1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
namespace Wikibase\Repo\ChangeOp; |
4
|
|
|
|
5
|
|
|
use InvalidArgumentException; |
6
|
|
|
use Site; |
7
|
|
|
use SiteLookup; |
8
|
|
|
use ValueValidators\Error; |
9
|
|
|
use ValueValidators\Result; |
10
|
|
|
use Wikibase\DataModel\Entity\EntityId; |
11
|
|
|
use Wikibase\DataModel\Entity\Item; |
12
|
|
|
use Wikibase\DataModel\Entity\ItemId; |
13
|
|
|
use Wikibase\DataModel\SiteLink; |
14
|
|
|
use Wikibase\Repo\Merge\StatementsMerger; |
15
|
|
|
use Wikibase\Repo\Merge\Validator\NoCrossReferencingStatements; |
16
|
|
|
use Wikibase\Repo\Validators\CompositeEntityValidator; |
17
|
|
|
use Wikibase\Repo\Validators\EntityConstraintProvider; |
18
|
|
|
use Wikibase\Repo\Validators\UniquenessViolation; |
19
|
|
|
|
20
|
|
|
/** |
21
|
|
|
* @license GPL-2.0-or-later |
22
|
|
|
* @author Addshore |
23
|
|
|
* @author Daniel Kinzler |
24
|
|
|
*/ |
25
|
|
|
class ChangeOpsMerge { |
26
|
|
|
|
27
|
|
|
/** |
28
|
|
|
* @var Item |
29
|
|
|
*/ |
30
|
|
|
private $fromItem; |
31
|
|
|
|
32
|
|
|
/** |
33
|
|
|
* @var Item |
34
|
|
|
*/ |
35
|
|
|
private $toItem; |
36
|
|
|
|
37
|
|
|
/** |
38
|
|
|
* @var ChangeOps |
39
|
|
|
*/ |
40
|
|
|
private $fromChangeOps; |
41
|
|
|
|
42
|
|
|
/** |
43
|
|
|
* @var ChangeOps |
44
|
|
|
*/ |
45
|
|
|
private $toChangeOps; |
46
|
|
|
|
47
|
|
|
/** |
48
|
|
|
* @var string[] |
49
|
|
|
*/ |
50
|
|
|
private $ignoreConflicts; |
51
|
|
|
|
52
|
|
|
/** |
53
|
|
|
* @var EntityConstraintProvider |
54
|
|
|
*/ |
55
|
|
|
private $constraintProvider; |
56
|
|
|
|
57
|
|
|
/** |
58
|
|
|
* @var ChangeOpFactoryProvider |
59
|
|
|
*/ |
60
|
|
|
private $changeOpFactoryProvider; |
61
|
|
|
|
62
|
|
|
/** |
63
|
|
|
* @var SiteLookup |
64
|
|
|
*/ |
65
|
|
|
private $siteLookup; |
66
|
|
|
|
67
|
|
|
public static $conflictTypes = [ 'description', 'sitelink', 'statement' ]; |
68
|
|
|
|
69
|
|
|
/** |
70
|
|
|
* @var StatementsMerger |
71
|
|
|
*/ |
72
|
|
|
private $statementsMerger; |
73
|
|
|
|
74
|
|
|
/** |
75
|
|
|
* @param Item $fromItem |
76
|
|
|
* @param Item $toItem |
77
|
|
|
* @param string[] $ignoreConflicts list of elements to ignore conflicts for |
78
|
|
|
* can only contain 'description' and or 'sitelink' and or 'statement' |
79
|
|
|
* @param EntityConstraintProvider $constraintProvider |
80
|
|
|
* @param ChangeOpFactoryProvider $changeOpFactoryProvider |
81
|
|
|
* @param SiteLookup $siteLookup |
82
|
|
|
* @param StatementsMerger $statementsMerger |
83
|
|
|
* |
84
|
|
|
* @todo Injecting ChangeOpFactoryProvider is an Abomination Unto Nuggan, we'll |
85
|
|
|
* need a MergeChangeOpsSequenceBuilder or some such. This will allow us |
86
|
|
|
* to merge different kinds of entities nicely, too. |
87
|
|
|
*/ |
88
|
|
|
public function __construct( |
89
|
|
|
Item $fromItem, |
90
|
|
|
Item $toItem, |
91
|
|
|
array $ignoreConflicts, |
92
|
|
|
EntityConstraintProvider $constraintProvider, |
93
|
|
|
ChangeOpFactoryProvider $changeOpFactoryProvider, |
94
|
|
|
SiteLookup $siteLookup, |
95
|
|
|
StatementsMerger $statementsMerger |
96
|
|
|
) { |
97
|
|
|
$this->assertValidIgnoreConflictValues( $ignoreConflicts ); |
98
|
|
|
|
99
|
|
|
$this->fromItem = $fromItem; |
100
|
|
|
$this->toItem = $toItem; |
101
|
|
|
$this->fromChangeOps = new ChangeOps(); |
102
|
|
