MediaWiki REL1_35
refreshLinks.php
Go to the documentation of this file.
1<?php
27
28require_once __DIR__ . '/Maintenance.php';
29
36 private const REPORTING_INTERVAL = 100;
37
39 protected $namespace = false;
40
41 public function __construct() {
42 parent::__construct();
43 $this->addDescription( 'Refresh link tables' );
44 $this->addOption( 'dfn-only', 'Delete links from nonexistent articles only' );
45 $this->addOption( 'new-only', 'Only affect articles with just a single edit' );
46 $this->addOption( 'redirects-only', 'Only fix redirects, not all links' );
47 $this->addOption( 'old-redirects-only', 'Only fix redirects with no redirect table entry' );
48 $this->addOption( 'e', 'Last page id to refresh', false, true );
49 $this->addOption( 'dfn-chunk-size', 'Maximum number of existent IDs to check per ' .
50 'query, default 100000', false, true );
51 $this->addOption( 'namespace', 'Only fix pages in this namespace', false, true );
52 $this->addOption( 'category', 'Only fix pages in this category', false, true );
53 $this->addOption( 'tracking-category', 'Only fix pages in this tracking category', false, true );
54 $this->addArg( 'start', 'Page_id to start from, default 1', false );
55 $this->setBatchSize( 100 );
56 }
57
58 public function execute() {
59 // Note that there is a difference between not specifying the start
60 // and end IDs and using the minimum and maximum values from the page
61 // table. In the latter case, deleteLinksFromNonexistent() will not
62 // delete entries for nonexistent IDs that fall outside the range.
63 $start = (int)$this->getArg( 0 ) ?: null;
64 $end = (int)$this->getOption( 'e' ) ?: null;
65 $dfnChunkSize = (int)$this->getOption( 'dfn-chunk-size', 100000 );
66 $ns = $this->getOption( 'namespace' );
67 if ( $ns === null ) {
68 $this->namespace = false;
69 } else {
70 $this->namespace = (int)$ns;
71 }
72 if ( ( $category = $this->getOption( 'category', false ) ) !== false ) {
73 $title = Title::makeTitleSafe( NS_CATEGORY, $category );
74 if ( !$title ) {
75 $this->fatalError( "'$category' is an invalid category name!\n" );
76 }
77 $this->refreshCategory( $title );
78 } elseif ( ( $category = $this->getOption( 'tracking-category', false ) ) !== false ) {
79 $this->refreshTrackingCategory( $category );
80 } elseif ( !$this->hasOption( 'dfn-only' ) ) {
81 $new = $this->hasOption( 'new-only' );
82 $redir = $this->hasOption( 'redirects-only' );
83 $oldRedir = $this->hasOption( 'old-redirects-only' );
84 $this->doRefreshLinks( $start, $new, $end, $redir, $oldRedir );
85 $this->deleteLinksFromNonexistent( null, null, $this->getBatchSize(), $dfnChunkSize );
86 } else {
87 $this->deleteLinksFromNonexistent( $start, $end, $this->getBatchSize(), $dfnChunkSize );
88 }
89 }
90
91 private function namespaceCond() {
92 return $this->namespace !== false
93 ? [ 'page_namespace' => $this->namespace ]
94 : [];
95 }
96
105 private function doRefreshLinks( $start, $newOnly = false,
106 $end = null, $redirectsOnly = false, $oldRedirectsOnly = false
107 ) {
108 $dbr = $this->getDB( DB_REPLICA, [ 'vslow' ] );
109
110 if ( $start === null ) {
111 $start = 1;
112 }
113
114 // Give extensions a chance to optimize settings
115 $this->getHookRunner()->onMaintenanceRefreshLinksInit( $this );
116
117 $what = $redirectsOnly ? "redirects" : "links";
118 $lbFactory = MediaWikiServices::getInstance()->getDBLoadBalancerFactory();
119
120 if ( $oldRedirectsOnly ) {
121 # This entire code path is cut-and-pasted from below. Hurrah.
