|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
declare( strict_types = 1 ); |
|
4
|
|
|
|
|
5
|
|
|
namespace Wikibase\Client\ChangeModification; |
|
6
|
|
|
|
|
7
|
|
|
use Job; |
|
8
|
|
|
use Wikibase\Client\RecentChanges\RecentChangeFactory; |
|
9
|
|
|
use Wikibase\Lib\Changes\RepoRevisionIdentifier; |
|
10
|
|
|
use Wikibase\Lib\Changes\RepoRevisionIdentifierFactory; |
|
11
|
|
|
use Wikimedia\Assert\Assert; |
|
12
|
|
|
use Wikimedia\Rdbms\ILoadBalancer; |
|
13
|
|
|
use Wikimedia\Rdbms\IResultWrapper; |
|
14
|
|
|
|
|
15
|
|
|
/** |
|
16
|
|
|
* Base class for Jobs handling modifications to a set of client changes (identified |
|
17
|
|
|
* by RepoRevisionIdentifiers). |
|
18
|
|
|
* |
|
19
|
|
|
* @license GPL-2.0-or-later |
|
20
|
|
|
* @author Marius Hoch |
|
21
|
|
|
*/ |
|
22
|
|
|
abstract class ChangeModificationNotificationJob extends Job { |
|
23
|
|
|
|
|
24
|
|
|
/** |
|
25
|
|
|
* @var ILoadBalancer |
|
26
|
|
|
*/ |
|
27
|
|
|
protected $loadBalancer; |
|
28
|
|
|
|
|
29
|
|
|
/** |
|
30
|
|
|
* @var RepoRevisionIdentifier[] |
|
31
|
|
|
*/ |
|
32
|
|
|
private $revisionIdentifiers; |
|
33
|
|
|
|
|
34
|
|
|
/** |
|
35
|
|
|
* @param string $jobName Name of this job. |
|
36
|
|
|
* @param ILoadBalancer $loadBalancer |
|
37
|
|
|
* @param array $params Contains the revisionIdentifiersJson to act upon. |
|
38
|
|
|
*/ |
|
39
|
|
|
public function __construct( string $jobName, ILoadBalancer $loadBalancer, array $params = [] ) { |
|
40
|
|
|
parent::__construct( $jobName, $params ); |
|
41
|
|
|
|
|
42
|
|
|
Assert::parameterType( 'array', $params, '$params' ); |
|
43
|
|
|
Assert::parameter( |
|
44
|
|
|
is_string( $params['revisionIdentifiersJson'] ) && $params['revisionIdentifiersJson'] !== '', |
|
45
|
|
|
'$params', |
|
46
|
|
|
'$params[\'revisionIdentifiersJson\'] must be a non-empty string.' |
|
47
|
|
|
); |
|
48
|
|
|
|
|
49
|
|
|
$this->revisionIdentifiers = $this->unpackRevisionIdentifiers( $params['revisionIdentifiersJson'] ); |
|
50
|
|
|
|
|
51
|
|
|
$this->loadBalancer = $loadBalancer; |
|
52
|
|
|
} |
|
53
|
|
|
|
|
54
|
|
|
/** |
|
55
|
|
|
* @param string $revisionIdentifiersJson |
|
56
|
|
|
* |
|
57
|
|
|
* @return RepoRevisionIdentifier[] |
|
58
|
|
|
*/ |
|
59
|
|
|
private function unpackRevisionIdentifiers( string $revisionIdentifiersJson ): array { |
|
60
|
|
|
$repoRevisionFactory = new RepoRevisionIdentifierFactory(); |
|
61
|
|
|
$revisionIdentifiersArray = json_decode( $revisionIdentifiersJson, true ); |
|
62
|
|
|
$revisionIdentifiers = []; |
|
63
|
|
|
foreach ( $revisionIdentifiersArray as $revisionIdentifierArray ) { |
|
64
|
|
|
$revisionIdentifiers[] = $repoRevisionFactory->newFromArray( $revisionIdentifierArray ); |
|
65
|
|
|
} |
|
66
|
|
|
|
|
67
|
|
|
return $revisionIdentifiers; |
|
68
|
|
|
} |
|
69
|
|
|
|
|
70
|
|
|
/** |
|
71
|
|
|
* @return bool success |
|
72
|
|
|
*/ |
|
73
|
|
|
public function run() { |
|
74
|
|
|
$relevantChanges = $this->getRelevantRecentChanges(); |
|
75
|
|
|
if ( $relevantChanges === [] ) { |
|
76
|
|
|
return true; |
|
77
|
|
|
} |
|
78
|
|
|
|
|
79
|
|
|
$this->modifyChanges( $relevantChanges ); |
|
80
|
|
|
|
|
81
|
|
|
return true; |
|
82
|
|
|
} |
|
83
|
|
|
|
|
84
|
|
|
/** |
|
85
|
|
|
* @param int[] $relevantChanges Ids of changes relevant for this job. |
|
86
|
|
|
* @return void |
|
87
|
|
|
*/ |
|
88
|
|
|
abstract protected function modifyChanges( array $relevantChanges ): void; |
|
89
|
|
|
|
|
90
|
|
|
/** |
|
91
|
|
|
* @return int[] Relevant rc_ids |
|
92
|
|
|
*/ |
|
93
|
|
|
protected function getRelevantRecentChanges(): array { |
|
94
|
|
|
$revisionIdentifiersByEntityId = $this->getGroupedRepoRevisionIdentifiers(); |
|
95
|
|
|
|
|
96
|
|
|
$toRedact = []; |
|
97
|
|
|
foreach ( $revisionIdentifiersByEntityId as $entityIdSerialization => $revisionIdentifiers ) { |
|
98
|
|
|
$candidateResults = $this->selectCandidateResults( |
|
99
|
|
|
$entityIdSerialization, |
|
100
|
|
|
$revisionIdentifiers |
|
101
|
|
|
); |
|
102
|
|
|
|
|
103
|
|
|
$toRedact = array_merge( |
|
104
|
|
|
$toRedact, |
|
105
|
|
|
$this->filterCandidateResults( $candidateResults, $revisionIdentifiers ) |
|
106
|
|
|
); |
|
107
|
|
|
} |
|
108
|
|
|
|
|
109
|
|
|
return $toRedact; |
|
110
|
|
|
} |
|
111
|
|
|
|
|
112
|
|
|
/** |
|
113
|
|
|
* @param string $entityIdSerialization |
|
114
|
|
|
* @param RepoRevisionIdentifier[] $revisionIdentifiers |
|
115
|
|
|
* |
|
116
|
|
|
* @return IResultWrapper |
|
117
|
|
|
*/ |
|
118
|
|
|
private function selectCandidateResults( |
|
119
|
|
|
string $entityIdSerialization, |
|
120
|
|
|
array $revisionIdentifiers |
|
121
|
|
|
): IResultWrapper { |
|
122
|
|
|
$dbr = $this->loadBalancer->getConnection( DB_REPLICA ); |
|
123
|
|
|
$rcParamPattern = $dbr->buildLike( |
|
124
|
|
|
$dbr->anyString(), |
|
125
|
|
|
'"', $entityIdSerialization, '"', |
|
126
|
|
|
$dbr->anyString() |
|
127
|
|
|
); |
|
128
|
|
|
|
|
129
|
|
|
$timestamps = []; |
|
130
|
|
|
foreach ( $revisionIdentifiers as $revisionIdentifier ) { |
|
131
|
|
|
$timestamps[] = $revisionIdentifier->getRevisionTimestamp(); |
|
132
|
|
|
} |
|
133
|
|
|
|
|
134
|
|
|
return $dbr->select( |
|
135
|
|
|
'recentchanges', |
|
136
|
|
|
[ 'rc_id', 'rc_params' ], |
|
137
|
|
|
[ |
|
138
|
|
|
'rc_timestamp' => $timestamps, |
|
139
|
|
|
"rc_params $rcParamPattern", |
|
140
|
|
|
'rc_source' => RecentChangeFactory::SRC_WIKIBASE |
|
141
|
|
|
], |
|
142
|
|
|
__METHOD__ |
|
143
|
|
|
); |
|
144
|
|
|
} |
|
145
|
|
|
|
|
146
|
|
|
/** |
|
147
|
|
|
* @param IResultWrapper $candidateResults |
|
148
|
|
|
* @param RepoRevisionIdentifier[] $revisionIdentifiers |
|
149
|
|
|
* |
|
150
|
|
|
* @return int[] |
|
151
|
|
|
*/ |
|
152
|
|
|
private function filterCandidateResults( IResultWrapper $candidateResults, array $revisionIdentifiers ): array { |
|
153
|
|
|
if ( $candidateResults->numRows() === 0 ) { |
|
154
|
|
|
return []; |
|
155
|
|
|
} |
|
156
|
|
|
$ids = []; |
|
157
|
|
|
foreach ( $revisionIdentifiers as $revisionIdentifier ) { |
|
158
|
|
|
$ids[$revisionIdentifier->getRevisionId()] = true; |
|
159
|
|
|
} |
|
160
|
|
|
|
|
161
|
|
|
$results = []; |
|
162
|
|
|
foreach ( $candidateResults as $rc ) { |
|
163
|
|
|
$metadata = $this->readRecentChangeParams( $rc->rc_params ); |
|
164
|
|
|
|
|
165
|
|
|
$rev_id = $metadata[ 'rev_id' ]; |
|
166
|
|
|
|
|
167
|
|
|
if ( isset( $ids[$rev_id] ) ) { |
|
168
|
|
|
$results[] = (int)$rc->rc_id; |
|
169
|
|
|
} |
|
170
|
|
|
} |
|
171
|
|
|
|
|
172
|
|
|
return $results; |
|
173
|
|
|
} |
|
174
|
|
|
|
|
175
|
|
|
/** |
|
176
|
|
|
* @return RepoRevisionIdentifier[][] |
|
177
|
|
|
*/ |
|
178
|
|
|
private function getGroupedRepoRevisionIdentifiers(): array { |
|
179
|
|
|
$revisionIdentifiersByEntityId = []; |
|
180
|
|
|
foreach ( $this->revisionIdentifiers as $revisionIdentifier ) { |
|
181
|
|
|
$entityIdSerialization = $revisionIdentifier->getEntityIdSerialization(); |
|
182
|
|
|
$revisionIdentifiersByEntityId[$entityIdSerialization][] = $revisionIdentifier; |
|
183
|
|
|
} |
|
184
|
|
|
|
|
185
|
|
|
return $revisionIdentifiersByEntityId; |
|
186
|
|
|
} |
|
187
|
|
|
|
|
188
|
|
|
/** |
|
189
|
|
|
* Extracts the metadata array from the value of an rc_params field. |
|
190
|
|
|
* |
|
191
|
|
|
* @param array|string $rc_params |
|
192
|
|
|
* |
|
193
|
|
|
* @return array |
|
194
|
|
|
*/ |
|
195
|
|
|
private function readRecentChangeParams( $rc_params ): array { |
|
196
|
|
|
if ( is_string( $rc_params ) ) { |
|
197
|
|
|
$rc_params = unserialize( $rc_params ); |
|
198
|
|
|
} |
|
199
|
|
|
|
|
200
|
|
|
if ( is_array( $rc_params ) && array_key_exists( 'wikibase-repo-change', $rc_params ) ) { |
|
201
|
|
|
$metadata = $rc_params['wikibase-repo-change']; |
|
202
|
|
|
} else { |
|
203
|
|
|
$metadata = []; |
|
204
|
|
|
} |
|
205
|
|
|
|
|
206
|
|
|
$metadata = array_merge( [ 'rev_id' => 0 ], $metadata ); |
|
207
|
|
|
return $metadata; |
|
208
|
|
|
} |
|
209
|
|
|
|
|
210
|
|
|
} |
|
211
|
|
|
|