|
$this->toChangeOps = new ChangeOps(); |
103
|
|
|
$this->ignoreConflicts = $ignoreConflicts; |
104
|
|
|
$this->constraintProvider = $constraintProvider; |
105
|
|
|
$this->siteLookup = $siteLookup; |
106
|
|
|
|
107
|
|
|
$this->changeOpFactoryProvider = $changeOpFactoryProvider; |
108
|
|
|
$this->statementsMerger = $statementsMerger; |
109
|
|
|
} |
110
|
|
|
|
111
|
|
|
/** |
112
|
|
|
* @param string[] $ignoreConflicts can contain strings 'description' or 'sitelink' |
113
|
|
|
* |
114
|
|
|
* @throws InvalidArgumentException |
115
|
|
|
*/ |
116
|
|
|
private function assertValidIgnoreConflictValues( array $ignoreConflicts ) { |
117
|
|
|
if ( array_diff( $ignoreConflicts, self::$conflictTypes ) ) { |
118
|
|
|
throw new InvalidArgumentException( |
119
|
|
|
'$ignoreConflicts array can only contain "description" and or "sitelink" and or "statement" values' |
120
|
|
|
); |
121
|
|
|
} |
122
|
|
|
} |
123
|
|
|
|
124
|
|
|
/** |
125
|
|
|
* @return FingerprintChangeOpFactory |
126
|
|
|
*/ |
127
|
|
|
private function getFingerprintChangeOpFactory() { |
128
|
|
|
return $this->changeOpFactoryProvider->getFingerprintChangeOpFactory(); |
129
|
|
|
} |
130
|
|
|
|
131
|
|
|
/** |
132
|
|
|
* @return SiteLinkChangeOpFactory |
133
|
|
|
*/ |
134
|
|
|
private function getSiteLinkChangeOpFactory() { |
135
|
|
|
return $this->changeOpFactoryProvider->getSiteLinkChangeOpFactory(); |
136
|
|
|
} |
137
|
|
|
|
138
|
|
|
/** |
139
|
|
|
* @throws ChangeOpException |
140
|
|
|
*/ |
141
|
|
|
public function apply() { |
142
|
|
|
// NOTE: we don't want to validate the ChangeOps individually, since they represent |
143
|
|
|
// data already present and saved on the system. Also, validating each would be |
144
|
|
|
// potentially expensive. |
145
|
|
|
|
146
|
|
|
$this->generateChangeOps(); |
147
|
|
|
|
148
|
|
|
$this->fromChangeOps->apply( $this->fromItem ); |
149
|
|
|
$this->toChangeOps->apply( $this->toItem ); |
150
|
|
|
|
151
|
|
|
$this->checkStatementLinks(); |
152
|
|
|
$this->statementsMerger->merge( $this->fromItem, $this->toItem ); |
153
|
|
|
|
154
|
|
|
//NOTE: we apply constraint checks on the modified items, but no |
155
|
|
|
// validation of individual change ops, since we are merging |
156
|
|
|
// two valid items. |
157
|
|
|
$this->applyConstraintChecks( $this->toItem, $this->fromItem->getId() ); |
|
|
|
|
158
|
|
|
|
159
|
|
|
return new DummyChangeOpResult(); |
160
|
|
|
} |
161
|
|
|
|
162
|
|
|
private function generateChangeOps() { |
163
|
|
|
$this->generateLabelsChangeOps(); |
164
|
|
|
$this->generateDescriptionsChangeOps(); |
165
|
|
|
$this->generateAliasesChangeOps(); |
166
|
|
|
$this->generateSitelinksChangeOps(); |
167
|
|
|
} |
168
|
|
|
|
169
|
|
|
private function generateLabelsChangeOps() { |
170
|
|
|
foreach ( $this->fromItem->getLabels()->toTextArray() as $langCode => $label ) { |
171
|
|
|
if ( !$this->toItem->getLabels()->hasTermForLanguage( $langCode ) |
172
|
|
|
|| $this->toItem->getLabels()->getByLanguage( $langCode )->getText() === $label |
173
|
|
|
) { |
174
|
|
|
$this->fromChangeOps->add( $this->getFingerprintChangeOpFactory()->newRemoveLabelOp( $langCode ) ); |
175
|
|
|
$this->toChangeOps->add( $this->getFingerprintChangeOpFactory()->newSetLabelOp( $langCode, $label ) ); |
176
|
|
|
} else { |
177
|
|
|
