Merge "Special:Upload should not crash on failing previews"
[mediawiki.git] / includes / WatchedItemQueryService.php
blobc80e4a53198ddcc8ba8df8aadd93561566ef75f6
1 <?php
3 use MediaWiki\Linker\LinkTarget;
4 use Wikimedia\Assert\Assert;
6 /**
7 * Class performing complex database queries related to WatchedItems.
9 * @since 1.28
11 * @file
12 * @ingroup Watchlist
14 * @license GNU GPL v2+
16 class WatchedItemQueryService {
18 const DIR_OLDER = 'older';
19 const DIR_NEWER = 'newer';
21 const INCLUDE_FLAGS = 'flags';
22 const INCLUDE_USER = 'user';
23 const INCLUDE_USER_ID = 'userid';
24 const INCLUDE_COMMENT = 'comment';
25 const INCLUDE_PATROL_INFO = 'patrol';
26 const INCLUDE_SIZES = 'sizes';
27 const INCLUDE_LOG_INFO = 'loginfo';
29 // FILTER_* constants are part of public API (are used in ApiQueryWatchlist and
30 // ApiQueryWatchlistRaw classes) and should not be changed.
31 // Changing values of those constants will result in a breaking change in the API
32 const FILTER_MINOR = 'minor';
33 const FILTER_NOT_MINOR = '!minor';
34 const FILTER_BOT = 'bot';
35 const FILTER_NOT_BOT = '!bot';
36 const FILTER_ANON = 'anon';
37 const FILTER_NOT_ANON = '!anon';
38 const FILTER_PATROLLED = 'patrolled';
39 const FILTER_NOT_PATROLLED = '!patrolled';
40 const FILTER_UNREAD = 'unread';
41 const FILTER_NOT_UNREAD = '!unread';
42 const FILTER_CHANGED = 'changed';
43 const FILTER_NOT_CHANGED = '!changed';
45 const SORT_ASC = 'ASC';
46 const SORT_DESC = 'DESC';
48 /**
49 * @var LoadBalancer
51 private $loadBalancer;
53 /** @var WatchedItemQueryServiceExtension[]|null */
54 private $extensions = null;
56 public function __construct( LoadBalancer $loadBalancer ) {
57 $this->loadBalancer = $loadBalancer;
60 /**
61 * @return WatchedItemQueryServiceExtension[]
63 private function getExtensions() {
64 if ( $this->extensions === null ) {
65 $this->extensions = [];
66 Hooks::run( 'WatchedItemQueryServiceExtensions', [ &$this->extensions, $this ] );
68 return $this->extensions;
71 /**
72 * @return IDatabase
73 * @throws MWException
75 private function getConnection() {
76 return $this->loadBalancer->getConnectionRef( DB_REPLICA, [ 'watchlist' ] );
79 /**
80 * @param User $user
81 * @param array $options Allowed keys:
82 * 'includeFields' => string[] RecentChange fields to be included in the result,
83 * self::INCLUDE_* constants should be used
84 * 'filters' => string[] optional filters to narrow down resulted items
85 * 'namespaceIds' => int[] optional namespace IDs to filter by
86 * (defaults to all namespaces)
87 * 'allRevisions' => bool return multiple revisions of the same page if true,
88 * only the most recent if false (default)
89 * 'rcTypes' => int[] which types of RecentChanges to include
90 * (defaults to all types), allowed values: RC_EDIT, RC_NEW,
91 * RC_LOG, RC_EXTERNAL, RC_CATEGORIZE
92 * 'onlyByUser' => string only list changes by a specified user
93 * 'notByUser' => string do not incluide changes by a specified user
94 * 'dir' => string in which direction to enumerate, accepted values:
95 * - DIR_OLDER list newest first
96 * - DIR_NEWER list oldest first
97 * 'start' => string (format accepted by wfTimestamp) requires 'dir' option,
98 * timestamp to start enumerating from
99 * 'end' => string (format accepted by wfTimestamp) requires 'dir' option,