122
123 $conds = [
124 "page_is_redirect=1",
125 "rd_from IS NULL",
126 self::intervalCond( $dbr, 'page_id', $start, $end ),
127 ] + $this->namespaceCond();
128
129 $res = $dbr->select(
130 [ 'page', 'redirect' ],
131 'page_id',
132 $conds,
133 __METHOD__,
134 [],
135 [ 'redirect' => [ "LEFT JOIN", "page_id=rd_from" ] ]
136 );
137 $num = $res->numRows();
138 $this->output( "Refreshing $num old redirects from $start...\n" );
139
140 $i = 0;
141
142 foreach ( $res as $row ) {
143 if ( !( ++$i % self::REPORTING_INTERVAL ) ) {
144 $this->output( "$i\n" );
145 $lbFactory->waitForReplication();
146 }
147 $this->fixRedirect( $row->page_id );
148 }
149 } elseif ( $newOnly ) {
150 $this->output( "Refreshing $what from " );
151 $res = $dbr->select( 'page',
152 [ 'page_id' ],
153 [
154 'page_is_new' => 1,
155 self::intervalCond( $dbr, 'page_id', $start, $end ),
156 ] + $this->namespaceCond(),
157 __METHOD__
158 );
159 $num = $res->numRows();
160 $this->output( "$num new articles...\n" );
161
162 $i = 0;
163 foreach ( $res as $row ) {
164 if ( !( ++$i % self::REPORTING_INTERVAL ) ) {
165 $this->output( "$i\n" );
166 $lbFactory->waitForReplication();
167 }
168 if ( $redirectsOnly ) {
169 $this->fixRedirect( $row->page_id );
170 } else {
171 self::fixLinksFromArticle( $row->page_id, $this->namespace );
172 }
173 }
174 } else {
175 if ( !$end ) {
176 $maxPage = $dbr->selectField( 'page', 'max(page_id)', '', __METHOD__ );
177 $maxRD = $dbr->selectField( 'redirect', 'max(rd_from)', '', __METHOD__ );
178 $end = max( $maxPage, $maxRD );
179 }
180 $this->output( "Refreshing redirects table.\n" );
181 $this->output( "Starting from page_id $start of $end.\n" );
182
183 for ( $id = $start; $id <= $end; $id++ ) {
184 if ( !( $id % self::REPORTING_INTERVAL ) ) {
185 $this->output( "$id\n" );
186 $lbFactory->waitForReplication();
187 }
188 $this->fixRedirect( $id );
189 }
190
191 if ( !$redirectsOnly ) {
192 $this->output( "Refreshing links tables.\n" );
193 $this->output( "Starting from page_id $start of $end.\n" );
194
195 for ( $id = $start; $id <= $end; $id++ ) {
196 if ( !( $id % self::REPORTING_INTERVAL ) ) {
197 $this->output( "$id\n" );
198 $lbFactory->waitForReplication();
199 }
200 self::fixLinksFromArticle( $id, $this->namespace );
201 }
202 }
203 }
204 }
205
218 private function fixRedirect( $id ) {
219 $page = WikiPage::newFromID( $id );
220 $dbw = $this->getDB( DB_MASTER );
221
222 if ( $page === null ) {
223 // This page doesn't exist (any more)
224 // Delete any redirect table entry for it
225 $dbw->delete( 'redirect', [ 'rd_from' => $id ],
226 __METHOD__ );
227
228 return;
229 } elseif ( $this->namespace !== false
230 && !$page->getTitle()->inNamespace( $this->namespace )
231 ) {
232 return;
233 }
234
235 $rt = null;
236 $content = $page->getContent( RevisionRecord::RAW );
237 if ( $content !== null ) {
238 $rt = $content->getUltimateRedirectTarget();
239 }
240
241 if ( $rt === null ) {
242 // The page is not a redirect
243 // Delete any redirect table entry for it
244 $dbw->delete( 'redirect', [ 'rd_from' => $id ], __METHOD__ );
245 $fieldValue = 0;
246 } else {
247 $page->insertRedirectEntry( $rt );
248 $fieldValue = 1;
249 }
250
251 // Update the page table to be sure it is an a consistent state
252 $dbw->update( 'page', [ 'page_is_redirect' => $fieldValue ],
253 [ 'page_id' => $id ], __METHOD__ );
254 }
255
261 public static function fixLinksFromArticle( $id, $ns = false ) {
262 $page = WikiPage::newFromID( $id );
263
264 MediaWikiServices::getInstance()->getLinkCache()->clear();
265
266 if ( $page === null ) {
267 return;
268 } elseif ( $ns !== false
269 && !$page->getTitle()->inNamespace( $ns ) ) {
270 return;
271 }
272
273 // Defer updates to post-send but then immediately execute deferred updates;
274 // this is the simplest way to run all updates immediately (including updates
275 // scheduled by other updates).
