1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
namespace WikibaseQuality\ExternalValidation\DumpMetaInformation; |
4
|
|
|
|
5
|
|
|
use Database; |
6
|
|
|
use InvalidArgumentException; |
7
|
|
|
use UnexpectedValueException; |
8
|
|
|
use ResultWrapper; |
9
|
|
|
use Wikibase\DataModel\Entity\ItemId; |
10
|
|
|
use Wikibase\DataModel\Entity\PropertyId; |
11
|
|
|
|
12
|
|
|
/** |
13
|
|
|
* Implements access to the actual database table that stores the dump information. |
14
|
|
|
* |
15
|
|
|
* @package WikibaseQuality\ExternalValidation\DumpMetaInformation |
16
|
|
|
* @author BP2014N1 |
17
|
|
|
* @license GNU GPL v2+ |
18
|
|
|
*/ |
19
|
|
|
class SqlDumpMetaInformationRepo implements DumpMetaInformationLookup, DumpMetaInformationStore { |
20
|
|
|
|
21
|
|
|
const META_TABLE_NAME = 'wbqev_dump_information'; |
22
|
|
|
const IDENTIFIER_PROPERTIES_TABLE_NAME = 'wbqev_identifier_properties'; |
23
|
|
|
|
24
|
|
|
/** |
25
|
|
|
* Gets DumpMetaInformation for specific dump ids from database |
26
|
|
|
* Returns array in the form 'dumpId' => DumpMetaInformation |
27
|
|
|
* |
28
|
|
|
* @param string[] $dumpIds |
29
|
|
|
* @return DumpMetaInformation[] |
30
|
|
|
* @throws InvalidArgumentException |
31
|
|
|
*/ |
32
|
|
|
public function getWithIds( array $dumpIds ) { |
33
|
|
|
foreach ( $dumpIds as $dumpId ) { |
34
|
|
|
if ( !is_string( $dumpId ) ) { |
35
|
|
|
throw new InvalidArgumentException( '$dumpIds must contain only strings.' ); |
36
|
|
|
} |
37
|
|
|
} |
38
|
|
|
|
39
|
|
|
if( count( $dumpIds ) > 0 ) { |
40
|
|
|
$db = wfGetDB( DB_SLAVE ); |
41
|
|
|
$result = $db->select( |
42
|
|
|
array( |
43
|
|
|
self::META_TABLE_NAME, |
44
|
|
|
self::IDENTIFIER_PROPERTIES_TABLE_NAME |
45
|
|
|
), |
46
|
|
|
'*', |
47
|
|
|
array( |
48
|
|
|
'id' => $dumpIds |
49
|
|
|
), |
50
|
|
|
__METHOD__, |
51
|
|
|
array(), |
52
|
|
|
array( |
53
|
|
|
self::IDENTIFIER_PROPERTIES_TABLE_NAME => array( |
54
|
|
|
'LEFT JOIN', |
55
|
|
|
'dump_id = id' |
56
|
|
|
) |
57
|
|
|
) |
58
|
|
|
); |
59
|
|
|
|
60
|
|
|
return $this->buildDumpMetaInformationFromResult( $result ); |
61
|
|
|
} |
62
|
|
|
|
63
|
|
|
return array(); |
64
|
|
|
} |
65
|
|
|
|
66
|
|
|
/** |
67
|
|
|
* Gets DumpMetaInformation for specific identifier properties from database |
68
|
|
|
* Returns array in the form 'dumpId' => DumpMetaInformation |
69
|
|
|
* |
70
|
|
|
* @param PropertyId[] $identifierPropertyIds |
71
|
|
|
* |
72
|
|
|
* @throws InvalidArgumentException |
73
|
|
|
* @return DumpMetaInformation[] |
74
|
|
|
*/ |
75
|
|
|
public function getWithIdentifierProperties( array $identifierPropertyIds ) { |
76
|
|
|
foreach ( $identifierPropertyIds as $propertyId ) { |
77
|
|
|
if ( !( $propertyId instanceof PropertyId ) ) { |
78
|
|
|
throw new InvalidArgumentException( '$identifierProperties must contain only PropertyIds.' ); |
79
|
|
|
} |
80
|
|
|
} |
81
|
|
|
|
82
|
|
|
if( count( $identifierPropertyIds ) > 0 ) { |
83
|
|
|
$db = wfGetDB( DB_SLAVE ); |
84
|
|
|
$identifierPropertyIds = array_map( |
85
|
|
|
function( PropertyId $id ) { |
86
|
|
|
return $id->getSerialization(); |
87
|
|
|
}, |
88
|
|
|
$identifierPropertyIds |
89
|
|
|
); |
90
|
|
|
$result = $db->select( |
91
|
|
|
self::IDENTIFIER_PROPERTIES_TABLE_NAME, |
92
|
|
|
'dump_id', |
93
|
|
|
array( |
94
|
|
|
'identifier_pid' => $identifierPropertyIds |
95
|
|
|
) |
96
|
|
|
); |
97
|
|
|
$dumpIds = array(); |
98
|
|
|
foreach ( $result as $row ) { |
99
|
|
|
if ( !in_array( $row->dump_id, $dumpIds ) ) { |
100
|
|
|
$dumpIds[] = $row->dump_id; |
101
|
|
|
} |
102
|
|
|
} |
103
|
|
|
|
104
|
|
|
return $this->getWithIds( $dumpIds ); |
105
|
|
|
} |
106
|
|
|
|
107
|
|
|
return array(); |
108
|
|
|
} |
109
|
|
|
|
110
|
|
|
/** |
111
|
|
|
* Gets all DumpMetaInformation from database |
112
|
|
|
* Returns array in the form 'dumpId' => DumpMetaInformation |
113
|
|
|
* |
114
|
|
|
* @return DumpMetaInformation[] |
115
|
|
|
*/ |
116
|
|
|
public function getAll() { |
117
|
|
|
$db = wfGetDB( DB_SLAVE ); |
118
|
|
|
$result = $db->select( |
119
|
|
|
array( |
120
|
|
|
self::META_TABLE_NAME, |
121
|
|
|
self::IDENTIFIER_PROPERTIES_TABLE_NAME |
122
|
|
|
), |
123
|
|
|
'*', |
124
|
|
|
array(), |
125
|
|
|
__METHOD__, |
126
|
|
|
array(), |
127
|
|
|
array( |
128
|
|
|
self::IDENTIFIER_PROPERTIES_TABLE_NAME => array( |
129
|
|
|
'LEFT JOIN', |
130
|
|
|
'dump_id = id' |
131
|
|
|
) |
132
|
|
|
) |
133
|
|
|
); |
134
|
|
|
|
135
|
|
|
return $this->buildDumpMetaInformationFromResult( $result ); |
136
|
|
|
} |
137
|
|
|
|
138
|
|
|
/** |
139
|
|
|
* @param ResultWrapper $result |
140
|
|
|
* @return null|array |
141
|
|
|
* @throws UnexpectedValueException |
142
|
|
|
*/ |
143
|
|
|
private function buildDumpMetaInformationFromResult( ResultWrapper $result ) { |
144
|
|
|
$aggregatedRows = array(); |
145
|
|
|
foreach ( $result as $row ) { |
146
|
|
|
if ( array_key_exists( $row->id, $aggregatedRows ) ) { |
147
|
|
|
$propertyId = new PropertyId( $row->identifier_pid ); |
148
|
|
|
$aggregatedRows[$row->id]->identifier_pid[] = $propertyId; |
149
|
|
|
} else { |
150
|
|
|
if ( $row->identifier_pid !== null ) { |
151
|
|
|
$propertyId = new PropertyId( $row->identifier_pid ); |
152
|
|
|
$row->identifier_pid = array( $propertyId ); |
153
|
|
|
} |
154
|
|
|
$aggregatedRows[$row->id] = $row; |
155
|
|
|
} |
156
|
|
|
} |
157
|
|
|
|
158
|
|
|
$dumpMetaInformation = array(); |
159
|
|
|
foreach ( $aggregatedRows as $row ) { |
160
|
|
|
$dumpId = $row->id; |
161
|
|
|
$sourceItemId = new ItemId( $row->source_qid ); |
162
|
|
|
$importDate = wfTimestamp( TS_MW, $row->import_date ); |
163
|
|
|
$language = $row->language; |
164
|
|
|
$sourceUrl = $row->source_url; |
165
|
|
|
$size = (int)$row->size; |
166
|
|
|
