1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
namespace Wikibase\TermStore\MediaWiki\PackagePrivate; |
4
|
|
|
|
5
|
|
|
use AppendIterator; |
6
|
|
|
use ArrayIterator; |
7
|
|
|
use Wikibase\TermStore\MediaWiki\PackagePrivate\Util\ReplicaMasterAwareRecordIdsAcquirer; |
8
|
|
|
use Wikimedia\Rdbms\IDatabase; |
9
|
|
|
use Wikimedia\Rdbms\ILoadBalancer; |
10
|
|
|
|
11
|
|
|
class DatabaseTermIdsAcquirer implements TermIdsAcquirer { |
12
|
|
|
|
13
|
|
|
const TABLE_TEXT = 'wbt_text'; |
14
|
|
|
const TABLE_TEXT_IN_LANG = 'wbt_text_in_lang'; |
15
|
|
|
const TABLE_TERM_IN_LANG = 'wbt_term_in_lang'; |
16
|
|
|
|
17
|
|
|
/** |
18
|
|
|
* @var ILoadBalancer |
19
|
|
|
*/ |
20
|
|
|
private $loadBalancer; |
21
|
|
|
|
22
|
|
|
/** |
23
|
|
|
* @var TypeIdsAcquirer |
24
|
|
|
*/ |
25
|
|
|
private $typeIdsAcquirer; |
26
|
|
|
|
27
|
|
|
public function __construct( |
28
|
|
|
ILoadBalancer $loadBalancer, |
29
|
|
|
TypeIdsAcquirer $typeIdsAcquirer |
30
|
|
|
) { |
31
|
|
|
$this->loadBalancer = $loadBalancer; |
32
|
|
|
$this->typeIdsAcquirer = $typeIdsAcquirer; |
33
|
|
|
} |
34
|
|
|
|
35
|
|
|
public function acquireTermIds( array $termsArray ): array { |
36
|
|
|
$termsArray = $this->mapToTextIds( $termsArray ); |
37
|
|
|
$termsArray = $this->mapToTextInLangIds( $termsArray ); |
38
|
|
|
$termsArray = $this->mapToTypeIds( $termsArray ); |
39
|
|
|
return $this->mapToTermInLangIds( $termsArray ); |
40
|
|
|
} |
41
|
|
|
|
42
|
|
|
/** |
43
|
|
|
* replace root keys containing type names in termsArray |
44
|
|
|
* with their respective ids in wbt_type table |
45
|
|
|
* |
46
|
|
|
* @param array $termsArray terms per type per language: |
47
|
|
|
* [ |
48
|
|
|
* 'type1' => [ ... ], |
49
|
|
|
* 'type2' => [ ... ], |
50
|
|
|
* ... |
51
|
|
|
* ] |
52
|
|
|
* |
53
|
|
|
* @return array |
54
|
|
|
* [ |
55
|
|
|
* <typeId1> => [ ... ], |
56
|
|
|
* <typeId2> => [ ... ], |
57
|
|
|
* ... |
58
|
|
|
* ] |
59
|
|
|
*/ |
60
|
|
|
private function mapToTypeIds( array $termsArray ) { |
61
|
|
|
$typeIds = $this->typeIdsAcquirer->acquireTypeIds( array_keys( $termsArray ) ); |
62
|
|
|
|
63
|
|
|
$termsArrayByTypeId = []; |
64
|
|
|
foreach ( $typeIds as $type => $typeId ) { |
65
|
|
|
$termsArrayByTypeId[$typeId] = $termsArray[$type]; |
66
|
|
|
} |
67
|
|
|
|
68
|
|
|
return $termsArrayByTypeId; |
69
|
|
|
} |
70
|
|
|
|
71
|
|
|
/** |
72
|
|
|
* replace text at termsArray leaves with their ids in wbt_text table |
73
|
|
|
* and return resulting array |
74
|
|
|
* |
75
|
|
|
* @param array $termsArray terms per type per language: |
76
|
|
|
* [ |
77
|
|
|
* 'type' => [ |
78
|
|
|
* [ 'language' => 'term' | [ 'term1', 'term2', ... ] ], ... |
79
|
|
|
* ], ... |
80
|
|
|
* ] |
81
|
|
|
* |
82
|
|
|
* @return array |
83
|
|
|
* [ |
84
|
|
|
* 'type' => [ |
85
|
|
|
* [ 'language' => [ <textId1>, <textId2>, ... ] ], ... |
86
|
|
|
* ], ... |
87
|
|
|
* ] |
88
|
|
|
*/ |
89
|
|
|
private function mapToTextIds( array $termsArray ) { |
90
|
|
|
$texts = []; |
91
|
|
|
|
92
|
|
|
array_walk_recursive( $termsArray, function ( $text ) use ( &$texts ) { |
