Add ability to override mb_strtoupper in Language::ucfirst
[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 ( MWNamespace::getCanonicalNamespaces() as $ns => $name ) {
121 // This includes $wgExtraNamespaces
122 if ( $name !== '' ) {
123 $spaces[$name] = $ns;
124 }
125 }
126 foreach ( $contLang->getNamespaces() as $ns => $name ) {
127 if ( $name !== '' ) {
128 $spaces[$name] = $ns;
129 }
130 }
131 foreach ( $wgNamespaceAliases as $name => $ns ) {
132 $spaces[$name] = $ns;
133 }
134 foreach ( $contLang->getNamespaceAliases() as $name => $ns ) {
135 $spaces[$name] = $ns;
136 }
137
138 // We'll need to check for lowercase keys as well,
139 // since we're doing case-sensitive searches in the db.
140 foreach ( $spaces as $name => $ns ) {
141 $moreNames = [];
142 $moreNames[] = $contLang->uc( $name );
143 $moreNames[] = $contLang->ucfirst( $contLang->lc( $name ) );
144 $moreNames[] = $contLang->ucwords( $name );
145 $moreNames[] = $contLang->ucwords( $contLang->lc( $name ) );
146 $moreNames[] = $contLang->ucwordbreaks( $name );
147 $moreNames[] = $contLang->ucwordbreaks( $contLang->lc( $name ) );
148 if ( !$wgCapitalLinks ) {
149 foreach ( $moreNames as $altName ) {
150 $moreNames[] = $contLang->lcfirst( $altName );
151 }
152 $moreNames[] = $contLang->lcfirst( $name );
153 }
154 foreach ( array_unique( $moreNames ) as $altName ) {
155 if ( $altName !== $name ) {
156 $spaces[$altName] = $ns;
157 }
158 }
159 }
160
161 // Sort by namespace index, and if there are two with the same index,
162 // break the tie by sorting by name
163 $origSpaces = $spaces;
164 uksort( $spaces, function ( $a, $b ) use ( $origSpaces ) {
165 return $origSpaces[$a] <=> $origSpaces[$b]
166 ?: $a <=> $b;
167 } );
168
169 $ok = true;
170 foreach ( $spaces as $name => $ns ) {
171 $ok = $this->checkNamespace( $ns, $name, $options ) && $ok;
172 }
173
174 $this->output( "{$this->totalPages} pages to fix, " .
175 "{$this->resolvablePages} were resolvable.\n\n" );
176
177 foreach ( $spaces as $name => $ns ) {
178 if ( $ns != 0 ) {
179 /* Fix up link destinations for non-interwiki links only.
180 *
181 * For example if a page has [[Foo:Bar]] and then a Foo namespace
182 * is introduced, pagelinks needs to be updated to have
183 * page_namespace = NS_FOO.
184 *
185 * If instead an interwiki prefix was introduced called "Foo",
186 * the link should instead be moved to the iwlinks table. If a new
187 * language is introduced called "Foo", or if there is a pagelink
188 * [[fr:Bar]] when interlanguage magic links are turned on, the
189 * link would have to be moved to the langlinks table. Let's put
190 * those cases in the too-hard basket for now. The consequences are
191 * not especially severe.
192 * @fixme Handle interwiki links, and pagelinks to Category:, File:
193 * which probably need reparsing.
194 */
195
196 $this->checkLinkTable( 'pagelinks', 'pl', $ns, $name, $options );
197 $this->checkLinkTable( 'templatelinks', 'tl', $ns, $name, $options );
198
199 // The redirect table has interwiki links randomly mixed in, we
200 // need to filter those out. For example [[w:Foo:Bar]] would
201 // have rd_interwiki=w and rd_namespace=0, which would match the
202 // query for a conflicting namespace "Foo" if filtering wasn't done.
203 $this->checkLinkTable( 'redirect', 'rd', $ns, $name, $options,
204 [ 'rd_interwiki' => null ] );
205 $this->checkLinkTable( 'redirect', 'rd', $ns, $name, $options,
206 [ 'rd_interwiki' => '' ] );
207 }
208 }
209
210 $this->output( "{$this->totalLinks} links to fix, " .
