MediaWiki master
populateIpChanges.php
Go to the documentation of this file.
1<?php
27require_once __DIR__ . '/Maintenance.php';
28
30use Wikimedia\IPUtils;
31
40 public function __construct() {
41 parent::__construct();
42
43 $this->addDescription( <<<TEXT
44This script will find all rows in the revision table where the user is an IP,
45and copy relevant fields to the ip_changes table. This backfilled data will
46then be available when querying for IP ranges at Special:Contributions.
47TEXT
48 );
49 $this->addOption( 'rev-id', 'The rev_id to start copying from. Default: 0', false, true );
50 $this->addOption(
51 'max-rev-id',
52 'The rev_id to stop at. Default: result of MAX(rev_id)',
53 false,
54 true
55 );
56 $this->addOption(
57 'throttle',
58 'Wait this many milliseconds after copying each batch of revisions. Default: 0',
59 false,
60 true
61 );
62 $this->addOption( 'force', 'Run regardless of whether the database says it\'s been run already' );
63 }
64
65 public function doDBUpdates() {
66 $dbw = $this->getDB( DB_PRIMARY );
67
68 if ( !$dbw->tableExists( 'ip_changes', __METHOD__ ) ) {
69 $this->fatalError( 'ip_changes table does not exist' );
70 }
71
72 $dbr = $this->getDB( DB_REPLICA, [ 'vslow' ] );
73 $throttle = intval( $this->getOption( 'throttle', 0 ) );
74 $maxRevId = intval( $this->getOption( 'max-rev-id', 0 ) );
75 $start = $this->getOption( 'rev-id', 0 );
76 $end = $maxRevId > 0
77 ? $maxRevId
78 : $dbw->newSelectQueryBuilder()
79 ->select( 'MAX(rev_id)' )
80 ->from( 'revision' )
81 ->caller( __METHOD__ )->fetchField();
82
83 if ( !$end ) {
84 $this->output( "No revisions found, aborting.\n" );
85 return true;
86 }
87
88 $blockStart = $start;
89 $attempted = 0;
90 $inserted = 0;
91
92 $this->output( "Copying IP revisions to ip_changes, from rev_id $start to rev_id $end\n" );
93
94 $actorMigration = ActorMigration::newMigration();
95 $actorQuery = $actorMigration->getJoin( 'rev_user' );
96 $revUserIsAnon = $actorMigration->isAnon( $actorQuery['fields']['rev_user'] );
97
98 while ( $blockStart <= $end ) {
99 $blockEnd = min( $blockStart + $this->getBatchSize(), $end );
100 $rows = $dbr->select(
101 [ 'revision' ] + $actorQuery['tables'],
102 [ 'rev_id', 'rev_timestamp', 'rev_user_text' => $actorQuery['fields']['rev_user_text'] ],
103 [ "rev_id BETWEEN " . (int)$blockStart . " AND " . (int)$blockEnd, $revUserIsAnon ],
104 __METHOD__,
105 [],
106 $actorQuery['joins']
107 );
108
109 $numRows = $rows->numRows();
110
111 if ( !$rows || $numRows === 0 ) {
112 $blockStart = $blockEnd + 1;
113 continue;
114 }
115
116 $this->output( "...checking $numRows revisions for IP edits that need copying, " .
117 "between rev_ids $blockStart and $blockEnd\n" );
118
119 $insertRows = [];
120 foreach ( $rows as $row ) {
121 // Make sure this is really an IP, e.g. not maintenance user or imported revision.
122 if ( IPUtils::isValid( $row->rev_user_text ) ) {
123 $insertRows[] = [
124 'ipc_rev_id' => $row->rev_id,
125 'ipc_rev_timestamp' => $row->rev_timestamp,
126 'ipc_hex' => IPUtils::toHex( $row->rev_user_text ),
127 ];
128
129 $attempted++;
130 }
131 }
132
133 if ( $insertRows ) {
134 $dbw->newInsertQueryBuilder()
135 ->insertInto( 'ip_changes' )
136 ->ignore()
137 ->rows( $insertRows )
138 ->caller( __METHOD__ )->execute();
139
140 $inserted += $dbw->affectedRows();
141 }
142
143 $this->waitForReplication();
144 usleep( $throttle * 1000 );
145
146 $blockStart = $blockEnd + 1;
147 }
148
149 $this->output( "Attempted to insert $attempted IP revisions, $inserted actually done.\n" );
150
151 return true;
152 }
153
154 protected function getUpdateKey() {
155 return 'populate ip_changes';
156 }
157}
158
159$maintClass = PopulateIpChanges::class;
160require_once RUN_MAINTENANCE_IF_MAIN;
getDB()
Class for scripts that perform database maintenance and want to log the update in updatelog so we can...
output( $out, $channel=null)
Throw some output to the user.
waitForReplication()
Wait for replica DBs to catch up.
getBatchSize()
Returns batch size.
addDescription( $text)
Set the description text.
addOption( $name, $description, $required=false, $withArg=false, $shortName=false, $multiOccurrence=false)
Add a parameter to the script.
getOption( $name, $default=null)
Get an option, or return the default.
fatalError( $msg, $exitCode=1)
Output a message and terminate the current script.
This is not intended to be a long-term part of MediaWiki; it will be deprecated and removed once acto...
Maintenance script that will find all rows in the revision table where rev_user = 0 (user is an IP),...
__construct()
Default constructor.
doDBUpdates()
Do the actual work.
getUpdateKey()
Get the update key name to go in the update log table.
const DB_REPLICA
Definition defines.php:26
const DB_PRIMARY
Definition defines.php:28