100 * timestamp to end enumerating
101 * 'watchlistOwner' => User user whose watchlist items should be listed if different
102 * than the one specified with $user param,
103 * requires 'watchlistOwnerToken' option
104 * 'watchlistOwnerToken' => string a watchlist token used to access another user's
105 * watchlist, used with 'watchlistOwnerToken' option
106 * 'limit' => int maximum numbers of items to return
107 * 'usedInGenerator' => bool include only RecentChange id field required by the
108 * generator ('rc_cur_id' or 'rc_this_oldid') if true, or all
109 * id fields ('rc_cur_id', 'rc_this_oldid', 'rc_last_oldid')
110 * if false (default)
111 * @param array|null &$startFrom Continuation value: [ string $rcTimestamp, int $rcId ]
112 * @return array of pairs ( WatchedItem $watchedItem, string[] $recentChangeInfo ),
113 * where $recentChangeInfo contains the following keys:
114 * - 'rc_id',
115 * - 'rc_namespace',
116 * - 'rc_title',
117 * - 'rc_timestamp',
118 * - 'rc_type',
119 * - 'rc_deleted',
120 * Additional keys could be added by specifying the 'includeFields' option
122 public function getWatchedItemsWithRecentChangeInfo(
123 User $user, array $options = [], &$startFrom = null
125 $options += [
126 'includeFields' => [],
127 'namespaceIds' => [],
128 'filters' => [],
129 'allRevisions' => false,
130 'usedInGenerator' => false
133 Assert::parameter(
134 !isset( $options['rcTypes'] )
135 || !array_diff( $options['rcTypes'], [ RC_EDIT, RC_NEW, RC_LOG, RC_EXTERNAL, RC_CATEGORIZE ] ),
136 '$options[\'rcTypes\']',
137 'must be an array containing only: RC_EDIT, RC_NEW, RC_LOG, RC_EXTERNAL and/or RC_CATEGORIZE'
139 Assert::parameter(
140 !isset( $options['dir'] ) || in_array( $options['dir'], [ self::DIR_OLDER, self::DIR_NEWER ] ),
141 '$options[\'dir\']',
142 'must be DIR_OLDER or DIR_NEWER'
144 Assert::parameter(
145 !isset( $options['start'] ) && !isset( $options['end'] ) && $startFrom === null
146 || isset( $options['dir'] ),
147 '$options[\'dir\']',
148 'must be provided when providing the "start" or "end" options or the $startFrom parameter'
150 Assert::parameter(
151 !isset( $options['startFrom'] ),
152 '$options[\'startFrom\']',
153 'must not be provided, use $startFrom instead'
155 Assert::parameter(
156 !isset( $startFrom ) || ( is_array( $startFrom ) && count( $startFrom ) === 2 ),
157 '$startFrom',
158 'must be a two-element array'
160 if ( array_key_exists( 'watchlistOwner', $options ) ) {
161 Assert::parameterType(
162 User::class,
163 $options['watchlistOwner'],
164 '$options[\'watchlistOwner\']'
166 Assert::parameter(
167 isset( $options['watchlistOwnerToken'] ),
168 '$options[\'watchlistOwnerToken\']',
169 'must be provided when providing watchlistOwner option'
173 $tables = [ 'recentchanges', 'watchlist' ];
174 if ( !$options['allRevisions'] ) {
175 $tables[] = 'page';
178 $db = $this->getConnection();
180 $fields = $this->getWatchedItemsWithRCInfoQueryFields( $options );
181 $conds = $this->getWatchedItemsWithRCInfoQueryConds( $db, $user, $options );
182 $dbOptions = $this->getWatchedItemsWithRCInfoQueryDbOptions( $options );
183 $joinConds = $this->getWatchedItemsWithRCInfoQueryJoinConds( $options );
185 if ( $startFrom !== null ) {
186 $conds[] = $this->getStartFromConds( $db, $options, $startFrom );