93
|
|
|
$texts[] = $text; |
94
|
|
|
} ); |
95
|
|
|
|
96
|
|
|
$textIds = $this->acquireTextIds( $texts ); |
97
|
|
|
|
98
|
|
|
array_walk_recursive( $termsArray, function ( &$text ) use ( $textIds ) { |
99
|
|
|
$text = $textIds[$text]; |
100
|
|
|
} ); |
101
|
|
|
|
102
|
|
|
return $termsArray; |
103
|
|
|
} |
104
|
|
|
|
105
|
|
|
private function acquireTextIds( array $texts ) { |
106
|
|
|
$textIdsAcquirer = new ReplicaMasterAwareRecordIdsAcquirer( |
107
|
|
|
$this->loadBalancer, 'wbt_text', 'wbx_id' ); |
108
|
|
|
|
109
|
|
|
$textRecords = []; |
110
|
|
|
foreach ( $texts as $text ) { |
111
|
|
|
$textRecords[] = [ 'wbx_text' => $text ]; |
112
|
|
|
} |
113
|
|
|
|
114
|
|
|
$acquiredIds = $textIdsAcquirer->acquireIds( $textRecords ); |
115
|
|
|
|
116
|
|
|
$textIds = []; |
117
|
|
|
foreach ( $acquiredIds as $acquiredId ) { |
118
|
|
|
$textIds[$acquiredId['wbx_text']] = $acquiredId['wbx_id']; |
119
|
|
|
} |
120
|
|
|
|
121
|
|
|
return $textIds; |
122
|
|
|
} |
123
|
|
|
|
124
|
|
|
/** |
125
|
|
|
* replace ( lang => [ textId, ... ] ) entries with their respective ids |
126
|
|
|
* in wbt_text_in_lang table and return resulting array |
127
|
|
|
* |
128
|
|
|
* @param array $termsArray text ids per type per langauge |
129
|
|
|
* [ |
130
|
|
|
* 'type' => [ |
131
|
|
|
* [ 'language' => [ <textId1>, <textId2>, ... ] ], ... |
132
|
|
|
* ], ... |
133
|
|
|
* ] |
134
|
|
|
* |
135
|
|
|
* @return array |
136
|
|
|
* [ |
137
|
|
|
* 'type' => [ <textInLangId1>, <textInLangId2>, ... ], |
138
|
|
|
* ... |
139
|
|
|
* ] |
140
|
|
|
*/ |
141
|
|
|
private function mapToTextInLangIds( array $termsArray ) { |
142
|
|
|
$flattenedLangTextIds = []; |
143
|
|
|
foreach ( $termsArray as $langTextIds ) { |
144
|
|
|
foreach ( $langTextIds as $lang => $textIds ) { |
145
|
|
|
if ( !isset( $flattenedLangTextIds[$lang] ) ) { |
146
|
|
|
$flattenedLangTextIds[$lang] = []; |
147
|
|
|
} |
148
|
|
|
|
149
|
|
|
$flattenedLangTextIds[$lang] = array_unique( |
150
|
|
|
array_merge( |
151
|
|
|
(array)$textIds, |
152
|
|
|
(array)$flattenedLangTextIds[$lang] |
153
|
|
|
) |
154
|
|
|
); |
155
|
|
|
|
156
|
|
|
} |
157
|
|
|
} |
158
|
|
|
|
159
|
|
|
$textInLangIds = $this->acquireTextInLangIds( $flattenedLangTextIds ); |
160
|
|
|
|
161
|
|
|
$newTermsArray = []; |
162
|
|
|
foreach ( $termsArray as $type => $langTextIds ) { |
163
|
|
|
$newTermsArray[$type] = []; |
164
|
|
|
foreach ( $langTextIds as $lang => $textIds ) { |
165
|
|
|
foreach ( (array)$textIds as $textId ) { |
166
|
|
|
$newTermsArray[$type][] = $textInLangIds[$lang][$textId]; |
167
|
|
|
} |
168
|
|
|
} |
169
|
|
|
} |
170
|
|
|
|
171
|
|
|
return $newTermsArray; |
172
|
|
|
} |
173
|
|
|
|
174
|
|
|
private function acquireTextInLangIds( array $langTextIds ) { |
175
|
|
|
$textInLangIdsAcquirer = new ReplicaMasterAwareRecordIdsAcquirer( |
176
|
|
|
$this->loadBalancer, 'wbt_text_in_lang', 'wbxl_id' ); |
177
|
|
|
|
178
|
|
|
$textInLangRecords = []; |
179
|
|
|
foreach ( $langTextIds as $lang => $textIds ) { |
180
|
|
|
foreach ( $textIds as $textId ) { |
