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 of the page we're updating
30 * @param $parserOutput ParserOutput: output from a full parse of this page
31 * @param $recursive Boolean: queue jobs for recursive updates?
33 function __construct( $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();
57 $this->mInterwikis
= $parserOutput->getInterwikiLinks();
59 # Convert the format of the interlanguage links
60 # I didn't want to change it in the ParserOutput, because that array is passed all
61 # the way back to the skin, so either a skin API break would be required, or an
62 # inefficient back-conversion.
63 $ill = $parserOutput->getLanguageLinks();
64 $this->mInterlangs
= array();
65 foreach ( $ill as $link ) {
66 list( $key, $title ) = explode( ':', $link, 2 );
67 $this->mInterlangs
[$key] = $title;
70 foreach ( $this->mCategories
as $cat => &$sortkey ) {
71 # If the sortkey is longer then 255 bytes,
72 # it truncated by DB, and then doesn't get
73 # matched when comparing existing vs current
74 # categories, causing bug 25254.
75 $sortkey = substr( $sortkey, 0, 255 );
78 $this->mRecursive
= $recursive;
80 wfRunHooks( 'LinksUpdateConstructed', array( &$this ) );
84 * Update link tables with outgoing links from an updated article
86 public function doUpdate() {
87 global $wgUseDumbLinkUpdate;
89 wfRunHooks( 'LinksUpdate', array( &$this ) );
90 if ( $wgUseDumbLinkUpdate ) {
91 $this->doDumbUpdate();
93 $this->doIncrementalUpdate();
95 wfRunHooks( 'LinksUpdateComplete', array( &$this ) );
98 protected function doIncrementalUpdate() {
99 wfProfileIn( __METHOD__
);
102 $existing = $this->getExistingLinks();
103 $this->incrTableUpdate( 'pagelinks', 'pl', $this->getLinkDeletions( $existing ),
104 $this->getLinkInsertions( $existing ) );
107 $existing = $this->getExistingImages();
109 $imageDeletes = $this->getImageDeletions( $existing );
110 $this->incrTableUpdate( 'imagelinks', 'il', $imageDeletes, $this->getImageInsertions( $existing ) );
112 # Invalidate all image description pages which had links added or removed
113 $imageUpdates = $imageDeletes +
array_diff_key( $this->mImages
, $existing );
114 $this->invalidateImageDescriptions( $imageUpdates );
117 $existing = $this->getExistingExternals();
118 $this->incrTableUpdate( 'externallinks', 'el', $this->getExternalDeletions( $existing ),
119 $this->getExternalInsertions( $existing ) );
122 $existing = $this->getExistingInterlangs();
123 $this->incrTableUpdate( 'langlinks', 'll', $this->getInterlangDeletions( $existing ),
124 $this->getInterlangInsertions( $existing ) );
126 # Inline interwiki links
127 $existing = $this->getExistingInterwikis();
128 $this->incrTableUpdate( 'iwlinks', 'iwl', $this->getInterwikiDeletions( $existing ),
129 $this->getInterwikiInsertions( $existing ) );
132 $existing = $this->getExistingTemplates();
133 $this->incrTableUpdate( 'templatelinks', 'tl', $this->getTemplateDeletions( $existing ),
134 $this->getTemplateInsertions( $existing ) );
137 $existing = $this->getExistingCategories();
139 $categoryDeletes = $this->getCategoryDeletions( $existing );
141 $this->incrTableUpdate( 'categorylinks', 'cl', $categoryDeletes, $this->getCategoryInsertions( $existing ) );
143 # Invalidate all categories which were added, deleted or changed (set symmetric difference)
144 $categoryInserts = array_diff_assoc( $this->mCategories
, $existing );
145 $categoryUpdates = $categoryInserts +
$categoryDeletes;
146 $this->invalidateCategories( $categoryUpdates );
147 $this->updateCategoryCounts( $categoryInserts, $categoryDeletes );
150 $existing = $this->getExistingProperties();
152 $propertiesDeletes = $this->getPropertyDeletions( $existing );
154 $this->incrTableUpdate( 'page_props', 'pp', $propertiesDeletes, $this->getPropertyInsertions( $existing ) );
156 # Invalidate the necessary pages
157 $changed = $propertiesDeletes +
array_diff_assoc( $this->mProperties
, $existing );
