|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Wikibase\Client\Changes; |
|
4
|
|
|
|
|
5
|
|
|
use Hooks; |
|
6
|
|
|
use InvalidArgumentException; |
|
7
|
|
|
use Psr\Log\LoggerInterface; |
|
8
|
|
|
use SiteLookup; |
|
9
|
|
|
use Title; |
|
10
|
|
|
use TitleFactory; |
|
11
|
|
|
use Wikibase\Client\Usage\PageEntityUsages; |
|
12
|
|
|
use Wikibase\Lib\Changes\Change; |
|
13
|
|
|
use Wikibase\Lib\Changes\EntityChange; |
|
14
|
|
|
|
|
15
|
|
|
/** |
|
16
|
|
|
* Interface for change handling. Whenever a change is detected, |
|
17
|
|
|
* it should be fed to this service which then takes care handling it. |
|
18
|
|
|
* |
|
19
|
|
|
* @see @ref md_docs_topics_change-propagation for an overview of the change propagation mechanism. |
|
20
|
|
|
* |
|
21
|
|
|
* @license GPL-2.0-or-later |
|
22
|
|
|
* @author Daniel Kinzler |
|
23
|
|
|
*/ |
|
24
|
|
|
class ChangeHandler { |
|
25
|
|
|
|
|
26
|
|
|
/** |
|
27
|
|
|
* @var AffectedPagesFinder |
|
28
|
|
|
*/ |
|
29
|
|
|
private $affectedPagesFinder; |
|
30
|
|
|
|
|
31
|
|
|
/** |
|
32
|
|
|
* @var TitleFactory |
|
33
|
|
|
*/ |
|
34
|
|
|
private $titleFactory; |
|
35
|
|
|
|
|
36
|
|
|
/** |
|
37
|
|
|
* @var PageUpdater |
|
38
|
|
|
*/ |
|
39
|
|
|
private $updater; |
|
40
|
|
|
|
|
41
|
|
|
/** |
|
42
|
|
|
* @var ChangeRunCoalescer |
|
43
|
|
|
*/ |
|
44
|
|
|
private $changeRunCoalescer; |
|
45
|
|
|
|
|
46
|
|
|
/** |
|
47
|
|
|
* @var SiteLookup |
|
48
|
|
|
*/ |
|
49
|
|
|
private $siteLookup; |
|
50
|
|
|
|
|
51
|
|
|
/** |
|
52
|
|
|
* @var LoggerInterface |
|
53
|
|
|
*/ |
|
54
|
|
|
private $logger; |
|
55
|
|
|
|
|
56
|
|
|
/** |
|
57
|
|
|
* @var bool |
|
58
|
|
|
*/ |
|
59
|
|
|
private $injectRecentChanges; |
|
60
|
|
|
|
|
61
|
|
|
/** |
|
62
|
|
|
* @param AffectedPagesFinder $affectedPagesFinder |
|
63
|
|
|
* @param TitleFactory $titleFactory |
|
64
|
|
|
* @param PageUpdater $updater |
|
65
|
|
|
* @param ChangeRunCoalescer $changeRunCoalescer |
|
66
|
|
|
* @param SiteLookup $siteLookup |
|
67
|
|
|
* @param LoggerInterface $logger |
|
68
|
|
|
* @param bool $injectRecentChanges |
|
69
|
|
|
* |
|
70
|
|
|
* @throws InvalidArgumentException |
|
71
|
|
|
*/ |
|
72
|
|
|
public function __construct( |
|
73
|
|
|
AffectedPagesFinder $affectedPagesFinder, |
|
74
|
|
|
TitleFactory $titleFactory, |
|
75
|
|
|
PageUpdater $updater, |
|
76
|
|
|
ChangeRunCoalescer $changeRunCoalescer, |
|
77
|
|
|
SiteLookup $siteLookup, |
|
78
|
|
|
LoggerInterface $logger, |
|
79
|
|
|
$injectRecentChanges = true |
|
80
|
|
|
) { |
|
81
|
|
|
if ( !is_bool( $injectRecentChanges ) ) { |
|
82
|
|
|
throw new InvalidArgumentException( '$injectRecentChanges must be a bool' ); |
|
83
|
|
|
} |
|
84
|
|
|
|
|
85
|
|
|
$this->affectedPagesFinder = $affectedPagesFinder; |
|
86
|
|
|
$this->titleFactory = $titleFactory; |
|
87
|
|
|
$this->updater = $updater; |
|
88
|
|
|
$this->changeRunCoalescer = $changeRunCoalescer; |
|
89
|
|
|
$this->siteLookup = $siteLookup; |
|
90
|
|
|
$this->logger = $logger; |
|
91
|
|
|
$this->injectRecentChanges = $injectRecentChanges; |
|
92
|
|
|
} |
|
93
|
|
|
|
|
94
|
|
|
/** |
|
95
|
|
|
* @param EntityChange[] $changes |
|
96
|
|
|
* @param array $rootJobParams any relevant root job parameters to be inherited by new jobs. |
|
97
|
|
|
*/ |
|
98
|
|
|
public function handleChanges( array $changes, array $rootJobParams = [] ) { |