$licenseItemId = new ItemId( $row->license_qid ); |
167
|
|
|
$identifierPropertyIds = $row->identifier_pid; |
168
|
|
|
|
169
|
|
|
$dumpMetaInformation[$row->dump_id] = |
170
|
|
|
new DumpMetaInformation( $dumpId, |
171
|
|
|
$sourceItemId, |
172
|
|
|
$identifierPropertyIds, |
173
|
|
|
$importDate, |
174
|
|
|
$language, |
175
|
|
|
$sourceUrl, |
176
|
|
|
$size, |
177
|
|
|
$licenseItemId |
178
|
|
|
); |
179
|
|
|
} |
180
|
|
|
|
181
|
|
|
return $dumpMetaInformation; |
182
|
|
|
} |
183
|
|
|
|
184
|
|
|
/** |
185
|
|
|
* Inserts or updates given dump meta information to database |
186
|
|
|
* |
187
|
|
|
* @param DumpMetaInformation $dumpMetaInformation |
188
|
|
|
*/ |
189
|
|
|
public function save( DumpMetaInformation $dumpMetaInformation ) { |
190
|
|
|
$db = wfGetDB( DB_SLAVE ); |
191
|
|
|
$dumpId = $dumpMetaInformation->getDumpId(); |
192
|
|
|
$accumulator = $this->getDumpInformationFields( $db, $dumpMetaInformation ); |
193
|
|
|
|
194
|
|
|
$existing = $db->selectRow( |
195
|
|
|
self::META_TABLE_NAME, |
196
|
|
|
array( 'id' ), |
197
|
|
|
array( 'id' => $dumpId ) |
198
|
|
|
); |
199
|
|
|
|
200
|
|
|
if ( $existing ) { |
201
|
|
|
$db->update( |
202
|
|
|
self::META_TABLE_NAME, |
203
|
|
|
$accumulator, |
204
|
|
|
array( 'id' => $dumpId ) |
205
|
|
|
); |
206
|
|
|
} else { |
207
|
|
|
$db->insert( |
208
|
|
|
self::META_TABLE_NAME, |
209
|
|
|
$accumulator |
210
|
|
|
); |
211
|
|
|
} |
212
|
|
|
|
213
|
|
|
$db->delete( |
214
|
|
|
self::IDENTIFIER_PROPERTIES_TABLE_NAME, |
215
|
|
|
array( 'dump_id' => $dumpId ) |
216
|
|
|
); |
217
|
|
|
$db->insert( |
218
|
|
|
self::IDENTIFIER_PROPERTIES_TABLE_NAME, |
219
|
|
|
array_map( |
220
|
|
|
function ( PropertyId $identifierPropertyId ) use ( $dumpId ) { |
221
|
|
|
return array( |
222
|
|
|
'dump_id' => $dumpId, |
223
|
|
|
'identifier_pid' => $identifierPropertyId->getSerialization() |
224
|
|
|
); |
225
|
|
|
}, |
226
|
|
|
$dumpMetaInformation->getIdentifierPropertyIds() |
227
|
|
|
) |
228
|
|
|
); |
229
|
|
|
} |
230
|
|
|
|
231
|
|
|
/** |
232
|
|
|
* @param Database $db |
233
|
|
|
* @param DumpMetaInformation $dumpMetaInformation |
234
|
|
|
* @return array |
235
|
|
|
*/ |
236
|
|
|
private function getDumpInformationFields( Database $db, DumpMetaInformation $dumpMetaInformation ) { |
237
|
|
|
return array( |
238
|
|
|
'id' => $dumpMetaInformation->getDumpId(), |
239
|
|
|
'source_qid' => $dumpMetaInformation->getSourceItemId()->getSerialization(), |
240
|
|
|
'import_date' => $db->timestamp( $dumpMetaInformation->getImportDate() ), |
241
|
|
|
'language' => $dumpMetaInformation->getLanguageCode(), |
242
|
|
|
'source_url' => $dumpMetaInformation->getSourceUrl(), |
243
|
|
|
'size' => $dumpMetaInformation->getSize(), |
244
|
|
|
'license_qid' => $dumpMetaInformation->getLicenseItemId()->getSerialization() |
245
|
|
|
); |
246
|
|
|
} |
247
|
|
|
|
248
|
|
|
} |
249
|
|
|
|