Merge "Remove parameter 'options' from hook 'SkinEditSectionLinks'"
[lhc/web/wiklou.git] / maintenance / namespaceDupes.php
1 <?php
2 /**
3 * Check for articles to fix after adding/deleting namespaces
4 *
5 * Copyright © 2005-2007 Brion Vibber <brion@pobox.com>
6 * https://www.mediawiki.org/
7 *
8 * This program is free software; you can redistribute it and/or modify
9 * it under the terms of the GNU General Public License as published by
10 * the Free Software Foundation; either version 2 of the License, or
11 * (at your option) any later version.
12 *
13 * This program is distributed in the hope that it will be useful,
14 * but WITHOUT ANY WARRANTY; without even the implied warranty of
15 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
16 * GNU General Public License for more details.
17 *
18 * You should have received a copy of the GNU General Public License along
19 * with this program; if not, write to the Free Software Foundation, Inc.,
20 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
21 * http://www.gnu.org/copyleft/gpl.html
22 *
23 * @file
24 * @ingroup Maintenance
25 */
26
27 require_once __DIR__ . '/Maintenance.php';
28
29 use MediaWiki\Linker\LinkTarget;
30 use MediaWiki\MediaWikiServices;
31 use Wikimedia\Rdbms\ResultWrapper;
32 use Wikimedia\Rdbms\IMaintainableDatabase;
33
34 /**
35 * Maintenance script that checks for articles to fix after
36 * adding/deleting namespaces.
37 *
38 * @ingroup Maintenance
39 */
40 class NamespaceDupes extends Maintenance {
41
42 /**
43 * @var IMaintainableDatabase
44 */
45 protected $db;
46
47 private $resolvablePages = 0;
48 private $totalPages = 0;
49
50 private $resolvableLinks = 0;
51 private $totalLinks = 0;
52
53 public function __construct() {
54 parent::__construct();
55 $this->addDescription( 'Find and fix pages affected by namespace addition/removal' );
56 $this->addOption( 'fix', 'Attempt to automatically fix errors' );
57 $this->addOption( 'merge', "Instead of renaming conflicts, do a history merge with " .
58 "the correct title" );
59 $this->addOption( 'add-suffix', "Dupes will be renamed with correct namespace with " .
60 "<text> appended after the article name", false, true );
61 $this->addOption( 'add-prefix', "Dupes will be renamed with correct namespace with " .
62 "<text> prepended before the article name", false, true );
63 $this->addOption( 'source-pseudo-namespace', "Move all pages with the given source " .
64 "prefix (with an implied colon following it). If --dest-namespace is not specified, " .
65 "the colon will be replaced with a hyphen.",
66 false, true );
67 $this->addOption( 'dest-namespace', "In combination with --source-pseudo-namespace, " .
68 "specify the namespace ID of the destination.", false, true );
69 $this->addOption( 'move-talk', "If this is specified, pages in the Talk namespace that " .
70 "begin with a conflicting prefix will be renamed, for example " .
71 "Talk:File:Foo -> File_Talk:Foo" );
72 }
73
74 public function execute() {
75 $this->db = $this->getDB( DB_MASTER );
76
77 $options = [
78 'fix' => $this->hasOption( 'fix' ),
79 'merge' => $this->hasOption( 'merge' ),
80 'add-suffix' => $this->getOption( 'add-suffix', '' ),
81 'add-prefix' => $this->getOption( 'add-prefix', '' ),
82 'move-talk' => $this->hasOption( 'move-talk' ),
83 'source-pseudo-namespace' => $this->getOption( 'source-pseudo-namespace', '' ),
84 'dest-namespace' => intval( $this->getOption( 'dest-namespace', 0 ) ) ];
85
86 if ( $options['source-pseudo-namespace'] !== '' ) {
87 $retval = $this->checkPrefix( $options );
88 } else {
89 $retval = $this->checkAll( $options );
90 }
91
92 if ( $retval ) {
93 $this->output( "\nLooks good!\n" );
94 } else {
95 $this->output( "\nOh noeees\n" );
96 }
97 }
98
99 /**
100 * Check all namespaces
101 *
102 * @param array $options Associative array of validated command-line options
103 *
104 * @return bool
105 */
106 private function checkAll( $options ) {
107 global $wgNamespaceAliases, $wgCapitalLinks;
108
109 $contLang = MediaWikiServices::getInstance()->getContentLanguage();
110 $spaces = [];
111
112 // List interwikis first, so they'll be overridden
113 // by any conflicting local namespaces.
114 foreach ( $this->getInterwikiList() as $prefix ) {
115 $name = $contLang->ucfirst( $prefix );
116 $spaces[$name] = 0;
117 }
118
119 // Now pull in all canonical and alias namespaces...