|
99
|
|
|
$changes = $this->changeRunCoalescer->transformChangeList( $changes ); |
|
100
|
|
|
|
|
101
|
|
|
if ( !Hooks::run( 'WikibaseHandleChanges', [ $changes, $rootJobParams ] ) ) { |
|
102
|
|
|
return; |
|
103
|
|
|
} |
|
104
|
|
|
|
|
105
|
|
|
foreach ( $changes as $change ) { |
|
106
|
|
|
if ( !Hooks::run( 'WikibaseHandleChange', [ $change, $rootJobParams ] ) ) { |
|
107
|
|
|
continue; |
|
108
|
|
|
} |
|
109
|
|
|
|
|
110
|
|
|
$this->handleChange( $change, $rootJobParams ); |
|
111
|
|
|
} |
|
112
|
|
|
} |
|
113
|
|
|
|
|
114
|
|
|
/** |
|
115
|
|
|
* Main entry point for handling changes |
|
116
|
|
|
* |
|
117
|
|
|
* @todo process multiple changes at once! |
|
118
|
|
|
* |
|
119
|
|
|
* @param EntityChange $change |
|
120
|
|
|
* @param array $rootJobParams any relevant root job parameters to be inherited by new jobs. |
|
121
|
|
|
*/ |
|
122
|
|
|
public function handleChange( EntityChange $change, array $rootJobParams = [] ) { |
|
123
|
|
|
$changeId = $this->getChangeIdForLog( $change ); |
|
124
|
|
|
|
|
125
|
|
|
$this->logger->debug( |
|
126
|
|
|
'{method}: handling change #{changeId} ({changeType})', |
|
127
|
|
|
[ |
|
128
|
|
|
'method' => __METHOD__, |
|
129
|
|
|
'changeId' => $changeId, |
|
130
|
|
|
'changeType' => $change->getType(), |
|
131
|
|
|
] |
|
132
|
|
|
); |
|
133
|
|
|
|
|
134
|
|
|
$usagesPerPage = $this->affectedPagesFinder->getAffectedUsagesByPage( $change ); |
|
135
|
|
|
|
|
136
|
|
|
$this->logger->debug( |
|
137
|
|
|
'{method}: updating {pageCount} page(s) for change #{changeId}.', |
|
138
|
|
|
[ |
|
139
|
|
|
'method' => __METHOD__, |
|
140
|
|
|
'changeId' => $changeId, |
|
141
|
|
|
'pageCount' => count( $usagesPerPage ), |
|
142
|
|
|
] |
|
143
|
|
|
); |
|
144
|
|
|
|
|
145
|
|
|
// if no usages we can abort early |
|
146
|
|
|
if ( $usagesPerPage === [] ) { |
|
147
|
|
|
return; |
|
148
|
|
|
} |
|
149
|
|
|
|
|
150
|
|
|
// Run all updates on all affected pages |
|
151
|
|
|
$titlesToUpdate = $this->getTitlesForUsages( $usagesPerPage ); |
|
152
|
|
|
|
|
153
|
|
|
// if no titles we can abort early |
|
154
|
|
|
if ( $titlesToUpdate === [] ) { |
|
155
|
|
|
return; |
|
156
|
|
|
} |
|
157
|
|
|
|
|
158
|
|
|
// NOTE: deduplicate |
|
159
|
|
|
$titleBatchSignature = $this->getTitleBatchSignature( $titlesToUpdate ); |
|
160
|
|
|
$rootJobParams['rootJobSignature'] = $titleBatchSignature; |
|
161
|
|
|
|
|
162
|
|
|
if ( !isset( $rootJobParams['rootJobTimestamp'] ) ) { |
|
163
|
|
|
$rootJobParams['rootJobTimestamp'] = wfTimestampNow(); |
|
164
|
|
|
} |
|
165
|
|
|
|
|
166
|
|
|
$this->updater->purgeWebCache( |
|
167
|
|
|
$titlesToUpdate, |
|
168
|
|
|
$rootJobParams, |
|
169
|
|
|
$change->getAction(), |
|
170
|
|
|
$change->hasField( 'user_id' ) ? 'uid:' . $change->getUserId() : 'uid:?' |
|
171
|
|
|
); |
|
172
|
|
|
|
|
173
|
|
|
// Removing root job timestamp to make it work: T233520 |
|
174
|
|
|
$refreshLinksRootParams = $rootJobParams; |
|
175
|
|
|
unset( $refreshLinksRootParams['rootJobTimestamp'] ); |
|
176
|
|
|
|
|
177
|
|
|
$this->updater->scheduleRefreshLinks( |
|
178
|
|
|
$titlesToUpdate, |
|
179
|
|
|
$refreshLinksRootParams, |
|
180
|
|
|
$change->getAction(), |
|
181
|
|
|
'uid:' . ( $change->getUserId() ?: '?' ) |
|
182
|
|
|
); |
|
183
|
|
|
|
|
184
|
|
|
