
|
|
|
} |
|
151
|
|
|
|
|
152
|
1 |
|
public function newEntityStoreInstaller() { |
|
153
|
1 |
|
return new EntityStoreInstaller( |
|
154
|
1 |
|
$this->connection->getSchemaManager(), |
|
155
|
1 |
|
new EntityStoreConfig( self::ENTITY_STORE_PREFIX ) |
|
156
|
|
|
); |
|
157
|
|
|
} |
|
158
|
|
|
|
|
159
|
|
|
public function newPagesImporter( PageImportReporter $reporter, |
|
160
|
|
|
StatsReporter $statsReporter, callable $onAborted = null ) { |
|
161
|
|
|
|
|
162
|
|
|
return new PagesImporter( |
|
163
|
|
|
$this->newPageImporter( $reporter ), |
|
164
|
|
|
$statsReporter, |
|
165
|
|
|
$onAborted |
|
166
|
|
|
); |
|
167
|
|
|
} |
|
168
|
|
|
|
|
169
|
|
|
public function newJsonEntityPageIterator( DumpReader $reader ) { |
|
170
|
|
|
$entityPageIterator = function( \Iterator $dumpIterator ) { |
|
171
|
|
|
foreach ( $dumpIterator as $entity ) { |
|
172
|
|
|
yield new EntityPage( |
|
173
|
|
|
json_encode( $entity ), |
|
174
|
|
|
$entity['type'] === 'property' ? 'Property:' . $entity['id'] : $entity['id'], |
|
175
|
|
|
0, |
|
176
|
|
|
0, |
|
177
|
|
|
( new \DateTime() )->format( 'Y-m-d\TH:i:s\Z' ) |
|
178
|
|
|
); |
|
179
|
|
|
} |
|
180
|
|
|
}; |
|
181
|
|
|
|
|
182
|
|
|
$dumpIterator = ( new JsonDumpFactory() )->newObjectDumpIterator( $reader ); |
|
183
|
|
|
return $entityPageIterator( $dumpIterator ); |
|
184
|
|
|
} |
|
185
|
|
|
|
|
186
|
|
|
public function newPageImporter( PageImportReporter $reporter ) { |
|
187
|
|
|
$loggingReporter = new LoggingReporter( $this->logger ); |
|
188
|
|
|
$compositeReporter = new CompositeReporter( $loggingReporter, $reporter ); |
|
189
|
|
|
|
|
190
|
|
|
return new PageImporter( |
|
191
|
|
|
$this->newLegacyEntityDeserializer(), |
|
192
|
|
|
$this->getEntityHandlers(), |
|
193
|
|
|
[ |
|
194
|
|
|
new EntityStoreEntityHandler( $this->newEntityStore() ) |
|
195
|
|
|
], |
|
196
|
|
|
$compositeReporter |
|
197
|
|
|
); |
|
198
|
|
|
} |
|
199
|
|
|
|
|
200
|
|
|
private function getEntityHandlers() { |
|
201
|
|
|
$handlers = [ |
|
202
|
|
|
new TermStoreEntityHandler( $this->newTermStoreWriter() ) |
|
203
|
|
|
]; |
|
204
|
|
|
|
|
205
|
|
|
if ( defined( 'WIKIBASE_QUERYENGINE_VERSION' ) ) { |
|
206
|
|
|
$handlers[] = new QueryEngineEntityHandler( $this->newQueryStoreWriter() ); |
|
207
|
|
|
} |
|
208
|
|
|
|
|
209
|
|
|
return $handlers; |
|
210
|
|
|
} |
|
211
|
|
|
|
|
212
|
1 |
|
public function newEntityStore() { |
|
213
|
1 |
|
return $this->newEntityStoreFactory()->newEntityStore(); |
|
214
|
|
|
} |
|
215
|
|
|
|
|
216
|
1 |
|
public function newItemStore() { |
|
217
|
1 |
|
return $this->newEntityStoreFactory()->newItemStore(); |
|
218
|
|
|
} |
|
219
|
|
|
|
|
220
|
1 |
|