120 foreach (
121 MediaWikiServices::getInstance()->getNamespaceInfo()->getCanonicalNamespaces()
122 as $ns => $name
123 ) {
124 // This includes $wgExtraNamespaces
125 if ( $name !== '' ) {
126 $spaces[$name] = $ns;
127 }
128 }
129 foreach ( $contLang->getNamespaces() as $ns => $name ) {
130 if ( $name !== '' ) {
131 $spaces[$name] = $ns;
132 }
133 }
134 foreach ( $wgNamespaceAliases as $name => $ns ) {
135 $spaces[$name] = $ns;
136 }
137 foreach ( $contLang->getNamespaceAliases() as $name => $ns ) {
138 $spaces[$name] = $ns;
139 }
140
141 // We'll need to check for lowercase keys as well,
142 // since we're doing case-sensitive searches in the db.
143 foreach ( $spaces as $name => $ns ) {
144 $moreNames = [];
145 $moreNames[] = $contLang->uc( $name );
146 $moreNames[] = $contLang->ucfirst( $contLang->lc( $name ) );
147 $moreNames[] = $contLang->ucwords( $name );
148 $moreNames[] = $contLang->ucwords( $contLang->lc( $name ) );
149 $moreNames[] = $contLang->ucwordbreaks( $name );
150 $moreNames[] = $contLang->ucwordbreaks( $contLang->lc( $name ) );
151 if ( !$wgCapitalLinks ) {
152 foreach ( $moreNames as $altName ) {
153 $moreNames[] = $contLang->lcfirst( $altName );
154 }
155 $moreNames[] = $contLang->lcfirst( $name );
156 }
157 foreach ( array_unique( $moreNames ) as $altName ) {
158 if ( $altName !== $name ) {
159 $spaces[$altName] = $ns;
160 }
161 }
162 }
163
164 // Sort by namespace index, and if there are two with the same index,
165 // break the tie by sorting by name
166 $origSpaces = $spaces;
167 uksort( $spaces, function ( $a, $b ) use ( $origSpaces ) {
168 return $origSpaces[$a] <=> $origSpaces[$b]
169 ?: $a <=> $b;
170 } );
171
172 $ok = true;
173 foreach ( $spaces as $name => $ns ) {
174 $ok = $this->checkNamespace( $ns, $name, $options ) && $ok;
175 }
176
177 $this->output( "{$this->totalPages} pages to fix, " .
178 "{$this->resolvablePages} were resolvable.\n\n" );
179
180 foreach ( $spaces as $name => $ns ) {
181 if ( $ns != 0 ) {
182 /* Fix up link destinations for non-interwiki links only.
183 *
184 * For example if a page has [[Foo:Bar]] and then a Foo namespace
185 * is introduced, pagelinks needs to be updated to have
186 * page_namespace = NS_FOO.
187 *
188 * If instead an interwiki prefix was introduced called "Foo",
189 * the link should instead be moved to the iwlinks table. If a new
190 * language is introduced called "Foo", or if there is a pagelink
191 * [[fr:Bar]] when interlanguage magic links are turned on, the
192 * link would have to be moved to the langlinks table. Let's put
193 * those cases in the too-hard basket for now. The consequences are
194 * not especially severe.
195 * @fixme Handle interwiki links, and pagelinks to Category:, File:
196 * which probably need reparsing.
197 */
198
199 $this->checkLinkTable( 'pagelinks', 'pl', $ns, $name, $options );
200 $this->checkLinkTable( 'templatelinks', 'tl', $ns, $name, $options );
201
202 // The redirect table has interwiki links randomly mixed in, we
203 // need to filter those out. For example [[w:Foo:Bar]] would
204 // have rd_interwiki=w and rd_namespace=0, which would match the
205 // query for a conflicting namespace "Foo" if filtering wasn't done.
206 $this->checkLinkTable( 'redirect', 'rd', $ns, $name, $options,
207 [ 'rd_interwiki' => null ] );
208 $this->checkLinkTable( 'redirect', 'rd', $ns, $name, $options,
209 [ 'rd_interwiki' => '' ] );
210 }
211 }
212
213 $this->output( "{$this->totalLinks} links to fix, " .