$this->fromChangeOps->add( $this->getFingerprintChangeOpFactory()->newRemoveLabelOp( $langCode ) ); |
178
|
|
|
$this->toChangeOps->add( $this->getFingerprintChangeOpFactory()->newAddAliasesOp( $langCode, [ $label ] ) ); |
179
|
|
|
} |
180
|
|
|
} |
181
|
|
|
} |
182
|
|
|
|
183
|
|
|
private function generateDescriptionsChangeOps() { |
184
|
|
|
foreach ( $this->fromItem->getDescriptions()->toTextArray() as $langCode => $desc ) { |
185
|
|
|
if ( !$this->toItem->getDescriptions()->hasTermForLanguage( $langCode ) |
186
|
|
|
|| $this->toItem->getDescriptions()->getByLanguage( $langCode )->getText() === $desc |
187
|
|
|
) { |
188
|
|
|
$this->fromChangeOps->add( $this->getFingerprintChangeOpFactory()->newRemoveDescriptionOp( $langCode ) ); |
189
|
|
|
$this->toChangeOps->add( $this->getFingerprintChangeOpFactory()->newSetDescriptionOp( $langCode, $desc ) ); |
190
|
|
|
} elseif ( !in_array( 'description', $this->ignoreConflicts ) ) { |
191
|
|
|
throw new ChangeOpException( "Conflicting descriptions for language {$langCode}" ); |
192
|
|
|
} |
193
|
|
|
} |
194
|
|
|
} |
195
|
|
|
|
196
|
|
|
private function generateAliasesChangeOps() { |
197
|
|
|
foreach ( $this->fromItem->getAliasGroups()->toTextArray() as $langCode => $aliases ) { |
198
|
|
|
$this->fromChangeOps->add( $this->getFingerprintChangeOpFactory()->newRemoveAliasesOp( $langCode, $aliases ) ); |
199
|
|
|
$this->toChangeOps->add( $this->getFingerprintChangeOpFactory()->newAddAliasesOp( $langCode, $aliases ) ); |
200
|
|
|
} |
201
|
|
|
} |
202
|
|
|
|
203
|
|
|
private function generateSitelinksChangeOps() { |
204
|
|
|
foreach ( $this->fromItem->getSiteLinkList()->toArray() as $fromSiteLink ) { |
205
|
|
|
$siteId = $fromSiteLink->getSiteId(); |
206
|
|
|
if ( !$this->toItem->getSiteLinkList()->hasLinkWithSiteId( $siteId ) ) { |
207
|
|
|
$this->generateSitelinksChangeOpsWithNoConflict( $fromSiteLink ); |
208
|
|
|
} else { |
209
|
|
|
$this->generateSitelinksChangeOpsWithConflict( $fromSiteLink ); |
210
|
|
|
} |
211
|
|
|
} |
212
|
|
|
} |
213
|
|
|
|
214
|
|
|
private function generateSitelinksChangeOpsWithNoConflict( SiteLink $fromSiteLink ) { |
215
|
|
|
$siteId = $fromSiteLink->getSiteId(); |
216
|
|
|
$this->fromChangeOps->add( $this->getSiteLinkChangeOpFactory()->newRemoveSiteLinkOp( $siteId ) ); |
217
|
|
|
$this->toChangeOps->add( |
218
|
|
|
$this->getSiteLinkChangeOpFactory()->newSetSiteLinkOp( |
219
|
|
|
$siteId, |
220
|
|
|
$fromSiteLink->getPageName(), |
221
|
|
|
$fromSiteLink->getBadges() |
222
|
|
|
) |
223
|
|
|
); |
224
|
|
|
} |
225
|
|
|
|
226
|
|
|
private function generateSitelinksChangeOpsWithConflict( SiteLink $fromSiteLink ) { |
227
|
|
|
$siteId = $fromSiteLink->getSiteId(); |
228
|
|
|
$toSiteLink = $this->toItem->getSiteLink( $siteId ); |
229
|
|
|
$fromPageName = $fromSiteLink->getPageName(); |
230
|
|
|
$toPageName = $toSiteLink->getPageName(); |
231
|
|
|
|
232
|
|
|
if ( $fromPageName !== $toPageName ) { |
233
|
|
|
$site = $this->getSite( $siteId ); |
234
|
|
|
$fromPageName = $site->normalizePageName( $fromPageName ); |
235
|
|
|
$toPageName = $site->normalizePageName( $toPageName ); |
236
|
|
|
} |
237
|
|
|
if ( $fromPageName === $toPageName ) { |
238
|
|
|
$this->fromChangeOps->add( $this->getSiteLinkChangeOpFactory()->newRemoveSiteLinkOp( $siteId ) ); |
239
|
|
|