189 foreach ( $this->getExtensions() as $extension ) {
190 $extension->modifyWatchedItemsWithRCInfoQuery(
191 $user, $options, $db,
192 $tables,
193 $fields,
194 $conds,
195 $dbOptions,
196 $joinConds
200 $res = $db->select(
201 $tables,
202 $fields,
203 $conds,
204 __METHOD__,
205 $dbOptions,
206 $joinConds
209 $limit = isset( $dbOptions['LIMIT'] ) ? $dbOptions['LIMIT'] : INF;
210 $items = [];
211 $startFrom = null;
212 foreach ( $res as $row ) {
213 if ( --$limit <= 0 ) {
214 $startFrom = [ $row->rc_timestamp, $row->rc_id ];
215 break;
218 $items[] = [
219 new WatchedItem(
220 $user,
221 new TitleValue( (int)$row->rc_namespace, $row->rc_title ),
222 $row->wl_notificationtimestamp
224 $this->getRecentChangeFieldsFromRow( $row )
228 foreach ( $this->getExtensions() as $extension ) {
229 $extension->modifyWatchedItemsWithRCInfo( $user, $options, $db, $items, $res, $startFrom );
232 return $items;
236 * For simple listing of user's watchlist items, see WatchedItemStore::getWatchedItemsForUser
238 * @param User $user
239 * @param array $options Allowed keys:
240 * 'sort' => string optional sorting by namespace ID and title
241 * one of the self::SORT_* constants
242 * 'namespaceIds' => int[] optional namespace IDs to filter by (defaults to all namespaces)
243 * 'limit' => int maximum number of items to return
244 * 'filter' => string optional filter, one of the self::FILTER_* contants
245 * 'from' => LinkTarget requires 'sort' key, only return items starting from
246 * those related to the link target
247 * 'until' => LinkTarget requires 'sort' key, only return items until
248 * those related to the link target
249 * 'startFrom' => LinkTarget requires 'sort' key, only return items starting from
250 * those related to the link target, allows to skip some link targets
251 * specified using the form option
252 * @return WatchedItem[]
254 public function getWatchedItemsForUser( User $user, array $options = [] ) {
255 if ( $user->isAnon() ) {
256 // TODO: should this just return an empty array or rather complain loud at this point
257 // as e.g. ApiBase::getWatchlistUser does?
258 return [];
261 $options += [ 'namespaceIds' => [] ];
263 Assert::parameter(
264 !isset( $options['sort'] ) || in_array( $options['sort'], [ self::SORT_ASC, self::SORT_DESC ] ),
265 '$options[\'sort\']',
266 'must be SORT_ASC or SORT_DESC'
268 Assert::parameter(
269 !isset( $options['filter'] ) || in_array(
270 $options['filter'], [ self::FILTER_CHANGED, self::FILTER_NOT_CHANGED ]
272 '$options[\'filter\']',
273 'must be FILTER_CHANGED or FILTER_NOT_CHANGED'
275 Assert::parameter(
276 !isset( $options['from'] ) && !isset( $options['until'] ) && !isset( $options['startFrom'] )
277 || isset( $options['sort'] ),
278 '$options[\'sort\']',
279 'must be provided if any of "from", "until", "startFrom" options is provided'
282 $db = $this->getConnection();
284 $conds = $this->getWatchedItemsForUserQueryConds( $db, $user, $options );
285 $dbOptions = $this->getWatchedItemsForUserQueryDbOptions( $options );
287 $res = $db->select(
288 'watchlist',
289 [ 'wl_namespace', 'wl_title', 'wl_notificationtimestamp' ],
290 $conds,
291 __METHOD__,
292 $dbOptions
295 $watchedItems = [];
296 foreach ( $res as $row ) {
297 // todo these could all be cached at some point?
298 $watchedItems[] = new WatchedItem(
299 $user,
300 new TitleValue( (int)$row->wl_namespace, $row->wl_title ),
301 $row->wl_notificationtimestamp
305 return $watchedItems;
308 private function getRecentChangeFieldsFromRow( stdClass $row ) {
309 // This can be simplified to single array_filter call filtering by key value,
310 // once we stop supporting PHP 5.5
311 $allFields = get_object_vars( $row );
312 $rcKeys = array_filter(
313 array_keys( $allFields ),
314 function( $key ) {
315 return substr( $key, 0, 3 ) === 'rc_';
318 return array_intersect_key( $allFields, array_flip( $rcKeys ) );
321 private function getWatchedItemsWithRCInfoQueryFields( array $options ) {
322 $fields = [
323 'rc_id',
324 'rc_namespace',
325 'rc_title',
326 'rc_timestamp',
327 'rc_type',
328 'rc_deleted',
329 'wl_notificationtimestamp'
332 $rcIdFields = [
333 'rc_cur_id',
334 'rc_this_oldid',
335 'rc_last_oldid',
337 if ( $options['usedInGenerator'] ) {
338 if ( $options['allRevisions'] ) {
339 $rcIdFields = [ 'rc_this_oldid' ];
340 } else {
341 $rcIdFields = [ 'rc_cur_id' ];
344 $fields = array_merge( $fields, $rcIdFields );
346 if ( in_array( self::INCLUDE_FLAGS, $options['includeFields'] ) ) {
347 $fields = array_merge( $fields, [ 'rc_type', 'rc_minor', 'rc_bot' ] );
349 if ( in_array( self::INCLUDE_USER, $options['includeFields'] ) ) {
350 $fields[] = 'rc_user_text';
352 if ( in_array( self::INCLUDE_USER_ID, $options['includeFields'] ) ) {
353 $fields[] = 'rc_user';
355 if ( in_array( self::INCLUDE_COMMENT, $options['includeFields'] ) ) {
356 $fields[] = 'rc_comment';
358 if ( in_array( self::INCLUDE_PATROL_INFO, $options['includeFields'] ) ) {
359 $fields = array_merge( $fields, [ 'rc_patrolled', 'rc_log_type' ] );
361 if ( in_array( self::INCLUDE_SIZES, $options['includeFields'] ) ) {
362 $fields = array_merge( $fields, [ 'rc_old_len', 'rc_new_len' ] );
364 if ( in_array( self::INCLUDE_LOG_INFO, $options['includeFields'] ) ) {
365 $fields = array_merge( $fields, [ 'rc_logid', 'rc_log_type', 'rc_log_action', 'rc_params' ] );
368 return $fields;
371 private function getWatchedItemsWithRCInfoQueryConds(
372 IDatabase $db,
373 User $user,
374 array $options
376 $watchlistOwnerId = $this->getWatchlistOwnerId( $user, $options );
377 $conds = [ 'wl_user' => $watchlistOwnerId ];
379 if ( !$options['allRevisions'] ) {
380 $conds[] = $db->makeList(
381 [ 'rc_this_oldid=page_latest', 'rc_type=' . RC_LOG ],
382 LIST_OR
386 if ( $options['namespaceIds'] ) {
387 $conds['wl_namespace'] = array_map( 'intval', $options['namespaceIds'] );
390 if ( array_key_exists( 'rcTypes', $options ) ) {
391 $conds['rc_type'] = array_map( 'intval', $options['rcTypes'] );
394 $conds = array_merge(
395 $conds,
396 $this->getWatchedItemsWithRCInfoQueryFilterConds( $user, $options )
399 $conds = array_merge( $conds, $this->getStartEndConds( $db, $options ) );
401 if ( !isset( $options['start'] ) && !isset( $options['end'] ) ) {
402 if ( $db->getType() === 'mysql' ) {
403 // This is an index optimization for mysql
404 $conds[] = "rc_timestamp > ''";
408 $conds = array_merge( $conds, $this->getUserRelatedConds( $db, $user, $options ) );
410 $deletedPageLogCond = $this->getExtraDeletedPageLogEntryRelatedCond( $db, $user );
411 if ( $deletedPageLogCond ) {
412 $conds[] = $deletedPageLogCond;
415 return $conds;
418 private function getWatchlistOwnerId( User $user, array $options ) {
419 if ( array_key_exists( 'watchlistOwner', $options ) ) {
420 /** @var User $watchlistOwner */
421 $watchlistOwner = $options['watchlistOwner'];
422 $ownersToken = $watchlistOwner->getOption( 'watchlisttoken' );
423 $token = $options['watchlistOwnerToken'];
424 if ( $ownersToken == '' || !hash_equals( $ownersToken, $token ) ) {
425 throw ApiUsageException::newWithMessage( null, 'apierror-bad-watchlist-token', 'bad_wltoken' );
427 return $watchlistOwner->getId();
429 return $user->getId();
432 private function getWatchedItemsWithRCInfoQueryFilterConds( User $user, array $options ) {
433 $conds = [];
435 if ( in_array( self::FILTER_MINOR, $options['filters'] ) ) {
436 $conds[] = 'rc_minor != 0';
437 } elseif ( in_array( self::FILTER_NOT_MINOR, $options['filters'] ) ) {
438 $conds[] = 'rc_minor = 0';
441 if ( in_array( self::FILTER_BOT, $options['filters'] ) ) {
442 $conds[] = 'rc_bot != 0';
443 } elseif ( in_array( self::FILTER_NOT_BOT, $options['filters'] ) ) {
444 $conds[] = 'rc_bot = 0';
447 if ( in_array( self::FILTER_ANON, $options['filters'] ) ) {
448 $conds[] = 'rc_user = 0';
449 } elseif ( in_array( self::FILTER_NOT_ANON, $options['filters'] ) ) {
450 $conds[] = 'rc_user != 0';
453 if ( $user->useRCPatrol() || $user->useNPPatrol() ) {
454 // TODO: not sure if this should simply ignore patrolled filters if user does not have the patrol
455 // right, or maybe rather fail loud at this point, same as e.g. ApiQueryWatchlist does?