214 "{$this->resolvableLinks} were resolvable.\n" );
215
216 return $ok;
217 }
218
219 /**
220 * Get the interwiki list
221 *
222 * @return array
223 */
224 private function getInterwikiList() {
225 $result = MediaWikiServices::getInstance()->getInterwikiLookup()->getAllPrefixes();
226 $prefixes = [];
227 foreach ( $result as $row ) {
228 $prefixes[] = $row['iw_prefix'];
229 }
230
231 return $prefixes;
232 }
233
234 /**
235 * Check a given prefix and try to move it into the given destination namespace
236 *
237 * @param int $ns Destination namespace id
238 * @param string $name
239 * @param array $options Associative array of validated command-line options
240 * @return bool
241 */
242 private function checkNamespace( $ns, $name, $options ) {
243 $targets = $this->getTargetList( $ns, $name, $options );
244 $count = $targets->numRows();
245 $this->totalPages += $count;
246 if ( $count == 0 ) {
247 return true;
248 }
249
250 $dryRunNote = $options['fix'] ? '' : ' DRY RUN ONLY';
251
252 $ok = true;
253 foreach ( $targets as $row ) {
254 // Find the new title and determine the action to take
255
256 $newTitle = $this->getDestinationTitle( $ns, $name,
257 $row->page_namespace, $row->page_title, $options );
258 $logStatus = false;
259 if ( !$newTitle ) {
260 $logStatus = 'invalid title';
261 $action = 'abort';
262 } elseif ( $newTitle->exists() ) {
263 if ( $options['merge'] ) {
264 if ( $this->canMerge( $row->page_id, $newTitle, $logStatus ) ) {
265 $action = 'merge';
266 } else {
267 $action = 'abort';
268 }
269 } elseif ( $options['add-prefix'] == '' && $options['add-suffix'] == '' ) {
270 $action = 'abort';
271 $logStatus = 'dest title exists and --add-prefix not specified';
272 } else {
273 $newTitle = $this->getAlternateTitle( $newTitle, $options );
274 if ( !$newTitle ) {
275 $action = 'abort';
276 $logStatus = 'alternate title is invalid';
277 } elseif ( $newTitle->exists() ) {
278 $action = 'abort';
279 $logStatus = 'title conflict';
280 } else {
281 $action = 'move';
282 $logStatus = 'alternate';
283 }
284 }
285 } else {
286 $action = 'move';
287 $logStatus = 'no conflict';
288 }
289
290 // Take the action or log a dry run message
291
292 $logTitle = "id={$row->page_id} ns={$row->page_namespace} dbk={$row->page_title}";
293 $pageOK = true;
294
295 switch ( $action ) {
296 case 'abort':
297 $this->output( "$logTitle *** $logStatus\n" );
298 $pageOK = false;
299 break;
300 case 'move':
301 $this->output( "$logTitle -> " .
302 $newTitle->getPrefixedDBkey() . " ($logStatus)$dryRunNote\n" );
303
304 if ( $options['fix'] ) {
305 $pageOK = $this->movePage( $row->page_id, $newTitle );
306 }
307 break;
308 case 'merge':
309 $this->output( "$logTitle => " .
310 $newTitle->getPrefixedDBkey() . " (merge)$dryRunNote\n" );
311
312 if ( $options['fix'] ) {
313 $pageOK = $this->mergePage( $row, $newTitle );
314 }
315 break;
316 }
317
318 if ( $pageOK ) {
319 $this->resolvablePages++;
320 } else {
321 $ok = false;
322 }
323 }
324
325 return $ok;
326 }
327
328 /**
329 * Check and repair the destination fields in a link table
330 * @param string $table The link table name
331 * @param string $fieldPrefix The field prefix in the link table
332 * @param int $ns Destination namespace id
333 * @param string $name
334 * @param array $options Associative array of validated command-line options
335 * @param array $extraConds Extra conditions for the SQL query
336 */
337 private function checkLinkTable( $table, $fieldPrefix, $ns, $name, $options,
338 $extraConds = []
339 ) {
340 $batchConds = [];
341 $fromField = "{$fieldPrefix}_from";
342 $namespaceField = "{$fieldPrefix}_namespace";
343 $titleField = "{$fieldPrefix}_title";
344 $batchSize = 500;
345 while ( true ) {
346 $res = $this->db->select(
347 $table,
348 [ $fromField, $namespaceField, $titleField ],
349 array_merge( $batchConds, $extraConds, [
350 $namespaceField => 0,
351 $titleField . $this->db->buildLike( "$name:", $this->db->anyString() )
352 ] ),
353 __METHOD__,
354 [
355 'ORDER BY' => [ $titleField, $fromField ],
356 'LIMIT' => $batchSize
357 ]
358 );
359
360 if ( $res->numRows() == 0 ) {
361 break;
362 }
363 foreach ( $res as $row ) {
364 $logTitle = "from={$row->$fromField} ns={$row->$namespaceField} " .