// NOTE: signature depends on change ID, effectively disabling deduplication |
|
185
|
|
|
$changeSignature = $this->getChangeSignature( $change ); |
|
186
|
|
|
$rootJobParams['rootJobSignature'] = $titleBatchSignature . '&' . $changeSignature; |
|
187
|
|
|
if ( $this->injectRecentChanges ) { |
|
188
|
|
|
$this->updater->injectRCRecords( $titlesToUpdate, $change, $rootJobParams ); |
|
189
|
|
|
} |
|
190
|
|
|
} |
|
191
|
|
|
|
|
192
|
|
|
/** |
|
193
|
|
|
* @param Title[] $titles |
|
194
|
|
|
* |
|
195
|
|
|
* @return string a signature based on the hash of the given titles |
|
196
|
|
|
*/ |
|
197
|
|
|
private function getTitleBatchSignature( array $titles ) { |
|
198
|
|
|
$pages = []; |
|
199
|
|
|
|
|
200
|
|
|
/** @see WikiPageUpdater::getPageParamForRefreshLinksJob */ |
|
201
|
|
|
foreach ( $titles as $title ) { |
|
202
|
|
|
$id = $title->getArticleID(); |
|
203
|
|
|
$pages[$id] = [ $title->getNamespace(), $title->getDBkey() ]; |
|
204
|
|
|
} |
|
205
|
|
|
|
|
206
|
|
|
ksort( $pages ); |
|
207
|
|
|
return 'title-batch:' . sha1( json_encode( $pages ) ); |
|
208
|
|
|
} |
|
209
|
|
|
|
|
210
|
|
|
/** |
|
211
|
|
|
* @param EntityChange $change |
|
212
|
|
|
* |
|
213
|
|
|
* @return string a signature representing the change's identity. |
|
214
|
|
|
*/ |
|
215
|
|
|
private function getChangeSignature( EntityChange $change ) { |
|
216
|
|
|
if ( $change->getId() ) { |
|
|
|
|
|
|
217
|
|
|
return 'change-id:' . $change->getId(); |
|
218
|
|
|
} else { |
|
219
|
|
|
// synthetic change! |
|
220
|
|
|
$changeData = $change->getFields(); |
|
221
|
|
|
|
|
222
|
|
|
if ( isset( $changeData['info']['change-ids'] ) ) { |
|
223
|
|
|
$ids = $changeData['info']['change-ids']; |
|
224
|
|
|
sort( $ids ); |
|
225
|
|
|
return 'change-batch:' . implode( ',', $ids ); |
|
226
|
|
|
} else { |
|
227
|
|
|
ksort( $changeData ); |
|
228
|
|
|
return 'change-hash:' . sha1( json_encode( $changeData ) ); |
|
229
|
|
|
} |
|
230
|
|
|
} |
|
231
|
|
|
} |
|
232
|
|
|
|
|
233
|
|
|
/** |
|
234
|
|
|
* @param PageEntityUsages[] $usagesPerPage |
|
235
|
|
|
* |
|
236
|
|
|
* @return Title[] |
|
237
|
|
|
*/ |
|
238
|
|
|
private function getTitlesForUsages( $usagesPerPage ) { |
|
239
|
|
|
$pageIds = []; |
|
240
|
|
|
|
|
241
|
|
|
foreach ( $usagesPerPage as $usages ) { |
|
242
|
|
|
$pageIds[] = $usages->getPageId(); |
|
243
|
|
|
} |
|
244
|
|
|
|
|
245
|
|
|
return $this->titleFactory->newFromIDs( $pageIds ); |
|
246
|
|
|
} |
|
247
|
|
|
|
|
248
|
|
|
/** |
|
249
|
|
|
* Returns a human readable change ID, containing multiple IDs in case of a |
|
250
|
|
|
* coalesced change. |
|
251
|
|
|
* |
|
252
|
|
|
* @param Change $change |
|
253
|
|
|
* |
|
254
|
|
|
* @return string |
|
255
|
|
|
*/ |
|
256
|
|
|
private function getChangeIdForLog( Change $change ) { |
|
257
|
|
|
if ( $change instanceof EntityChange ) { |
|
258
|
|
|
$info = $change->getInfo(); |
|
259
|
|
|
|
|
260
|
|
|
if ( isset( $info['change-ids'] ) ) { |
|
261
|
|
|
return implode( '|', $info['change-ids'] ); |
|
262
|
|
|
} |
|
263
|
|
|
} |
|
264
|
|
|
|
|
265
|
|
|
return $change->getId(); |
|
266
|
|
|
} |
|
267
|
|
|
|
|
268
|
|
|
} |
|
269
|
|
|
|
In PHP, under loose comparison (like
==, or!=, orswitchconditions), values of different types might be equal.For
integervalues, zero is a special case, in particular the following results might be unexpected: