3 * Updater for link tracking tables after a page edit.
5 * This program is free software; you can redistribute it and/or modify
6 * it under the terms of the GNU General Public License as published by
7 * the Free Software Foundation; either version 2 of the License, or
8 * (at your option) any later version.
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 * GNU General Public License for more details.
15 * You should have received a copy of the GNU General Public License along
16 * with this program; if not, write to the Free Software Foundation, Inc.,
17 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
18 * http://www.gnu.org/copyleft/gpl.html
24 * See docs/deferred.txt
26 * @todo document (e.g. one-sentence top-level class description).
28 class LinksUpdate
extends SqlDataUpdate
{
29 // @todo make members protected, but make sure extensions don't break
31 /** @var int Page ID of the article linked from */
34 /** @var Title Title object of the article linked from */
37 /** @var ParserOutput */
38 public $mParserOutput;
40 /** @var array Map of title strings to IDs for the links in the document */
43 /** @var array DB keys of the images used, in the array key only */
46 /** @var array Map of title strings to IDs for the template references, including broken ones */
49 /** @var array URLs of external links, array key only */
52 /** @var array Map of category names to sort keys */
55 /** @var array Map of language codes to titles */
58 /** @var array Map of arbitrary name to value */
61 /** @var bool Whether to queue jobs for recursive updates */
65 * @var null|array Added links if calculated.
67 private $linkInsertions = null;
70 * @var null|array Deleted links if calculated.
72 private $linkDeletions = null;
77 * @param Title $title Title of the page we're updating
78 * @param ParserOutput $parserOutput Output from a full parse of this page
79 * @param bool $recursive Queue jobs for recursive updates?
82 function __construct( $title, $parserOutput, $recursive = true ) {
83 parent
::__construct( false ); // no implicit transaction
85 if ( !( $title instanceof Title
) ) {
86 throw new MWException( "The calling convention to LinksUpdate::LinksUpdate() has changed. " .
87 "Please see Article::editUpdates() for an invocation example.\n" );
90 if ( !( $parserOutput instanceof ParserOutput
) ) {
91 throw new MWException( "The calling convention to LinksUpdate::__construct() has changed. " .
92 "Please see WikiPage::doEditUpdates() for an invocation example.\n" );
95 $this->mTitle
= $title;
96 $this->mId
= $title->getArticleID();
99 throw new MWException( "The Title object did not provide an article " .
100 "ID. Perhaps the page doesn't exist?" );
103 $this->mParserOutput
= $parserOutput;
105 $this->mLinks
= $parserOutput->getLinks();
106 $this->mImages
= $parserOutput->getImages();
107 $this->mTemplates
= $parserOutput->getTemplates();
108 $this->mExternals
= $parserOutput->getExternalLinks();
109 $this->mCategories
= $parserOutput->getCategories();
110 $this->mProperties
= $parserOutput->getProperties();
111 $this->mInterwikis
= $parserOutput->getInterwikiLinks();
113 # Convert the format of the interlanguage links
114 # I didn't want to change it in the ParserOutput, because that array is passed all
115 # the way back to the skin, so either a skin API break would be required, or an
116 # inefficient back-conversion.
117 $ill = $parserOutput->getLanguageLinks();
118 $this->mInterlangs
= array();
119 foreach ( $ill as $link ) {
120 list( $key, $title ) = explode( ':', $link, 2 );
121 $this->mInterlangs
[$key] = $title;
124 foreach ( $this->mCategories
as &$sortkey ) {
125 # If the sortkey is longer then 255 bytes,
126 # it truncated by DB, and then doesn't get
127 # matched when comparing existing vs current
128 # categories, causing bug 25254.
129 # Also. substr behaves weird when given "".
130 if ( $sortkey !== '' ) {
131 $sortkey = substr( $sortkey, 0, 255 );
135 $this->mRecursive
= $recursive;
137 Hooks
::run( 'LinksUpdateConstructed', array( &$this ) );
141 * Update link tables with outgoing links from an updated article
143 public function doUpdate() {
144 Hooks
::run( 'LinksUpdate', array( &$this ) );
145 $this->doIncrementalUpdate();
146 Hooks
::run( 'LinksUpdateComplete', array( &$this ) );
149 protected function doIncrementalUpdate() {
152 $existing = $this->getExistingLinks();
153 $this->linkDeletions
= $this->getLinkDeletions( $existing );
154 $this->linkInsertions
= $this->getLinkInsertions( $existing );
155 $this->incrTableUpdate( 'pagelinks', 'pl', $this->linkDeletions
, $this->linkInsertions
);
158 $existing = $this->getExistingImages();
160 $imageDeletes = $this->getImageDeletions( $existing );
161 $this->incrTableUpdate( 'imagelinks', 'il', $imageDeletes,
162 $this->getImageInsertions( $existing ) );
164 # Invalidate all image description pages which had links added or removed
165 $imageUpdates = $imageDeletes +
array_diff_key( $this->mImages
, $existing );
166 $this->invalidateImageDescriptions( $imageUpdates );
169 $existing = $this->getExistingExternals();
170 $this->incrTableUpdate( 'externallinks', 'el', $this->getExternalDeletions( $existing ),
171 $this->getExternalInsertions( $existing ) );
174 $existing = $this->getExistingInterlangs();