181
|
|
|
$textInLangRecords[] = [ 'wbxl_text_id' => $textId, 'wbxl_language' => $lang ]; |
182
|
|
|
} |
183
|
|
|
} |
184
|
|
|
|
185
|
|
|
$acquiredIds = $textInLangIdsAcquirer->acquireIds( $textInLangRecords ); |
186
|
|
|
|
187
|
|
|
$textInLangIds = []; |
188
|
|
|
foreach ( $acquiredIds as $acquiredId ) { |
189
|
|
|
$textInLangIds[$acquiredId['wbxl_language']][$acquiredId['wbxl_text_id']] |
190
|
|
|
= $acquiredId['wbxl_id']; |
191
|
|
|
} |
192
|
|
|
|
193
|
|
|
return $textInLangIds; |
194
|
|
|
} |
195
|
|
|
|
196
|
|
|
/** |
197
|
|
|
* replace root ( type => [ textInLangId, ... ] ) entries with their respective ids |
198
|
|
|
* in wbt_term_in_lang table and return resulting array |
199
|
|
|
* |
200
|
|
|
* @param array $termsArray text in lang ids per type |
201
|
|
|
* [ |
202
|
|
|
* 'type' => [ <textInLangId1>, <textInLangId2>, ... ], |
203
|
|
|
* ... |
204
|
|
|
* ] |
205
|
|
|
* |
206
|
|
|
* @return array |
207
|
|
|
* [ |
208
|
|
|
* <termInLang1>, |
209
|
|
|
* <termInLang2>, |
210
|
|
|
* ... |
211
|
|
|
* ] |
212
|
|
|
*/ |
213
|
|
|
private function mapToTermInLangIds( array $termsArray ) { |
214
|
|
|
$flattenedTypeTextInLangIds = []; |
215
|
|
|
foreach ( $termsArray as $typeId => $textInLangIds ) { |
216
|
|
|
if ( !isset( $flattenedTypeTextInLangIds[$typeId] ) ) { |
217
|
|
|
$flattenedTypeTextInLangIds[$typeId] = []; |
218
|
|
|
} |
219
|
|
|
|
220
|
|
|
$flattenedTypeTextInLangIds[$typeId] = array_unique( |
221
|
|
|
array_merge( |
222
|
|
|
(array)$textInLangIds, |
223
|
|
|
(array)$flattenedTypeTextInLangIds[$typeId] |
224
|
|
|
) |
225
|
|
|
); |
226
|
|
|
} |
227
|
|
|
|
228
|
|
|
$termInLangIds = $this->acquireTermInLangIds( $flattenedTypeTextInLangIds ); |
229
|
|
|
|
230
|
|
|
$newTermsArray = []; |
231
|
|
|
foreach ( $termsArray as $typeId => $textInLangIds ) { |
232
|
|
|
foreach ( $textInLangIds as $textInLangId ) { |
233
|
|
|
$newTermsArray[] = $termInLangIds[$typeId][$textInLangId]; |
234
|
|
|
} |
235
|
|
|
} |
236
|
|
|
|
237
|
|
|
return $newTermsArray; |
238
|
|
|
} |
239
|
|
|
|
240
|
|
|
private function acquireTermInLangIds( array $typeTextInLangIds ) { |
241
|
|
|
$termInLangIdsAcquirer = new ReplicaMasterAwareRecordIdsAcquirer( |
242
|
|
|
$this->loadBalancer, 'wbt_term_in_lang', 'wbtl_id' ); |
243
|
|
|
|
244
|
|
|
$termInLangRecords = []; |
245
|
|
|
foreach ( $typeTextInLangIds as $typeId => $textInLangIds ) { |
246
|
|
|
foreach ( $textInLangIds as $textInLangId ) { |
247
|
|
|
$termInLangRecords[] = [ |
248
|
|
|
'wbtl_text_in_lang_id' => $textInLangId, |
249
|
|
|
'wbtl_type_id' => (string)$typeId |
250
|
|
|
]; |
251
|
|
|
} |
252
|
|
|
} |
253
|
|
|
|
254
|
|
|
$acquiredIds = $termInLangIdsAcquirer->acquireIds( $termInLangRecords ); |
255
|
|
|
|
256
|
|
|
$termInLangIds = []; |
257
|
|
|
foreach ( $acquiredIds as $acquiredId ) { |
258
|
|
|
$termInLangIds[$acquiredId['wbtl_type_id']][$acquiredId['wbtl_text_in_lang_id']] |
259
|
|
|
= $acquiredId['wbtl_id']; |
260
|
|
|
} |
261
|
|
|
|
262
|
|
|
return $termInLangIds; |
263
|
|
|
} |
264
|
|
|
|
265
|
|
|
} |
266
|
|
|
|