365 "dbk={$row->$titleField}";
366 $destTitle = $this->getDestinationTitle( $ns, $name,
367 $row->$namespaceField, $row->$titleField, $options );
368 $this->totalLinks++;
369 if ( !$destTitle ) {
370 $this->output( "$table $logTitle *** INVALID\n" );
371 continue;
372 }
373 $this->resolvableLinks++;
374 if ( !$options['fix'] ) {
375 $this->output( "$table $logTitle -> " .
376 $destTitle->getPrefixedDBkey() . " DRY RUN\n" );
377 continue;
378 }
379
380 $this->db->update( $table,
381 // SET
382 [
383 $namespaceField => $destTitle->getNamespace(),
384 $titleField => $destTitle->getDBkey()
385 ],
386 // WHERE
387 [
388 $namespaceField => 0,
389 $titleField => $row->$titleField,
390 $fromField => $row->$fromField
391 ],
392 __METHOD__,
393 [ 'IGNORE' ]
394 );
395 $this->output( "$table $logTitle -> " .
396 $destTitle->getPrefixedDBkey() . "\n" );
397 }
398 $encLastTitle = $this->db->addQuotes( $row->$titleField );
399 $encLastFrom = $this->db->addQuotes( $row->$fromField );
400
401 $batchConds = [
402 "$titleField > $encLastTitle " .
403 "OR ($titleField = $encLastTitle AND $fromField > $encLastFrom)" ];
404
405 wfWaitForSlaves();
406 }
407 }
408
409 /**
410 * Move the given pseudo-namespace, either replacing the colon with a hyphen
411 * (useful for pseudo-namespaces that conflict with interwiki links) or move
412 * them to another namespace if specified.
413 * @param array $options Associative array of validated command-line options
414 * @return bool
415 */
416 private function checkPrefix( $options ) {
417 $prefix = $options['source-pseudo-namespace'];
418 $ns = $options['dest-namespace'];
419 $this->output( "Checking prefix \"$prefix\" vs namespace $ns\n" );
420
421 return $this->checkNamespace( $ns, $prefix, $options );
422 }
423
424 /**
425 * Find pages in main and talk namespaces that have a prefix of the new
426 * namespace so we know titles that will need migrating
427 *
428 * @param int $ns Destination namespace id
429 * @param string $name Prefix that is being made a namespace
430 * @param array $options Associative array of validated command-line options
431 *
432 * @return ResultWrapper
433 */
434 private function getTargetList( $ns, $name, $options ) {
435 if (
436 $options['move-talk'] &&
437 MediaWikiServices::getInstance()->getNamespaceInfo()->isSubject( $ns )
438 ) {
439 $checkNamespaces = [ NS_MAIN, NS_TALK ];
440 } else {
441 $checkNamespaces = NS_MAIN;
442 }
443
444 return $this->db->select( 'page',
445 [
446 'page_id',
447 'page_title',
448 'page_namespace',
449 ],
450 [
451 'page_namespace' => $checkNamespaces,
452 'page_title' . $this->db->buildLike( "$name:", $this->db->anyString() ),
453 ],
454 __METHOD__
455 );
456 }
457
458 /**
459 * Get the preferred destination title for a given target page.
460 * @param int $ns The destination namespace ID
461 * @param string $name The conflicting prefix
462 * @param int $sourceNs The source namespace
463 * @param int $sourceDbk The source DB key (i.e. page_title)
464 * @param array $options Associative array of validated command-line options
465 * @return Title|false
466 */
467 private function getDestinationTitle( $ns, $name, $sourceNs, $sourceDbk, $options ) {
468 $dbk = substr( $sourceDbk, strlen( "$name:" ) );
469 if ( $ns == 0 ) {
470 // An interwiki; try an alternate encoding with '-' for ':'
471 $dbk = "$name-" . $dbk;
472 }
473 $destNS = $ns;
474 $nsInfo = MediaWikiServices::getInstance()->getNamespaceInfo();
475 if ( $sourceNs == NS_TALK && $nsInfo->isSubject( $ns ) ) {
476 // This is an associated talk page moved with the --move-talk feature.
477 $destNS = $nsInfo->getTalk( $destNS );
478 }
479 $newTitle = Title::makeTitleSafe( $destNS, $dbk );
480 if ( !$newTitle || !$newTitle->canExist() ) {
481 return false;
482 }
483 return $newTitle;
484 }
485
486 /**
487 * Get an alternative title to move a page to. This is used if the
488 * preferred destination title already exists.