158 $this->invalidateProperties( $changed );
160 # Refresh links of all pages including this page
161 # This will be in a separate transaction
162 if ( $this->mRecursive
) {
163 $this->queueRecursiveJobs();
166 wfProfileOut( __METHOD__
);
170 * Link update which clears the previous entries and inserts new ones
171 * May be slower or faster depending on level of lock contention and write speed of DB
172 * Also useful where link table corruption needs to be repaired, e.g. in refreshLinks.php
174 protected function doDumbUpdate() {
175 wfProfileIn( __METHOD__
);
177 # Refresh category pages and image description pages
178 $existing = $this->getExistingCategories();
179 $categoryInserts = array_diff_assoc( $this->mCategories
, $existing );
180 $categoryDeletes = array_diff_assoc( $existing, $this->mCategories
);
181 $categoryUpdates = $categoryInserts +
$categoryDeletes;
182 $existing = $this->getExistingImages();
183 $imageUpdates = array_diff_key( $existing, $this->mImages
) +
array_diff_key( $this->mImages
, $existing );
185 $this->dumbTableUpdate( 'pagelinks', $this->getLinkInsertions(), 'pl_from' );
186 $this->dumbTableUpdate( 'imagelinks', $this->getImageInsertions(), 'il_from' );
187 $this->dumbTableUpdate( 'categorylinks', $this->getCategoryInsertions(), 'cl_from' );
188 $this->dumbTableUpdate( 'templatelinks', $this->getTemplateInsertions(), 'tl_from' );
189 $this->dumbTableUpdate( 'externallinks', $this->getExternalInsertions(), 'el_from' );
190 $this->dumbTableUpdate( 'langlinks', $this->getInterlangInsertions(),'ll_from' );
191 $this->dumbTableUpdate( 'iwlinks', $this->getInterwikiInsertions(),'iwl_from' );
192 $this->dumbTableUpdate( 'page_props', $this->getPropertyInsertions(), 'pp_page' );
194 # Update the cache of all the category pages and image description
195 # pages which were changed, and fix the category table count
196 $this->invalidateCategories( $categoryUpdates );
197 $this->updateCategoryCounts( $categoryInserts, $categoryDeletes );
198 $this->invalidateImageDescriptions( $imageUpdates );
200 # Refresh links of all pages including this page
201 # This will be in a separate transaction
202 if ( $this->mRecursive
) {
203 $this->queueRecursiveJobs();
206 wfProfileOut( __METHOD__
);
209 function queueRecursiveJobs() {
210 global $wgUpdateRowsPerJob;
211 wfProfileIn( __METHOD__
);
213 $cache = $this->mTitle
->getBacklinkCache();
214 $batches = $cache->partition( 'templatelinks', $wgUpdateRowsPerJob );
216 wfProfileOut( __METHOD__
);
220 foreach ( $batches as $batch ) {
221 list( $start, $end ) = $batch;
226 $jobs[] = new RefreshLinksJob2( $this->mTitle
, $params );
228 Job
::batchInsert( $jobs );
230 wfProfileOut( __METHOD__
);
234 * Invalidate the cache of a list of pages from a single namespace
236 * @param $namespace Integer
237 * @param $dbkeys Array
239 function invalidatePages( $namespace, $dbkeys ) {
240 if ( !count( $dbkeys ) ) {
245 * Determine which pages need to be updated
246 * This is necessary to prevent the job queue from smashing the DB with
247 * large numbers of concurrent invalidations of the same page
249 $now = $this->mDb
->timestamp();
251 $res = $this->mDb
->select( 'page', array( 'page_id' ),
253 'page_namespace' => $namespace,
254 'page_title IN (' . $this->mDb
->makeList( $dbkeys ) . ')',
255 'page_touched < ' . $this->mDb
->addQuotes( $now )
258 foreach ( $res as $row ) {
259 $ids[] = $row->page_id
;
261 if ( !count( $ids ) ) {
267 * We still need the page_touched condition, in case the row has changed since
268 * the non-locking select above.
270 $this->mDb
->update( 'page', array( 'page_touched' => $now ),
272 'page_id IN (' . $this->mDb
->makeList( $ids ) . ')',
273 'page_touched < ' . $this->mDb
->addQuotes( $now )
278 function invalidateCategories( $cats ) {
279 $this->invalidatePages( NS_CATEGORY
, array_keys( $cats ) );
283 * Update all the appropriate counts in the category table.
284 * @param $added associative array of category name => sort key
285 * @param $deleted associative array of category name => sort key
287 function updateCategoryCounts( $added, $deleted ) {
288 $a = new Article($this->mTitle
);
289 $a->updateCategoryCounts(
290 array_keys( $added ), array_keys( $deleted )
294 function invalidateImageDescriptions( $images ) {
295 $this->invalidatePages( NS_FILE
, array_keys( $images ) );
298 function dumbTableUpdate( $table, $insertions, $fromField ) {
299 $this->mDb
->delete( $table, array( $fromField => $this->mId
), __METHOD__
);
300 if ( count( $insertions ) ) {
301 # The link array was constructed without FOR UPDATE, so there may
302 # be collisions. This may cause minor link table inconsistencies,
303 # which is better than crippling the site with lock contention.
304 $this->mDb
->insert( $table, $insertions, __METHOD__
, array( 'IGNORE' ) );
309 * Update a table by doing a delete query then an insert query
312 function incrTableUpdate( $table, $prefix, $deletions, $insertions ) {
313 if ( $table == 'page_props' ) {
314 $fromField = 'pp_page';
316 $fromField = "{$prefix}_from";
318 $where = array( $fromField => $this->mId
);
319 if ( $table == 'pagelinks' ||
$table == 'templatelinks' ||
$table == 'iwlinks' ) {
320 if ( $table == 'iwlinks' ) {
321 $baseKey = 'iwl_prefix';
323 $baseKey = "{$prefix}_namespace";
325 $clause = $this->mDb
->makeWhereFrom2d( $deletions, $baseKey, "{$prefix}_title" );
332 if ( $table == 'langlinks' ) {
333 $toField = 'll_lang';
334 } elseif ( $table == 'page_props' ) {
335 $toField = 'pp_propname';
337 $toField = $prefix . '_to';
339 if ( count( $deletions ) ) {
340 $where[] = "$toField IN (" . $this->mDb
->makeList( array_keys( $deletions ) ) . ')';
346 $this->mDb
->delete( $table, $where, __METHOD__
);
348 if ( count( $insertions ) ) {
349 $this->mDb
->insert( $table, $insertions, __METHOD__
, 'IGNORE' );
355 * Get an array of pagelinks insertions for passing to the DB
356 * Skips the titles specified by the 2-D array $existing
359 function getLinkInsertions( $existing = array() ) {
361 foreach( $this->mLinks
as $ns => $dbkeys ) {
362 $diffs = isset( $existing[$ns] ) ?
array_diff_key( $dbkeys, $existing[$ns] ) : $dbkeys;
363 foreach ( $diffs as $dbk => $id ) {
365 'pl_from' => $this->mId
,
366 'pl_namespace' => $ns,
375 * Get an array of template insertions. Like getLinkInsertions()
378 function getTemplateInsertions( $existing = array() ) {
380 foreach( $this->mTemplates
as $ns => $dbkeys ) {
381 $diffs = isset( $existing[$ns] ) ?
array_diff_key( $dbkeys, $existing[$ns] ) : $dbkeys;
382 foreach ( $diffs as $dbk => $id ) {
384 'tl_from' => $this->mId
,
385 'tl_namespace' => $ns,
394 * Get an array of image insertions
395 * Skips the names specified in $existing
398 function getImageInsertions( $existing = array() ) {
400 $diffs = array_diff_key( $this->mImages
, $existing );
401 foreach( $diffs as $iname => $dummy ) {
403 'il_from' => $this->mId
,
411 * Get an array of externallinks insertions. Skips the names specified in $existing
414 function getExternalInsertions( $existing = array() ) {
416 $diffs = array_diff_key( $this->mExternals
, $existing );
417 foreach( $diffs as $url => $dummy ) {
419 'el_from' => $this->mId
,
421 'el_index' => wfMakeUrlIndex( $url ),
428 * Get an array of category insertions
430 * @param $existing Array mapping existing category names to sort keys. If both
431 * match a link in $this, the link will be omitted from the output
434 function getCategoryInsertions( $existing = array() ) {
435 global $wgContLang, $wgCategoryCollation;
436 $diffs = array_diff_assoc( $this->mCategories
, $existing );
438 foreach ( $diffs as $name => $sortkey ) {
439 $nt = Title
::makeTitleSafe( NS_CATEGORY
, $name );
440 $wgContLang->findVariantLink( $name, $nt, true );
442 if ( $this->mTitle
->getNamespace() == NS_CATEGORY
) {
444 } elseif ( $this->mTitle
->getNamespace() == NS_FILE
) {
450 # TODO: This is kind of wrong, because someone might set a sort
451 # key prefix that's the same as the default sortkey for the
452 # title. This should be fixed by refactoring code to replace
453 # $sortkey in this array by a prefix, but it's basically harmless
454 # (Title::moveTo() has had the same issue for a long time).
455 if ( $this->mTitle
->getCategorySortkey() == $sortkey ) {
457 $sortkey = $wgContLang->convertToSortkey( $sortkey );
459 # Treat custom sortkeys as a prefix, so that if multiple
460 # things are forced to sort as '*' or something, they'll
461 # sort properly in the category rather than in page_id
464 $sortkey = $wgContLang->convertToSortkey(
465 $this->mTitle
->getCategorySortkey( $prefix ) );
469 'cl_from' => $this->mId
,
471 'cl_sortkey' => $sortkey,
472 'cl_timestamp' => $this->mDb
->timestamp(),
473 'cl_sortkey_prefix' => $prefix,
474 'cl_collation' => $wgCategoryCollation,
482 * Get an array of interlanguage link insertions
484 * @param $existing Array mapping existing language codes to titles
487 function getInterlangInsertions( $existing = array() ) {
488 $diffs = array_diff_assoc( $this->mInterlangs
, $existing );
490 foreach( $diffs as $lang => $title ) {
492 'll_from' => $this->mId
,
501 * Get an array of page property insertions
503 function getPropertyInsertions( $existing = array() ) {
504 $diffs = array_diff_assoc( $this->mProperties
, $existing );
506 foreach ( $diffs as $name => $value ) {
508 'pp_page' => $this->mId
,
509 'pp_propname' => $name,
510 'pp_value' => $value,
517 * Get an array of interwiki insertions for passing to the DB
518 * Skips the titles specified by the 2-D array $existing
521 function getInterwikiInsertions( $existing = array() ) {
523 foreach( $this->mInterwikis
as $prefix => $dbkeys ) {
524 $diffs = isset( $existing[$prefix] ) ?
array_diff_key( $dbkeys, $existing[$prefix] ) : $dbkeys;
525 foreach ( $diffs as $dbk => $id ) {
527 'iwl_from' => $this->mId
,
528 'iwl_prefix' => $prefix,
537 * Given an array of existing links, returns those links which are not in $this
538 * and thus should be deleted.
541 function getLinkDeletions( $existing ) {
543 foreach ( $existing as $ns => $dbkeys ) {
544 if ( isset( $this->mLinks
[$ns] ) ) {
545 $del[$ns] = array_diff_key( $existing[$ns], $this->mLinks
[$ns] );
547 $del[$ns] = $existing[$ns];
554 * Given an array of existing templates, returns those templates which are not in $this
555 * and thus should be deleted.
558 function getTemplateDeletions( $existing ) {
560 foreach ( $existing as $ns => $dbkeys ) {
561 if ( isset( $this->mTemplates
[$ns] ) ) {
562 $del[$ns] = array_diff_key( $existing[$ns], $this->mTemplates
[$ns] );
564 $del[$ns] = $existing[$ns];
571 * Given an array of existing images, returns those images which are not in $this
572 * and thus should be deleted.
575 function getImageDeletions( $existing ) {
576 return array_diff_key( $existing, $this->mImages
);
580 * Given an array of existing external links, returns those links which are not
581 * in $this and thus should be deleted.
584 function getExternalDeletions( $existing ) {
585 return array_diff_key( $existing, $this->mExternals
);
589 * Given an array of existing categories, returns those categories which are not in $this
590 * and thus should be deleted.
593 function getCategoryDeletions( $existing ) {
594 return array_diff_assoc( $existing, $this->mCategories
);
598 * Given an array of existing interlanguage links, returns those links which are not
599 * in $this and thus should be deleted.
602 function getInterlangDeletions( $existing ) {
603 return array_diff_assoc( $existing, $this->mInterlangs
);
607 * Get array of properties which should be deleted.
610 function getPropertyDeletions( $existing ) {
611 return array_diff_assoc( $existing, $this->mProperties
);
615 * Given an array of existing interwiki links, returns those links which are not in $this
616 * and thus should be deleted.
619 function getInterwikiDeletions( $existing ) {
621 foreach ( $existing as $prefix => $dbkeys ) {
622 if ( isset( $this->mInterwikis
[$prefix] ) ) {
623 $del[$prefix] = array_diff_key( $existing[$prefix], $this->mInterwikis
[$prefix] );
625 $del[$prefix] = $existing[$prefix];
632 * Get an array of existing links, as a 2-D array
635 function getExistingLinks() {
636 $res = $this->mDb
->select( 'pagelinks', array( 'pl_namespace', 'pl_title' ),
637 array( 'pl_from' => $this->mId
), __METHOD__
, $this->mOptions
);
639 foreach ( $res as $row ) {
640 if ( !isset( $arr[$row->pl_namespace
] ) ) {
641 $arr[$row->pl_namespace
] = array();
643 $arr[$row->pl_namespace
][$row->pl_title
] = 1;
649 * Get an array of existing templates, as a 2-D array
652 function getExistingTemplates() {
653 $res = $this->mDb
->select( 'templatelinks', array( 'tl_namespace', 'tl_title' ),
654 array( 'tl_from' => $this->mId
), __METHOD__
, $this->mOptions
);
656 foreach ( $res as $row ) {
657 if ( !isset( $arr[$row->tl_namespace
] ) ) {
658 $arr[$row->tl_namespace
] = array();
660 $arr[$row->tl_namespace
][$row->tl_title
] = 1;
666 * Get an array of existing images, image names in the keys
669 function getExistingImages() {
670 $res = $this->mDb
->select( 'imagelinks', array( 'il_to' ),
671 array( 'il_from' => $this->mId
), __METHOD__
, $this->mOptions
);
673 foreach ( $res as $row ) {
674 $arr[$row->il_to
] = 1;
680 * Get an array of existing external links, URLs in the keys
683 function getExistingExternals() {
684 $res = $this->mDb
->select( 'externallinks', array( 'el_to' ),
685 array( 'el_from' => $this->mId
), __METHOD__
, $this->mOptions
);
687 foreach ( $res as $row ) {
688 $arr[$row->el_to
] = 1;
694 * Get an array of existing categories, with the name in the key and sort key in the value.
697 function getExistingCategories() {
698 $res = $this->mDb
->select( 'categorylinks', array( 'cl_to', 'cl_sortkey_prefix' ),
699 array( 'cl_from' => $this->mId
), __METHOD__
, $this->mOptions
);
701 foreach ( $res as $row ) {
702 if ( $row->cl_sortkey_prefix
!== '' ) {
703 $arr[$row->cl_to
] = $row->cl_sortkey_prefix
;
705 $arr[$row->cl_to
] = $this->mTitle
->getCategorySortkey();
712 * Get an array of existing interlanguage links, with the language code in the key and the
713 * title in the value.
716 function getExistingInterlangs() {
717 $res = $this->mDb
->select( 'langlinks', array( 'll_lang', 'll_title' ),
718 array( 'll_from' => $this->mId
), __METHOD__
, $this->mOptions
);
720 foreach ( $res as $row ) {
721 $arr[$row->ll_lang
] = $row->ll_title
;
727 * Get an array of existing inline interwiki links, as a 2-D array
728 * @return array (prefix => array(dbkey => 1))
730 protected function getExistingInterwikis() {
731 $res = $this->mDb
->select( 'iwlinks', array( 'iwl_prefix', 'iwl_title' ),
732 array( 'iwl_from' => $this->mId
), __METHOD__
, $this->mOptions
);
734 foreach ( $res as $row ) {
735 if ( !isset( $arr[$row->iwl_prefix
] ) ) {
736 $arr[$row->iwl_prefix
] = array();
738 $arr[$row->iwl_prefix
][$row->iwl_title
] = 1;
744 * Get an array of existing categories, with the name in the key and sort key in the value.
747 function getExistingProperties() {
748 $res = $this->mDb
->select( 'page_props', array( 'pp_propname', 'pp_value' ),
749 array( 'pp_page' => $this->mId
), __METHOD__
, $this->mOptions
);
751 foreach ( $res as $row ) {
752 $arr[$row->pp_propname
] = $row->pp_value
;
759 * Return the title object of the page being updated
761 function getTitle() {
762 return $this->mTitle
;
766 * Return the list of images used as generated by the parser
768 public function getImages() {
769 return $this->mImages
;
773 * Invalidate any necessary link lists related to page property changes
775 function invalidateProperties( $changed ) {
776 global $wgPagePropLinkInvalidations;
778 foreach ( $changed as $name => $value ) {
779 if ( isset( $wgPagePropLinkInvalidations[$name] ) ) {
780 $inv = $wgPagePropLinkInvalidations[$name];
781 if ( !is_array( $inv ) ) {
782 $inv = array( $inv );
784 foreach ( $inv as $table ) {
785 $update = new HTMLCacheUpdate( $this->mTitle
, $table );