private function newEntityStoreFactory() { |
|
221
|
1 |
|
return new EntityStoreFactory( |
|
222
|
1 |
|
$this->connection, |
|
223
|
1 |
|
new EntityStoreConfig( self::ENTITY_STORE_PREFIX ) |
|
224
|
|
|
); |
|
225
|
|
|
} |
|
226
|
|
|
|
|
227
|
1 |
|
public function newTermStoreWriter() { |
|
228
|
1 |
|
return new TermStoreWriter( |
|
229
|
1 |
|
$this->connection, |
|
230
|
1 |
|
new TermStoreConfig( self::TERMS_STORE_PREFIX ) |
|
231
|
|
|
); |
|
232
|
|
|
} |
|
233
|
|
|
|
|
234
|
1 |
|
private function newDataValueDeserializer() { |
|
235
|
1 |
|
return new DataValueDeserializer( [ |
|
236
|
1 |
|
'boolean' => BooleanValue::class, |
|
237
|
|
|
'number' => NumberValue::class, |
|
238
|
|
|
'string' => StringValue::class, |
|
239
|
|
|
'unknown' => UnknownValue::class, |
|
240
|
|
|
'globecoordinate' => GlobeCoordinateValue::class, |
|
241
|
|
|
'monolingualtext' => MonolingualTextValue::class, |
|
242
|
|
|
'multilingualtext' => MultilingualTextValue::class, |
|
243
|
|
|
'quantity' => QuantityValue::class, |
|
244
|
|
|
'time' => TimeValue::class, |
|
245
|
|
|
'wikibase-entityid' => EntityIdValue::class, |
|
246
|
|
|
] ); |
|
247
|
|
|
} |
|
248
|
|
|
|
|
249
|
1 |
|
private function newEntityIdParser() { |
|
250
|
1 |
|
return new BasicEntityIdParser(); |
|
251
|
|
|
} |
|
252
|
|
|
|
|
253
|
1 |
|
public function newLegacyEntityDeserializer() { |
|
254
|
1 |
|
$factory = new DeserializerFactory( |
|
255
|
1 |
|
$this->newDataValueDeserializer(), |
|
256
|
1 |
|
$this->newEntityIdParser() |
|
257
|
|
|
); |
|
258
|
|
|
|
|
259
|
1 |
|
return $factory->newEntityDeserializer(); |
|
260
|
|
|
} |
|
261
|
|
|
|
|
262
|
|
|
public function newCurrentEntityDeserializer() { |
|
263
|
|
|
$factory = new \Wikibase\DataModel\DeserializerFactory( |
|
264
|
|
|
$this->newDataValueDeserializer(), |
|
265
|
|
|
$this->newEntityIdParser() |
|
266
|
|
|
); |
|
267
|
|
|
|
|
268
|
|
|
return $factory->newEntityDeserializer(); |
|
269
|
|
|
} |
|
270
|
|
|
|
|
271
|
|
|
public function newCurrentEntitySerializer() { |
|
272
|
|
|
$factory = new SerializerFactory( |
|
273
|
|
|
new DataValueSerializer() |
|
274
|
|
|
); |
|
275
|
|
|
|
|
276
|
|
|
return $factory->newEntitySerializer(); |
|
277
|
|
|
} |
|
278
|
|
|
|
|
279
|
|
|
public function newQueryStoreWriter() { |
|
280
|
|
|
return $this->newSqlStore()->newWriter( $this->connection ); |
|
281
|
|
|
} |
|
282
|
|
|
|
|
283
|
1 |
|
public function newTermStoreInstaller() { |
|
284
|
1 |
|
return new TermStoreInstaller( |
|
285
|
1 |
|
$this->connection->getSchemaManager(), |
|
286
|
1 |
|
new TermStoreConfig( self::TERMS_STORE_PREFIX ) |
|
287
|
|
|
); |
|
288
|
|
|
} |
|
289
|
|
|
|
|
290
|
|
|
} |
|
291
|
|
|
|