489 *
490 * @param LinkTarget $linkTarget
491 * @param array $options Associative array of validated command-line options
492 * @return Title|bool
493 */
494 private function getAlternateTitle( LinkTarget $linkTarget, $options ) {
495 $prefix = $options['add-prefix'];
496 $suffix = $options['add-suffix'];
497 if ( $prefix == '' && $suffix == '' ) {
498 return false;
499 }
500 while ( true ) {
501 $dbk = $prefix . $linkTarget->getDBkey() . $suffix;
502 $title = Title::makeTitleSafe( $linkTarget->getNamespace(), $dbk );
503 if ( !$title ) {
504 return false;
505 }
506 if ( !$title->exists() ) {
507 return $title;
508 }
509 }
510 }
511
512 /**
513 * Move a page
514 *
515 * @param integer $id The page_id
516 * @param LinkTarget $newLinkTarget The new title link target
517 * @return bool
518 */
519 private function movePage( $id, LinkTarget $newLinkTarget ) {
520 $this->db->update( 'page',
521 [
522 "page_namespace" => $newLinkTarget->getNamespace(),
523 "page_title" => $newLinkTarget->getDBkey(),
524 ],
525 [
526 "page_id" => $id,
527 ],
528 __METHOD__ );
529
530 // Update *_from_namespace in links tables
531 $fromNamespaceTables = [
532 [ 'pagelinks', 'pl' ],
533 [ 'templatelinks', 'tl' ],
534 [ 'imagelinks', 'il' ] ];
535 foreach ( $fromNamespaceTables as $tableInfo ) {
536 list( $table, $fieldPrefix ) = $tableInfo;
537 $this->db->update( $table,
538 // SET
539 [ "{$fieldPrefix}_from_namespace" => $newLinkTarget->getNamespace() ],
540 // WHERE
541 [ "{$fieldPrefix}_from" => $id ],
542 __METHOD__ );
543 }
544
545 return true;
546 }
547
548 /**
549 * Determine if we can merge a page.
550 * We check if an inaccessible revision would become the latest and
551 * deny the merge if so -- it's theoretically possible to update the
552 * latest revision, but opens a can of worms -- search engine updates,
553 * recentchanges review, etc.
554 *
555 * @param integer $id The page_id
556 * @param LinkTarget $linkTarget The new link target
557 * @param string $logStatus This is set to the log status message on failure
558 * @return bool
559 */
560 private function canMerge( $id, LinkTarget $linkTarget, &$logStatus ) {
561 $latestDest = Revision::newFromTitle( $linkTarget, 0, Revision::READ_LATEST );
562 $latestSource = Revision::newFromPageId( $id, 0, Revision::READ_LATEST );
563 if ( $latestSource->getTimestamp() > $latestDest->getTimestamp() ) {
564 $logStatus = 'cannot merge since source is later';
565 return false;
566 } else {
567 return true;
568 }
569 }
570
571 /**
572 * Merge page histories
573 *
574 * @param stdClass $row Page row
575 * @param Title $newTitle The new title
576 * @return bool
577 */
578 private function mergePage( $row, Title $newTitle ) {
579 $id = $row->page_id;
580
581 // Construct the WikiPage object we will need later, while the
582 // page_id still exists. Note that this cannot use makeTitleSafe(),
583 // we are deliberately constructing an invalid title.
584 $sourceTitle = Title::makeTitle( $row->page_namespace, $row->page_title );
585 $sourceTitle->resetArticleID( $id );
586 $wikiPage = new WikiPage( $sourceTitle );
587 $wikiPage->loadPageData( 'fromdbmaster' );
588
589 $destId = $newTitle->getArticleID();
590 $this->beginTransaction( $this->db, __METHOD__ );
591 $this->db->update( 'revision',
592 // SET
593 [ 'rev_page' => $destId ],
594 // WHERE
595 [ 'rev_page' => $id ],
596 __METHOD__ );
597
598 $this->db->delete( 'page', [ 'page_id' => $id ], __METHOD__ );
599
600 $this->commitTransaction( $this->db, __METHOD__ );
601
602 /* Call LinksDeletionUpdate to delete outgoing links from the old title,
603 * and update category counts.
604 *
605 * Calling external code with a fake broken Title is a fairly dubious
606 * idea. It's necessary because it's quite a lot of code to duplicate,
607 * but that also makes it fragile since it would be easy for someone to
608 * accidentally introduce an assumption of title validity to the code we
609 * are calling.
610 */
611 DeferredUpdates::addUpdate( new LinksDeletionUpdate( $wikiPage ) );
612 DeferredUpdates::doUpdates();
613
614 return true;
615 }
616 }
617
618 $maintClass = NamespaceDupes::class;
619 require_once RUN_MAINTENANCE_IF_MAIN;