211 "{$this->resolvableLinks} were resolvable.\n" );
212
213 return $ok;
214 }
215
216 /**
217 * Get the interwiki list
218 *
219 * @return array
220 */
221 private function getInterwikiList() {
222 $result = MediaWikiServices::getInstance()->getInterwikiLookup()->getAllPrefixes();
223 $prefixes = [];
224 foreach ( $result as $row ) {
225 $prefixes[] = $row['iw_prefix'];
226 }
227
228 return $prefixes;
229 }
230
231 /**
232 * Check a given prefix and try to move it into the given destination namespace
233 *
234 * @param int $ns Destination namespace id
235 * @param string $name
236 * @param array $options Associative array of validated command-line options
237 * @return bool
238 */
239 private function checkNamespace( $ns, $name, $options ) {
240 $targets = $this->getTargetList( $ns, $name, $options );
241 $count = $targets->numRows();
242 $this->totalPages += $count;
243 if ( $count == 0 ) {
244 return true;
245 }
246
247 $dryRunNote = $options['fix'] ? '' : ' DRY RUN ONLY';
248
249 $ok = true;
250 foreach ( $targets as $row ) {
251 // Find the new title and determine the action to take
252
253 $newTitle = $this->getDestinationTitle( $ns, $name,
254 $row->page_namespace, $row->page_title, $options );
255 $logStatus = false;
256 if ( !$newTitle ) {
257 $logStatus = 'invalid title';
258 $action = 'abort';
259 } elseif ( $newTitle->exists() ) {
260 if ( $options['merge'] ) {
261 if ( $this->canMerge( $row->page_id, $newTitle, $logStatus ) ) {
262 $action = 'merge';
263 } else {
264 $action = 'abort';
265 }
266 } elseif ( $options['add-prefix'] == '' && $options['add-suffix'] == '' ) {
267 $action = 'abort';
268 $logStatus = 'dest title exists and --add-prefix not specified';
269 } else {
270 $newTitle = $this->getAlternateTitle( $newTitle, $options );
271 if ( !$newTitle ) {
272 $action = 'abort';
273 $logStatus = 'alternate title is invalid';
274 } elseif ( $newTitle->exists() ) {
275 $action = 'abort';
276 $logStatus = 'title conflict';
277 } else {
278 $action = 'move';
279 $logStatus = 'alternate';
280 }
281 }
282 } else {
283 $action = 'move';
284 $logStatus = 'no conflict';
285 }
286
287 // Take the action or log a dry run message
288
289 $logTitle = "id={$row->page_id} ns={$row->page_namespace} dbk={$row->page_title}";
290 $pageOK = true;
291
292 switch ( $action ) {
293 case 'abort':
294 $this->output( "$logTitle *** $logStatus\n" );
295 $pageOK = false;
296 break;
297 case 'move':
298 $this->output( "$logTitle -> " .
299 $newTitle->getPrefixedDBkey() . " ($logStatus)$dryRunNote\n" );
300
301 if ( $options['fix'] ) {
302 $pageOK = $this->movePage( $row->page_id, $newTitle );
303 }
304 break;
305 case 'merge':
306 $this->output( "$logTitle => " .
307 $newTitle->getPrefixedDBkey() . " (merge)$dryRunNote\n" );
308
309 if ( $options['fix'] ) {
310 $pageOK = $this->mergePage( $row, $newTitle );
311 }
312 break;
313 }
314
315 if ( $pageOK ) {
316 $this->resolvablePages++;
317 } else {
318 $ok = false;
319 }
320 }
321
322 return $ok;
323 }
324
325 /**
326 * Check and repair the destination fields in a link table
327 * @param string $table The link table name
328 * @param string $fieldPrefix The field prefix in the link table
329 * @param int $ns Destination namespace id
330 * @param string $name
331 * @param array $options Associative array of validated command-line options
332 * @param array $extraConds Extra conditions for the SQL query
333 */
334 private function checkLinkTable( $table, $fieldPrefix, $ns, $name, $options,
335 $extraConds = []
336 ) {
337 $batchConds = [];
338 $fromField = "{$fieldPrefix}_from";
339 $namespaceField = "{$fieldPrefix}_namespace";
340 $titleField = "{$fieldPrefix}_title";
341 $batchSize = 500;
342 while ( true ) {
343 $res = $this->db->select(
344 $table,
345 [ $fromField, $namespaceField, $titleField ],
346 array_merge( $batchConds, $extraConds, [
347 $namespaceField => 0,
348 $titleField . $this->db->buildLike( "$name:", $this->db->anyString() )
349 ] ),
350 __METHOD__,
351 [
352 'ORDER BY' => [ $titleField, $fromField ],
353 'LIMIT' => $batchSize
354 ]
355 );
356
357 if ( $res->numRows() == 0 ) {
358 break;
359 }
360 foreach ( $res as $row ) {
361 $logTitle = "from={$row->$fromField} ns={$row->$namespaceField} " .