$this->toChangeOps->add( |
240
|
|
|
$this->getSiteLinkChangeOpFactory()->newSetSiteLinkOp( |
241
|
|
|
$siteId, |
242
|
|
|
$fromPageName, |
243
|
|
|
array_unique( array_merge( $fromSiteLink->getBadges(), $toSiteLink->getBadges() ) ) |
244
|
|
|
) |
245
|
|
|
); |
246
|
|
|
} elseif ( !in_array( 'sitelink', $this->ignoreConflicts ) ) { |
247
|
|
|
throw new ChangeOpException( "Conflicting sitelinks for {$siteId}" ); |
248
|
|
|
} |
249
|
|
|
} |
250
|
|
|
|
251
|
|
|
/** |
252
|
|
|
* @param string $siteId |
253
|
|
|
* |
254
|
|
|
* @throws ChangeOpException |
255
|
|
|
* @return Site |
256
|
|
|
*/ |
257
|
|
|
private function getSite( $siteId ) { |
258
|
|
|
$site = $this->siteLookup->getSite( $siteId ); |
259
|
|
|
if ( $site === null ) { |
260
|
|
|
throw new ChangeOpException( "Conflicting sitelinks for {$siteId}, Failed to normalize" ); |
261
|
|
|
} |
262
|
|
|
return $site; |
263
|
|
|
} |
264
|
|
|
|
265
|
|
|
/** |
266
|
|
|
* @param Item $item |
267
|
|
|
* @param ItemId $fromId |
268
|
|
|
* |
269
|
|
|
* @throws ChangeOpValidationException if it would not be possible to save the updated items. |
270
|
|
|
*/ |
271
|
|
|
private function applyConstraintChecks( Item $item, ItemId $fromId ) { |
272
|
|
|
$constraintValidator = new CompositeEntityValidator( |
273
|
|
|
$this->constraintProvider->getUpdateValidators( $item->getType() ) |
274
|
|
|
); |
275
|
|
|
|
276
|
|
|
$result = $constraintValidator->validateEntity( $item ); |
277
|
|
|
$errors = $result->getErrors(); |
278
|
|
|
|
279
|
|
|
$errors = $this->removeConflictsWithEntity( $errors, $fromId ); |
280
|
|
|
|
281
|
|
|
if ( !empty( $errors ) ) { |
282
|
|
|
$result = Result::newError( $errors ); |
283
|
|
|
throw new ChangeOpValidationException( $result ); |
284
|
|
|
} |
285
|
|
|
} |
286
|
|
|
|
287
|
|
|
/** |
288
|
|
|
* Strip any conflicts with the given $fromId from the array of Error objects |
289
|
|
|
* |
290
|
|
|
* @param Error[] $errors |
291
|
|
|
* @param EntityId $fromId |
292
|
|
|
* |
293
|
|
|
* @return Error[] |
294
|
|
|
*/ |
295
|
|
|
private function removeConflictsWithEntity( array $errors, EntityId $fromId ) { |
296
|
|
|
$filtered = []; |
297
|
|
|
|
298
|
|
|
foreach ( $errors as $error ) { |
299
|
|
|
if ( $error instanceof UniquenessViolation |
300
|
|
|
&& $fromId->equals( $error->getConflictingEntity() ) |
301
|
|
|
) { |
302
|
|
|
continue; |
303
|
|
|
} |
304
|
|
|
|
305
|
|
|
$filtered[] = $error; |
306
|
|
|
} |
307
|
|
|
|
308
|
|
|
return $filtered; |
309
|
|
|
} |
310
|
|
|
|
311
|
|
|
private function checkStatementLinks() { |
312
|
|
|
if ( in_array( 'statement', $this->ignoreConflicts ) ) { |
313
|
|
|
return; |
314
|
|
|
} |
315
|
|
|
|
316
|
|
|
$validator = new NoCrossReferencingStatements(); |
317
|
|
|
if ( $validator->validate( $this->fromItem, $this->toItem ) ) { |
318
|
|
|
return; |
319
|
|
|
} |
320
|
|
|
|
321
|
|
|
throw new ChangeOpException( |
322
|
|
|
'The two items cannot be merged because one of them links to the other using the properties: ' . |
323
|
|
|
implode( ', ', $validator->getViolations() ) |
324
|
|
|
); |
325
|
|
|
} |
326
|
|
|
|
327
|
|
|
} |
328
|
|
|
|
Unless you are absolutely sure that the expression can never be null because of other conditions, we strongly recommend to add an additional type check to your code: