92856e9f42ce98384a6b93208cf143c70048b196
3 * See docs/deferred.txt
5 * @todo document (e.g. one-sentence top-level class description).
12 var $mId, //!< Page ID of the article linked from
13 $mTitle, //!< Title object of the article linked from
14 $mLinks, //!< Map of title strings to IDs for the links in the document
15 $mImages, //!< DB keys of the images used, in the array key only
16 $mTemplates, //!< Map of title strings to IDs for the template references, including broken ones
17 $mExternals, //!< URLs of external links, array key only
18 $mCategories, //!< Map of category names to sort keys
19 $mInterlangs, //!< Map of language codes to titles
20 $mProperties, //!< Map of arbitrary name to value
21 $mDb, //!< Database connection reference
22 $mOptions, //!< SELECT options to be used (array)
23 $mRecursive; //!< Whether to queue jobs for recursive updates
29 * @param Title $title Title of the page we're updating
30 * @param ParserOutput $parserOutput Output from a full parse of this page
31 * @param bool $recursive Queue jobs for recursive updates?
33 function LinksUpdate( $title, $parserOutput, $recursive = true ) {
34 global $wgAntiLockFlags;
36 if ( $wgAntiLockFlags & ALF_NO_LINK_LOCK
) {
37 $this->mOptions
= array();
39 $this->mOptions
= array( 'FOR UPDATE' );
41 $this->mDb
= wfGetDB( DB_MASTER
);
43 if ( !is_object( $title ) ) {
44 throw new MWException( "The calling convention to LinksUpdate::LinksUpdate() has changed. " .
45 "Please see Article::editUpdates() for an invocation example.\n" );
47 $this->mTitle
= $title;
48 $this->mId
= $title->getArticleID();
50 $this->mParserOutput
= $parserOutput;
51 $this->mLinks
= $parserOutput->getLinks();
52 $this->mImages
= $parserOutput->getImages();
53 $this->mTemplates
= $parserOutput->getTemplates();
54 $this->mExternals
= $parserOutput->getExternalLinks();
55 $this->mCategories
= $parserOutput->getCategories();
56 $this->mProperties
= $parserOutput->getProperties();
58 # Convert the format of the interlanguage links
59 # I didn't want to change it in the ParserOutput, because that array is passed all
60 # the way back to the skin, so either a skin API break would be required, or an
61 # inefficient back-conversion.
62 $ill = $parserOutput->getLanguageLinks();
63 $this->mInterlangs
= array();
64 foreach ( $ill as $link ) {
65 list( $key, $title ) = explode( ':', $link, 2 );
66 $this->mInterlangs
[$key] = $title;
69 $this->mRecursive
= $recursive;
70 $this->mTouchTmplLinks
= false;
72 wfRunHooks( 'LinksUpdateConstructed', array( &$this ) );
76 * Update link tables with outgoing links from an updated article
78 public function doUpdate() {
79 global $wgUseDumbLinkUpdate;
81 wfRunHooks( 'LinksUpdate', array( &$this ) );
82 if ( $wgUseDumbLinkUpdate ) {
83 $this->doDumbUpdate();
85 $this->doIncrementalUpdate();
87 wfRunHooks( 'LinksUpdateComplete', array( &$this ) );
90 protected function doIncrementalUpdate() {
91 wfProfileIn( __METHOD__
);
94 $existing = $this->getExistingLinks();
95 $this->incrTableUpdate( 'pagelinks', 'pl', $this->getLinkDeletions( $existing ),
96 $this->getLinkInsertions( $existing ) );
99 $existing = $this->getExistingImages();
101 $imageDeletes = $this->getImageDeletions( $existing );
102 $this->incrTableUpdate( 'imagelinks', 'il', $imageDeletes, $this->getImageInsertions( $existing ) );
104 # Invalidate all image description pages which had links added or removed
105 $imageUpdates = $imageDeletes +
array_diff_key( $this->mImages
, $existing );
106 $this->invalidateImageDescriptions( $imageUpdates );
109 $existing = $this->getExistingExternals();
110 $this->incrTableUpdate( 'externallinks', 'el', $this->getExternalDeletions( $existing ),
111 $this->getExternalInsertions( $existing ) );
114 $existing = $this->getExistingInterlangs();
115 $this->incrTableUpdate( 'langlinks', 'll', $this->getInterlangDeletions( $existing ),
116 $this->getInterlangInsertions( $existing ) );
119 $existing = $this->getExistingTemplates();
120 $this->incrTableUpdate( 'templatelinks', 'tl', $this->getTemplateDeletions( $existing ),
121 $this->getTemplateInsertions( $existing ) );
124 $existing = $this->getExistingCategories();
126 $categoryDeletes = $this->getCategoryDeletions( $existing );
128 $this->incrTableUpdate( 'categorylinks', 'cl', $categoryDeletes, $this->getCategoryInsertions( $existing ) );
130 # Invalidate all categories which were added, deleted or changed (set symmetric difference)
131 $categoryInserts = array_diff_assoc( $this->mCategories
, $existing );
132 $categoryUpdates = $categoryInserts +
$categoryDeletes;
133 $this->invalidateCategories( $categoryUpdates );
134 $this->updateCategoryCounts( $categoryInserts, $categoryDeletes );
137 $existing = $this->getExistingProperties();
139 $propertiesDeletes = $this->getPropertyDeletions( $existing );
141 $this->incrTableUpdate( 'page_props', 'pp', $propertiesDeletes, $this->getPropertyInsertions( $existing ) );
143 # Invalidate the necessary pages
144 $changed = $propertiesDeletes +
array_diff_assoc( $this->mProperties
, $existing );
145 $this->invalidateProperties( $changed );
147 # Refresh links of all pages including this page
148 # This will be in a separate transaction
149 if ( $this->mRecursive
) {
150 $this->queueRecursiveJobs();
153 wfProfileOut( __METHOD__
);
157 * Link update which clears the previous entries and inserts new ones
158 * May be slower or faster depending on level of lock contention and write speed of DB
159 * Also useful where link table corruption needs to be repaired, e.g. in refreshLinks.php
161 protected function doDumbUpdate() {
162 wfProfileIn( __METHOD__
);
164 # Refresh category pages and image description pages
165 $existing = $this->getExistingCategories();
166 $categoryInserts = array_diff_assoc( $this->mCategories
, $existing );
167 $categoryDeletes = array_diff_assoc( $existing, $this->mCategories
);
168 $categoryUpdates = $categoryInserts +
$categoryDeletes;
169 $existing = $this->getExistingImages();
170 $imageUpdates = array_diff_key( $existing, $this->mImages
) +
array_diff_key( $this->mImages
, $existing );
172 $this->dumbTableUpdate( 'pagelinks', $this->getLinkInsertions(), 'pl_from' );
173 $this->dumbTableUpdate( 'imagelinks', $this->getImageInsertions(), 'il_from' );
174 $this->dumbTableUpdate( 'categorylinks', $this->getCategoryInsertions(), 'cl_from' );
175 $this->dumbTableUpdate( 'templatelinks', $this->getTemplateInsertions(), 'tl_from' );
176 $this->dumbTableUpdate( 'externallinks', $this->getExternalInsertions(), 'el_from' );
177 $this->dumbTableUpdate( 'langlinks', $this->getInterlangInsertions(),'ll_from' );
178 $this->dumbTableUpdate( 'page_props', $this->getPropertyInsertions(), 'pp_page' );
180 # Update the cache of all the category pages and image description
181 # pages which were changed, and fix the category table count
182 $this->invalidateCategories( $categoryUpdates );
183 $this->updateCategoryCounts( $categoryInserts, $categoryDeletes );
184 $this->invalidateImageDescriptions( $imageUpdates );
186 # Refresh links of all pages including this page
187 # This will be in a separate transaction
188 if ( $this->mRecursive
) {
189 $this->queueRecursiveJobs();
192 wfProfileOut( __METHOD__
);
195 function queueRecursiveJobs() {
196 global $wgUpdateRowsPerJob;
197 wfProfileIn( __METHOD__
);
199 $cache = $this->mTitle
->getBacklinkCache();
200 $batches = $cache->partition( 'templatelinks', $wgUpdateRowsPerJob );
205 foreach ( $batches as $batch ) {
206 list( $start, $end ) = $batch;
211 $jobs[] = new RefreshLinksJob2( $this->mTitle
, $params );
213 Job
::batchInsert( $jobs );
215 wfProfileOut( __METHOD__
);
219 * Invalidate the cache of a list of pages from a single namespace
221 * @param integer $namespace
222 * @param array $dbkeys
224 function invalidatePages( $namespace, $dbkeys ) {
225 if ( !count( $dbkeys ) ) {
230 * Determine which pages need to be updated
231 * This is necessary to prevent the job queue from smashing the DB with
232 * large numbers of concurrent invalidations of the same page
234 $now = $this->mDb
->timestamp();
236 $res = $this->mDb
->select( 'page', array( 'page_id' ),
238 'page_namespace' => $namespace,
239 'page_title IN (' . $this->mDb
->makeList( $dbkeys ) . ')',
240 'page_touched < ' . $this->mDb
->addQuotes( $now )
243 while ( $row = $this->mDb
->fetchObject( $res ) ) {
244 $ids[] = $row->page_id
;
246 if ( !count( $ids ) ) {
252 * We still need the page_touched condition, in case the row has changed since
253 * the non-locking select above.
255 $this->mDb
->update( 'page', array( 'page_touched' => $now ),
257 'page_id IN (' . $this->mDb
->makeList( $ids ) . ')',
258 'page_touched < ' . $this->mDb
->addQuotes( $now )
263 function invalidateCategories( $cats ) {
264 $this->invalidatePages( NS_CATEGORY
, array_keys( $cats ) );
268 * Update all the appropriate counts in the category table.
269 * @param $added associative array of category name => sort key
270 * @param $deleted associative array of category name => sort key
272 function updateCategoryCounts( $added, $deleted ) {
273 $a = new Article($this->mTitle
);
274 $a->updateCategoryCounts(
275 array_keys( $added ), array_keys( $deleted )
279 function invalidateImageDescriptions( $images ) {
280 $this->invalidatePages( NS_FILE
, array_keys( $images ) );
283 function dumbTableUpdate( $table, $insertions, $fromField ) {
284 $this->mDb
->delete( $table, array( $fromField => $this->mId
), __METHOD__
);
285 if ( count( $insertions ) ) {
286 # The link array was constructed without FOR UPDATE, so there may
287 # be collisions. This may cause minor link table inconsistencies,
288 # which is better than crippling the site with lock contention.
289 $this->mDb
->insert( $table, $insertions, __METHOD__
, array( 'IGNORE' ) );
294 * Make a WHERE clause from a 2-d NS/dbkey array
296 * @param array $arr 2-d array indexed by namespace and DB key
297 * @param string $prefix Field name prefix, without the underscore
299 function makeWhereFrom2d( &$arr, $prefix ) {
301 $lb->setArray( $arr );
302 return $lb->constructSet( $prefix, $this->mDb
);
306 * Update a table by doing a delete query then an insert query
309 function incrTableUpdate( $table, $prefix, $deletions, $insertions ) {
310 if ( $table == 'page_props' ) {
311 $fromField = 'pp_page';
313 $fromField = "{$prefix}_from";
315 $where = array( $fromField => $this->mId
);
316 if ( $table == 'pagelinks' ||
$table == 'templatelinks' ) {
317 $clause = $this->makeWhereFrom2d( $deletions, $prefix );
324 if ( $table == 'langlinks' ) {
325 $toField = 'll_lang';
326 } elseif ( $table == 'page_props' ) {
327 $toField = 'pp_propname';
329 $toField = $prefix . '_to';
331 if ( count( $deletions ) ) {
332 $where[] = "$toField IN (" . $this->mDb
->makeList( array_keys( $deletions ) ) . ')';
338 $this->mDb
->delete( $table, $where, __METHOD__
);
340 if ( count( $insertions ) ) {
341 $this->mDb
->insert( $table, $insertions, __METHOD__
, 'IGNORE' );
347 * Get an array of pagelinks insertions for passing to the DB
348 * Skips the titles specified by the 2-D array $existing
351 function getLinkInsertions( $existing = array() ) {
353 foreach( $this->mLinks
as $ns => $dbkeys ) {
354 # array_diff_key() was introduced in PHP 5.1, there is a compatibility function
355 # in GlobalFunctions.php
356 $diffs = isset( $existing[$ns] ) ?
array_diff_key( $dbkeys, $existing[$ns] ) : $dbkeys;
357 foreach ( $diffs as $dbk => $id ) {
359 'pl_from' => $this->mId
,
360 'pl_namespace' => $ns,
369 * Get an array of template insertions. Like getLinkInsertions()
372 function getTemplateInsertions( $existing = array() ) {
374 foreach( $this->mTemplates
as $ns => $dbkeys ) {
375 $diffs = isset( $existing[$ns] ) ?
array_diff_key( $dbkeys, $existing[$ns] ) : $dbkeys;
376 foreach ( $diffs as $dbk => $id ) {
378 'tl_from' => $this->mId
,
379 'tl_namespace' => $ns,
388 * Get an array of image insertions
389 * Skips the names specified in $existing
392 function getImageInsertions( $existing = array() ) {
394 $diffs = array_diff_key( $this->mImages
, $existing );
395 foreach( $diffs as $iname => $dummy ) {
397 'il_from' => $this->mId
,
405 * Get an array of externallinks insertions. Skips the names specified in $existing
408 function getExternalInsertions( $existing = array() ) {
410 $diffs = array_diff_key( $this->mExternals
, $existing );
411 foreach( $diffs as $url => $dummy ) {
413 'el_from' => $this->mId
,
415 'el_index' => wfMakeUrlIndex( $url ),
422 * Get an array of category insertions
423 * @param array $existing Array mapping existing category names to sort keys. If both
424 * match a link in $this, the link will be omitted from the output
427 function getCategoryInsertions( $existing = array() ) {
429 $diffs = array_diff_assoc( $this->mCategories
, $existing );
431 foreach ( $diffs as $name => $sortkey ) {
432 $nt = Title
::makeTitleSafe( NS_CATEGORY
, $name );
433 $wgContLang->findVariantLink( $name, $nt, true );
434 // for category redirection
435 if ( $nt->isRedirect() ) {
436 $at = new Article( $nt );
437 $nt = $at->getRedirectTarget();
438 // we only redirect a category to another category
439 if ( $nt->getNamespace() == NS_CATEGORY
)
440 $name = $nt->getText();
443 'cl_from' => $this->mId
,
445 'cl_sortkey' => $sortkey,
446 'cl_timestamp' => $this->mDb
->timestamp()
453 * Get an array of interlanguage link insertions
454 * @param array $existing Array mapping existing language codes to titles
457 function getInterlangInsertions( $existing = array() ) {
458 $diffs = array_diff_assoc( $this->mInterlangs
, $existing );
460 foreach( $diffs as $lang => $title ) {
462 'll_from' => $this->mId
,
471 * Get an array of page property insertions
473 function getPropertyInsertions( $existing = array() ) {
474 $diffs = array_diff_assoc( $this->mProperties
, $existing );
476 foreach ( $diffs as $name => $value ) {
478 'pp_page' => $this->mId
,
479 'pp_propname' => $name,
480 'pp_value' => $value,
488 * Given an array of existing links, returns those links which are not in $this
489 * and thus should be deleted.
492 function getLinkDeletions( $existing ) {
494 foreach ( $existing as $ns => $dbkeys ) {
495 if ( isset( $this->mLinks
[$ns] ) ) {
496 $del[$ns] = array_diff_key( $existing[$ns], $this->mLinks
[$ns] );
498 $del[$ns] = $existing[$ns];
505 * Given an array of existing templates, returns those templates which are not in $this
506 * and thus should be deleted.
509 function getTemplateDeletions( $existing ) {
511 foreach ( $existing as $ns => $dbkeys ) {
512 if ( isset( $this->mTemplates
[$ns] ) ) {
513 $del[$ns] = array_diff_key( $existing[$ns], $this->mTemplates
[$ns] );
515 $del[$ns] = $existing[$ns];
522 * Given an array of existing images, returns those images which are not in $this
523 * and thus should be deleted.
526 function getImageDeletions( $existing ) {
527 return array_diff_key( $existing, $this->mImages
);
531 * Given an array of existing external links, returns those links which are not
532 * in $this and thus should be deleted.
535 function getExternalDeletions( $existing ) {
536 return array_diff_key( $existing, $this->mExternals
);
540 * Given an array of existing categories, returns those categories which are not in $this
541 * and thus should be deleted.
544 function getCategoryDeletions( $existing ) {
545 return array_diff_assoc( $existing, $this->mCategories
);
549 * Given an array of existing interlanguage links, returns those links which are not
550 * in $this and thus should be deleted.
553 function getInterlangDeletions( $existing ) {
554 return array_diff_assoc( $existing, $this->mInterlangs
);
558 * Get array of properties which should be deleted.
561 function getPropertyDeletions( $existing ) {
562 return array_diff_assoc( $existing, $this->mProperties
);
566 * Get an array of existing links, as a 2-D array
569 function getExistingLinks() {
570 $res = $this->mDb
->select( 'pagelinks', array( 'pl_namespace', 'pl_title' ),
571 array( 'pl_from' => $this->mId
), __METHOD__
, $this->mOptions
);
573 while ( $row = $this->mDb
->fetchObject( $res ) ) {
574 if ( !isset( $arr[$row->pl_namespace
] ) ) {
575 $arr[$row->pl_namespace
] = array();
577 $arr[$row->pl_namespace
][$row->pl_title
] = 1;
579 $this->mDb
->freeResult( $res );
584 * Get an array of existing templates, as a 2-D array
587 function getExistingTemplates() {
588 $res = $this->mDb
->select( 'templatelinks', array( 'tl_namespace', 'tl_title' ),
589 array( 'tl_from' => $this->mId
), __METHOD__
, $this->mOptions
);
591 while ( $row = $this->mDb
->fetchObject( $res ) ) {
592 if ( !isset( $arr[$row->tl_namespace
] ) ) {
593 $arr[$row->tl_namespace
] = array();
595 $arr[$row->tl_namespace
][$row->tl_title
] = 1;
597 $this->mDb
->freeResult( $res );
602 * Get an array of existing images, image names in the keys
605 function getExistingImages() {
606 $res = $this->mDb
->select( 'imagelinks', array( 'il_to' ),
607 array( 'il_from' => $this->mId
), __METHOD__
, $this->mOptions
);
609 while ( $row = $this->mDb
->fetchObject( $res ) ) {
610 $arr[$row->il_to
] = 1;
612 $this->mDb
->freeResult( $res );
617 * Get an array of existing external links, URLs in the keys
620 function getExistingExternals() {
621 $res = $this->mDb
->select( 'externallinks', array( 'el_to' ),
622 array( 'el_from' => $this->mId
), __METHOD__
, $this->mOptions
);
624 while ( $row = $this->mDb
->fetchObject( $res ) ) {
625 $arr[$row->el_to
] = 1;
627 $this->mDb
->freeResult( $res );
632 * Get an array of existing categories, with the name in the key and sort key in the value.
635 function getExistingCategories() {
636 $res = $this->mDb
->select( 'categorylinks', array( 'cl_to', 'cl_sortkey' ),
637 array( 'cl_from' => $this->mId
), __METHOD__
, $this->mOptions
);
639 while ( $row = $this->mDb
->fetchObject( $res ) ) {
640 $arr[$row->cl_to
] = $row->cl_sortkey
;
642 $this->mDb
->freeResult( $res );
647 * Get an array of existing interlanguage links, with the language code in the key and the
648 * title in the value.
651 function getExistingInterlangs() {
652 $res = $this->mDb
->select( 'langlinks', array( 'll_lang', 'll_title' ),
653 array( 'll_from' => $this->mId
), __METHOD__
, $this->mOptions
);
655 while ( $row = $this->mDb
->fetchObject( $res ) ) {
656 $arr[$row->ll_lang
] = $row->ll_title
;
662 * Get an array of existing categories, with the name in the key and sort key in the value.
665 function getExistingProperties() {
666 $res = $this->mDb
->select( 'page_props', array( 'pp_propname', 'pp_value' ),
667 array( 'pp_page' => $this->mId
), __METHOD__
, $this->mOptions
);
669 while ( $row = $this->mDb
->fetchObject( $res ) ) {
670 $arr[$row->pp_propname
] = $row->pp_value
;
672 $this->mDb
->freeResult( $res );
678 * Return the title object of the page being updated
680 function getTitle() {
681 return $this->mTitle
;
685 * Invalidate any necessary link lists related to page property changes
687 function invalidateProperties( $changed ) {
688 global $wgPagePropLinkInvalidations;
690 foreach ( $changed as $name => $value ) {
691 if ( isset( $wgPagePropLinkInvalidations[$name] ) ) {
692 $inv = $wgPagePropLinkInvalidations[$name];
693 if ( !is_array( $inv ) ) {
694 $inv = array( $inv );
696 foreach ( $inv as $table ) {
697 $update = new HTMLCacheUpdate( $this->mTitle
, $table );