276 $page->doSecondaryDataUpdates( [
277 'defer' => DeferredUpdates::POSTSEND,
278 'recursive' => false,
279 ] );
280 DeferredUpdates::doUpdates();
281 }
282
294 private function deleteLinksFromNonexistent( $start = null, $end = null, $batchSize = 100,
295 $chunkSize = 100000
296 ) {
297 MediaWikiServices::getInstance()->getDBLoadBalancerFactory()->waitForReplication();
298 $this->output( "Deleting illegal entries from the links tables...\n" );
299 $dbr = $this->getDB( DB_REPLICA, [ 'vslow' ] );
300 do {
301 // Find the start of the next chunk. This is based only
302 // on existent page_ids.
303 $nextStart = $dbr->selectField(
304 'page',
305 'page_id',
306 [ self::intervalCond( $dbr, 'page_id', $start, $end ) ]
307 + $this->namespaceCond(),
308 __METHOD__,
309 [ 'ORDER BY' => 'page_id', 'OFFSET' => $chunkSize ]
310 );
311
312 if ( $nextStart !== false ) {
313 // To find the end of the current chunk, subtract one.
314 // This will serve to limit the number of rows scanned in
315 // dfnCheckInterval(), per query, to at most the sum of
316 // the chunk size and deletion batch size.
317 $chunkEnd = $nextStart - 1;
318 } else {
319 // This is the last chunk. Check all page_ids up to $end.
320 $chunkEnd = $end;
321 }
322
323 $fmtStart = $start !== null ? "[$start" : '(-INF';
324 $fmtChunkEnd = $chunkEnd !== null ? "$chunkEnd]" : 'INF)';
325 $this->output( " Checking interval $fmtStart, $fmtChunkEnd\n" );
326 $this->dfnCheckInterval( $start, $chunkEnd, $batchSize );
327
328 $start = $nextStart;
329
330 } while ( $nextStart !== false );
331 }
332
339 private function dfnCheckInterval( $start = null, $end = null, $batchSize = 100 ) {
340 $dbw = $this->getDB( DB_MASTER );
341 $dbr = $this->getDB( DB_REPLICA, [ 'vslow' ] );
342
343 $linksTables = [
344 // table name => page_id field
345 'pagelinks' => 'pl_from',
346 'imagelinks' => 'il_from',
347 'categorylinks' => 'cl_from',
348 'templatelinks' => 'tl_from',
349 'externallinks' => 'el_from',
350 'iwlinks' => 'iwl_from',
351 'langlinks' => 'll_from',
352 'redirect' => 'rd_from',
353 'page_props' => 'pp_page',
354 ];
355
356 $lbFactory = MediaWikiServices::getInstance()->getDBLoadBalancerFactory();
357 foreach ( $linksTables as $table => $field ) {
358 $this->output( " $table: 0" );
359 $tableStart = $start;
360 $counter = 0;
361 do {
362 $ids = $dbr->selectFieldValues(
363 $table,
364 $field,
365 [
366 self::intervalCond( $dbr, $field, $tableStart, $end ),
367 "$field NOT IN ({$dbr->selectSQLText( 'page', 'page_id', [], __METHOD__ )})",
368 ],
369 __METHOD__,
370 [ 'DISTINCT', 'ORDER BY' => $field, 'LIMIT' => $batchSize ]
371 );
372
373 $numIds = count( $ids );
374 if ( $numIds ) {
375 $counter += $numIds;
376 $dbw->delete( $table, [ $field => $ids ], __METHOD__ );
377 $this->output( ", $counter" );
378 $tableStart = $ids[$numIds - 1] + 1;
379 $lbFactory->waitForReplication();
380 }
381
382 } while ( $numIds >= $batchSize && ( $end === null || $tableStart <= $end ) );
383
384 $this->output( " deleted.\n" );
385 }
386 }
387
400 private static function intervalCond( IDatabase $db, $var, $start, $end ) {
401 if ( $start === null && $end === null ) {
402 return "$var IS NOT NULL";
403 } elseif ( $end === null ) {
404 return "$var >= {$db->addQuotes( $start )}";
405 } elseif ( $start === null ) {
406 return "$var <= {$db->addQuotes( $end )}";
407 } else {
408 return "$var BETWEEN {$db->addQuotes( $start )} AND {$db->addQuotes( $end )}";
409 }
410 }
411
417 private function refreshTrackingCategory( $category ) {
418 $cats = $this->getPossibleCategories( $category );
419
420 if ( !$cats ) {
421 $this->error( "Tracking category '$category' is disabled\n" );
422 // Output to stderr but don't bail out,
423 }
424
425 foreach ( $cats as $cat ) {
426 $this->refreshCategory( $cat );
427 }
428 }
429
435 private function refreshCategory( Title $category ) {
436 $this->output( "Refreshing pages in category '{$category->getText()}'...\n" );
437
438 $dbr = $this->getDB( DB_REPLICA );
439 $conds = [
440 'page_id=cl_from',
441 'cl_to' => $category->getDBkey(),
442 ];
443 if ( $this->namespace !== false ) {
444 $conds['page_namespace'] = $this->namespace;
445 }
446
447 $i = 0;
448 $timestamp = '';
449 $lastId = 0;
450 $lbFactory = MediaWikiServices::getInstance()->getDBLoadBalancerFactory();
451 do {
452 $finalConds = $conds;
453 $timestamp = $dbr->addQuotes( $timestamp );
454 $finalConds[] =
455 "(cl_timestamp > $timestamp OR (cl_timestamp = $timestamp AND cl_from > $lastId))";
456 $res = $dbr->select( [ 'page', 'categorylinks' ],
457 [ 'page_id', 'cl_timestamp' ],
458 $finalConds,
459 __METHOD__,
460 [
461 'ORDER BY' => [ 'cl_timestamp', 'cl_from' ],
462 'LIMIT' => $this->getBatchSize(),
463 ]
464 );
465
466 foreach ( $res as $row ) {
467 if ( !( ++$i % self::REPORTING_INTERVAL ) ) {
468 $this->output( "$i\n" );
469 $lbFactory->waitForReplication();
470 }
471 $lastId = $row->page_id;
472 $timestamp = $row->cl_timestamp;
473 self::fixLinksFromArticle( $row->page_id );
474 }
475
476 } while ( $res->numRows() == $this->getBatchSize() );
477 }
478
485 private function getPossibleCategories( $categoryKey ) {
486 $trackingCategories = new TrackingCategories( $this->getConfig() );
487 $cats = $trackingCategories->getTrackingCategories();
488 if ( isset( $cats[$categoryKey] ) ) {
489 return $cats[$categoryKey]['cats'];
490 }
491 $this->fatalError( "Unknown tracking category {$categoryKey}\n" );
492 }
493}
494
495$maintClass = RefreshLinks::class;
496require_once RUN_MAINTENANCE_IF_MAIN;
getDB()
const RUN_MAINTENANCE_IF_MAIN
Abstract maintenance class for quickly writing and churning out maintenance scripts with minimal effo...
error( $err, $die=0)
Throw an error to the user.
addArg( $arg, $description, $required=true)
Add some args that are needed.
output( $out, $channel=null)
Throw some output to the user.
getHookRunner()
Get a HookRunner for running core hooks.
hasOption( $name)
Checks to see if a particular option was set.
getBatchSize()
Returns batch size.
getArg( $argId=0, $default=null)
Get an argument.
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.
setBatchSize( $s=0)
Set the batch size.
fatalError( $msg, $exitCode=1)
Output a message and terminate the current script.
MediaWikiServices is the service locator for the application scope of MediaWiki.
Page revision base class.
Represents a title within MediaWiki.
Definition Title.php:42
getDBkey()
Get the main part with underscores.
Definition Title.php:1032
This class performs some operations related to tracking categories, such as creating a list of all su...
const NS_CATEGORY
Definition Defines.php:84
Basic database interface for live and lazy-loaded relation database handles.
Definition IDatabase.php:38
const REPORTING_INTERVAL
const DB_REPLICA
Definition defines.php:25
const DB_MASTER
Definition defines.php:29
$content
Definition router.php:76