362 "dbk={$row->$titleField}";
363 $destTitle = $this->getDestinationTitle( $ns, $name,
364 $row->$namespaceField, $row->$titleField, $options );
365 $this->totalLinks++;
366 if ( !$destTitle ) {
367 $this->output( "$table $logTitle *** INVALID\n" );
368 continue;
369 }
370 $this->resolvableLinks++;
371 if ( !$options['fix'] ) {
372 $this->output( "$table $logTitle -> " .
373 $destTitle->getPrefixedDBkey() . " DRY RUN\n" );
374 continue;
375 }
376
377 $this->db->update( $table,
378 // SET
379 [
380 $namespaceField => $destTitle->getNamespace(),
381 $titleField => $destTitle->getDBkey()
382 ],
383 // WHERE
384 [
385 $namespaceField => 0,
386 $titleField => $row->$titleField,
387 $fromField => $row->$fromField
388 ],
389 __METHOD__,
390 [ 'IGNORE' ]
391 );
392 $this->output( "$table $logTitle -> " .
393 $destTitle->getPrefixedDBkey() . "\n" );
394 }
395 $encLastTitle = $this->db->addQuotes( $row->$titleField );
396 $encLastFrom = $this->db->addQuotes( $row->$fromField );
397
398 $batchConds = [
399 "$titleField > $encLastTitle " .
400 "OR ($titleField = $encLastTitle AND $fromField > $encLastFrom)" ];
401
402 wfWaitForSlaves();
403 }
404 }
405
406 /**
407 * Move the given pseudo-namespace, either replacing the colon with a hyphen
408 * (useful for pseudo-namespaces that conflict with interwiki links) or move
409 * them to another namespace if specified.
410 * @param array $options Associative array of validated command-line options
411 * @return bool
412 */
413 private function checkPrefix( $options ) {
414 $prefix = $options['source-pseudo-namespace'];
415 $ns = $options['dest-namespace'];
416 $this->output( "Checking prefix \"$prefix\" vs namespace $ns\n" );
417
418 return $this->checkNamespace( $ns, $prefix, $options );
419 }
420
421 /**
422 * Find pages in main and talk namespaces that have a prefix of the new
423 * namespace so we know titles that will need migrating
424 *
425 * @param int $ns Destination namespace id
426 * @param string $name Prefix that is being made a namespace
427 * @param array $options Associative array of validated command-line options
428 *
429 * @return ResultWrapper
430 */
431 private function getTargetList( $ns, $name, $options ) {
432 if ( $options['move-talk'] && MWNamespace::isSubject( $ns ) ) {
433 $checkNamespaces = [ NS_MAIN, NS_TALK ];
434 } else {
435 $checkNamespaces = NS_MAIN;
436 }
437
438 return $this->db->select( 'page',
439 [
440 'page_id',
441 'page_title',
442 'page_namespace',
443 ],
444 [
445 'page_namespace' => $checkNamespaces,
446 'page_title' . $this->db->buildLike( "$name:", $this->db->anyString() ),
447 ],
448 __METHOD__
449 );
450 }
451
452 /**
453 * Get the preferred destination title for a given target page.
454 * @param int $ns The destination namespace ID
455 * @param string $name The conflicting prefix
456 * @param int $sourceNs The source namespace
457 * @param int $sourceDbk The source DB key (i.e. page_title)
458 * @param array $options Associative array of validated command-line options
459 * @return Title|false
460 */
461 private function getDestinationTitle( $ns, $name, $sourceNs, $sourceDbk, $options ) {
462 $dbk = substr( $sourceDbk, strlen( "$name:" ) );
463 if ( $ns == 0 ) {
464 // An interwiki; try an alternate encoding with '-' for ':'
465 $dbk = "$name-" . $dbk;
466 }
467 $destNS = $ns;
468 if ( $sourceNs == NS_TALK && MWNamespace::isSubject( $ns ) ) {
469 // This is an associated talk page moved with the --move-talk feature.
470 $destNS = MWNamespace::getTalk( $destNS );
471 }
472 $newTitle = Title::makeTitleSafe( $destNS, $dbk );
473 if ( !$newTitle || !$newTitle->canExist() ) {
474 return false;
475 }
476 return $newTitle;
477 }
478
479 /**
480 * Get an alternative title to move a page to. This is used if the
481 * preferred destination title already exists.