456 if ( in_array( self::FILTER_PATROLLED, $options['filters'] ) ) {
457 $conds[] = 'rc_patrolled != 0';
458 } elseif ( in_array( self::FILTER_NOT_PATROLLED, $options['filters'] ) ) {
459 $conds[] = 'rc_patrolled = 0';
463 if ( in_array( self::FILTER_UNREAD, $options['filters'] ) ) {
464 $conds[] = 'rc_timestamp >= wl_notificationtimestamp';
465 } elseif ( in_array( self::FILTER_NOT_UNREAD, $options['filters'] ) ) {
466 // TODO: should this be changed to use Database::makeList?
467 $conds[] = 'wl_notificationtimestamp IS NULL OR rc_timestamp < wl_notificationtimestamp';
470 return $conds;
473 private function getStartEndConds( IDatabase $db, array $options ) {
474 if ( !isset( $options['start'] ) && !isset( $options['end'] ) ) {
475 return [];
478 $conds = [];
480 if ( isset( $options['start'] ) ) {
481 $after = $options['dir'] === self::DIR_OLDER ? '<=' : '>=';
482 $conds[] = 'rc_timestamp ' . $after . ' ' .
483 $db->addQuotes( $db->timestamp( $options['start'] ) );
485 if ( isset( $options['end'] ) ) {
486 $before = $options['dir'] === self::DIR_OLDER ? '>=' : '<=';
487 $conds[] = 'rc_timestamp ' . $before . ' ' .
488 $db->addQuotes( $db->timestamp( $options['end'] ) );
491 return $conds;
494 private function getUserRelatedConds( IDatabase $db, User $user, array $options ) {
495 if ( !array_key_exists( 'onlyByUser', $options ) && !array_key_exists( 'notByUser', $options ) ) {
496 return [];
499 $conds = [];
501 if ( array_key_exists( 'onlyByUser', $options ) ) {
502 $conds['rc_user_text'] = $options['onlyByUser'];
503 } elseif ( array_key_exists( 'notByUser', $options ) ) {
504 $conds[] = 'rc_user_text != ' . $db->addQuotes( $options['notByUser'] );
507 // Avoid brute force searches (bug 17342)
508 $bitmask = 0;
509 if ( !$user->isAllowed( 'deletedhistory' ) ) {
510 $bitmask = Revision::DELETED_USER;
511 } elseif ( !$user->isAllowedAny( 'suppressrevision', 'viewsuppressed' ) ) {
512 $bitmask = Revision::DELETED_USER | Revision::DELETED_RESTRICTED;
514 if ( $bitmask ) {
515 $conds[] = $db->bitAnd( 'rc_deleted', $bitmask ) . " != $bitmask";
518 return $conds;
521 private function getExtraDeletedPageLogEntryRelatedCond( IDatabase $db, User $user ) {
522 // LogPage::DELETED_ACTION hides the affected page, too. So hide those
523 // entirely from the watchlist, or someone could guess the title.
524 $bitmask = 0;
525 if ( !$user->isAllowed( 'deletedhistory' ) ) {
526 $bitmask = LogPage::DELETED_ACTION;
527 } elseif ( !$user->isAllowedAny( 'suppressrevision', 'viewsuppressed' ) ) {
528 $bitmask = LogPage::DELETED_ACTION | LogPage::DELETED_RESTRICTED;
530 if ( $bitmask ) {
531 return $db->makeList( [
532 'rc_type != ' . RC_LOG,
533 $db->bitAnd( 'rc_deleted', $bitmask ) . " != $bitmask",
534 ], LIST_OR );
536 return '';
539 private function getStartFromConds( IDatabase $db, array $options, array $startFrom ) {
540 $op = $options['dir'] === self::DIR_OLDER ? '<' : '>';
541 list( $rcTimestamp, $rcId ) = $startFrom;
542 $rcTimestamp = $db->addQuotes( $db->timestamp( $rcTimestamp ) );
543 $rcId = (int)$rcId;
544 return $db->makeList(
546 "rc_timestamp $op $rcTimestamp",
547 $db->makeList(
549 "rc_timestamp = $rcTimestamp",
550 "rc_id $op= $rcId"
552 LIST_AND
555 LIST_OR
559 private function getWatchedItemsForUserQueryConds( IDatabase $db, User $user, array $options ) {
560 $conds = [ 'wl_user' => $user->getId() ];
561 if ( $options['namespaceIds'] ) {
562 $conds['wl_namespace'] = array_map( 'intval', $options['namespaceIds'] );
564 if ( isset( $options['filter'] ) ) {
565 $filter = $options['filter'];
566 if ( $filter === self::FILTER_CHANGED ) {
567 $conds[] = 'wl_notificationtimestamp IS NOT NULL';
568 } else {
569 $conds[] = 'wl_notificationtimestamp IS NULL';
573 if ( isset( $options['from'] ) ) {
574 $op = $options['sort'] === self::SORT_ASC ? '>' : '<';
575 $conds[] = $this->getFromUntilTargetConds( $db, $options['from'], $op );
577 if ( isset( $options['until'] ) ) {
578 $op = $options['sort'] === self::SORT_ASC ? '<' : '>';
579 $conds[] = $this->getFromUntilTargetConds( $db, $options['until'], $op );
581 if ( isset( $options['startFrom'] ) ) {
582 $op = $options['sort'] === self::SORT_ASC ? '>' : '<';
583 $conds[] = $this->getFromUntilTargetConds( $db, $options['startFrom'], $op );
586 return $conds;
590 * Creates a query condition part for getting only items before or after the given link target
591 * (while ordering using $sort mode)
593 * @param IDatabase $db
594 * @param LinkTarget $target
595 * @param string $op comparison operator to use in the conditions
596 * @return string
598 private function getFromUntilTargetConds( IDatabase $db, LinkTarget $target, $op ) {
599 return $db->makeList(
601 "wl_namespace $op " . $target->getNamespace(),
602 $db->makeList(
604 'wl_namespace = ' . $target->getNamespace(),
605 "wl_title $op= " . $db->addQuotes( $target->getDBkey() )
607 LIST_AND
610 LIST_OR
614 private function getWatchedItemsWithRCInfoQueryDbOptions( array $options ) {
615 $dbOptions = [];
617 if ( array_key_exists( 'dir', $options ) ) {
618 $sort = $options['dir'] === self::DIR_OLDER ? ' DESC' : '';
619 $dbOptions['ORDER BY'] = [ 'rc_timestamp' . $sort, 'rc_id' . $sort ];
622 if ( array_key_exists( 'limit', $options ) ) {
623 $dbOptions['LIMIT'] = (int)$options['limit'] + 1;
626 return $dbOptions;
629 private function getWatchedItemsForUserQueryDbOptions( array $options ) {
630 $dbOptions = [];
631 if ( array_key_exists( 'sort', $options ) ) {
632 $dbOptions['ORDER BY'] = [
633 "wl_namespace {$options['sort']}",
634 "wl_title {$options['sort']}"
636 if ( count( $options['namespaceIds'] ) === 1 ) {
637 $dbOptions['ORDER BY'] = "wl_title {$options['sort']}";
640 if ( array_key_exists( 'limit', $options ) ) {
641 $dbOptions['LIMIT'] = (int)$options['limit'];
643 return $dbOptions;
646 private function getWatchedItemsWithRCInfoQueryJoinConds( array $options ) {
647 $joinConds = [
648 'watchlist' => [ 'INNER JOIN',
650 'wl_namespace=rc_namespace',
651 'wl_title=rc_title'
655 if ( !$options['allRevisions'] ) {
656 $joinConds['page'] = [ 'LEFT JOIN', 'rc_cur_id=page_id' ];
658 return $joinConds;