482 *
483 * @param LinkTarget $linkTarget
484 * @param array $options Associative array of validated command-line options
485 * @return Title|bool
486 */
487 private function getAlternateTitle( LinkTarget $linkTarget, $options ) {
488 $prefix = $options['add-prefix'];
489 $suffix = $options['add-suffix'];
490 if ( $prefix == '' && $suffix == '' ) {
491 return false;
492 }
493 while ( true ) {
494 $dbk = $prefix . $linkTarget->getDBkey() . $suffix;
495 $title = Title::makeTitleSafe( $linkTarget->getNamespace(), $dbk );
496 if ( !$title ) {
497 return false;
498 }
499 if ( !$title->exists() ) {
500 return $title;
501 }
502 }
503 }
504
505 /**
506 * Move a page
507 *
508 * @param integer $id The page_id
509 * @param LinkTarget $newLinkTarget The new title link target
510 * @return bool
511 */
512 private function movePage( $id, LinkTarget $newLinkTarget ) {
513 $this->db->update( 'page',
514 [
515 "page_namespace" => $newLinkTarget->getNamespace(),
516 "page_title" => $newLinkTarget->getDBkey(),
517 ],
518 [
519 "page_id" => $id,
520 ],
521 __METHOD__ );
522
523 // Update *_from_namespace in links tables
524 $fromNamespaceTables = [
525 [ 'pagelinks', 'pl' ],
526 [ 'templatelinks', 'tl' ],
527 [ 'imagelinks', 'il' ] ];
528 foreach ( $fromNamespaceTables as $tableInfo ) {
529 list( $table, $fieldPrefix ) = $tableInfo;
530 $this->db->update( $table,
531 // SET
532 [ "{$fieldPrefix}_from_namespace" => $newLinkTarget->getNamespace() ],
533 // WHERE
534 [ "{$fieldPrefix}_from" => $id ],
535 __METHOD__ );
536 }
537
538 return true;
539 }
540
541 /**
542 * Determine if we can merge a page.
543 * We check if an inaccessible revision would become the latest and
544 * deny the merge if so -- it's theoretically possible to update the
545 * latest revision, but opens a can of worms -- search engine updates,
546 * recentchanges review, etc.
547 *
548 * @param integer $id The page_id
549 * @param LinkTarget $linkTarget The new link target
550 * @param string $logStatus This is set to the log status message on failure
551 * @return bool
552 */
553 private function canMerge( $id, LinkTarget $linkTarget, &$logStatus ) {
554 $latestDest = Revision::newFromTitle( $linkTarget, 0, Revision::READ_LATEST );
555 $latestSource = Revision::newFromPageId( $id, 0, Revision::READ_LATEST );
556 if ( $latestSource->getTimestamp() > $latestDest->getTimestamp() ) {
557 $logStatus = 'cannot merge since source is later';
558 return false;
559 } else {
560 return true;
561 }
562 }
563
564 /**
565 * Merge page histories
566 *
567 * @param stdClass $row Page row
568 * @param Title $newTitle The new title
569 * @return bool
570 */
571 private function mergePage( $row, Title $newTitle ) {
572 $id = $row->page_id;
573
574 // Construct the WikiPage object we will need later, while the
575 // page_id still exists. Note that this cannot use makeTitleSafe(),
576 // we are deliberately constructing an invalid title.
577 $sourceTitle = Title::makeTitle( $row->page_namespace, $row->page_title );
578 $sourceTitle->resetArticleID( $id );
579 $wikiPage = new WikiPage( $sourceTitle );
580 $wikiPage->loadPageData( 'fromdbmaster' );
581
582 $destId = $newTitle->getArticleID();
583 $this->beginTransaction( $this->db, __METHOD__ );
584 $this->db->update( 'revision',
585 // SET
586 [ 'rev_page' => $destId ],
587 // WHERE
588 [ 'rev_page' => $id ],
589 __METHOD__ );
590
591 $this->db->delete( 'page', [ 'page_id' => $id ], __METHOD__ );
592
593 $this->commitTransaction( $this->db, __METHOD__ );
594
595 /* Call LinksDeletionUpdate to delete outgoing links from the old title,
596 * and update category counts.
597 *
598 * Calling external code with a fake broken Title is a fairly dubious
599 * idea. It's necessary because it's quite a lot of code to duplicate,
600 * but that also makes it fragile since it would be easy for someone to
601 * accidentally introduce an assumption of title validity to the code we
602 * are calling.
603 */
604 DeferredUpdates::addUpdate( new LinksDeletionUpdate( $wikiPage ) );
605 DeferredUpdates::doUpdates();
606
607 return true;
608 }
609 }
610
611 $maintClass = NamespaceDupes::class;
612 require_once RUN_MAINTENANCE_IF_MAIN;