175 $this->incrTableUpdate( 'langlinks', 'll', $this->getInterlangDeletions( $existing ),
176 $this->getInterlangInsertions( $existing ) );
178 # Inline interwiki links
179 $existing = $this->getExistingInterwikis();
180 $this->incrTableUpdate( 'iwlinks', 'iwl', $this->getInterwikiDeletions( $existing ),
181 $this->getInterwikiInsertions( $existing ) );
184 $existing = $this->getExistingTemplates();
185 $this->incrTableUpdate( 'templatelinks', 'tl', $this->getTemplateDeletions( $existing ),
186 $this->getTemplateInsertions( $existing ) );
189 $existing = $this->getExistingCategories();
191 $categoryDeletes = $this->getCategoryDeletions( $existing );
193 $this->incrTableUpdate( 'categorylinks', 'cl', $categoryDeletes,
194 $this->getCategoryInsertions( $existing ) );
196 # Invalidate all categories which were added, deleted or changed (set symmetric difference)
197 $categoryInserts = array_diff_assoc( $this->mCategories
, $existing );
198 $categoryUpdates = $categoryInserts +
$categoryDeletes;
199 $this->invalidateCategories( $categoryUpdates );
200 $this->updateCategoryCounts( $categoryInserts, $categoryDeletes );
203 $existing = $this->getExistingProperties();
205 $propertiesDeletes = $this->getPropertyDeletions( $existing );
207 $this->incrTableUpdate( 'page_props', 'pp', $propertiesDeletes,
208 $this->getPropertyInsertions( $existing ) );
210 # Invalidate the necessary pages
211 $changed = $propertiesDeletes +
array_diff_assoc( $this->mProperties
, $existing );
212 $this->invalidateProperties( $changed );
214 # Update the links table freshness for this title
215 $this->updateLinksTimestamp();
217 # Refresh links of all pages including this page
218 # This will be in a separate transaction
219 if ( $this->mRecursive
) {
220 $this->queueRecursiveJobs();
226 * Queue recursive jobs for this page
228 * Which means do LinksUpdate on all pages that include the current page,
229 * using the job queue.
231 protected function queueRecursiveJobs() {
232 self
::queueRecursiveJobsForTable( $this->mTitle
, 'templatelinks' );
233 if ( $this->mTitle
->getNamespace() == NS_FILE
) {
234 // Process imagelinks in case the title is or was a redirect
235 self
::queueRecursiveJobsForTable( $this->mTitle
, 'imagelinks' );
238 $bc = $this->mTitle
->getBacklinkCache();
239 // Get jobs for cascade-protected backlinks for a high priority queue.
240 // If meta-templates change to using a new template, the new template
241 // should be implicitly protected as soon as possible, if applicable.
242 // These jobs duplicate a subset of the above ones, but can run sooner.
243 // Which ever runs first generally no-ops the other one.
245 foreach ( $bc->getCascadeProtectedLinks() as $title ) {
246 $jobs[] = new RefreshLinksJob( $title, array( 'prioritize' => true ) );
248 JobQueueGroup
::singleton()->push( $jobs );
252 * Queue a RefreshLinks job for any table.
254 * @param Title $title Title to do job for
255 * @param string $table Table to use (e.g. 'templatelinks')
257 public static function queueRecursiveJobsForTable( Title
$title, $table ) {
258 if ( $title->getBacklinkCache()->hasLinks( $table ) ) {
259 $job = new RefreshLinksJob(
264 ) + Job
::newRootJobParams( // "overall" refresh links job info
265 "refreshlinks:{$table}:{$title->getPrefixedText()}"
269 JobQueueGroup
::singleton()->push( $job );
276 function invalidateCategories( $cats ) {
277 $this->invalidatePages( NS_CATEGORY
, array_keys( $cats ) );
281 * Update all the appropriate counts in the category table.
282 * @param array $added Associative array of category name => sort key
283 * @param array $deleted Associative array of category name => sort key
285 function updateCategoryCounts( $added, $deleted ) {
286 $a = WikiPage
::factory( $this->mTitle
);
287 $a->updateCategoryCounts(
288 array_keys( $added ), array_keys( $deleted )
293 * @param array $images
295 function invalidateImageDescriptions( $images ) {
296 $this->invalidatePages( NS_FILE
, array_keys( $images ) );
300 * Update a table by doing a delete query then an insert query
301 * @param string $table Table name
302 * @param string $prefix Field name prefix
303 * @param array $deletions
304 * @param array $insertions Rows to insert
306 function incrTableUpdate( $table, $prefix, $deletions, $insertions ) {
307 if ( $table == 'page_props' ) {
308 $fromField = 'pp_page';
310 $fromField = "{$prefix}_from";
312 $where = array( $fromField => $this->mId
);
313 if ( $table == 'pagelinks' ||
$table == 'templatelinks' ||
$table == 'iwlinks' ) {
314 if ( $table == 'iwlinks' ) {
315 $baseKey = 'iwl_prefix';
317 $baseKey = "{$prefix}_namespace";
319 $clause = $this->mDb
->makeWhereFrom2d( $deletions, $baseKey, "{$prefix}_title" );
326 if ( $table == 'langlinks' ) {
327 $toField = 'll_lang';
328 } elseif ( $table == 'page_props' ) {
329 $toField = 'pp_propname';
331 $toField = $prefix . '_to';
333 if ( count( $deletions ) ) {
334 $where[$toField] = array_keys( $deletions );
340 $this->mDb
->delete( $table, $where, __METHOD__
);
342 if ( count( $insertions ) ) {
343 $this->mDb
->insert( $table, $insertions, __METHOD__
, 'IGNORE' );
344 Hooks
::run( 'LinksUpdateAfterInsert', array( $this, $table, $insertions ) );
349 * Get an array of pagelinks insertions for passing to the DB
350 * Skips the titles specified by the 2-D array $existing
351 * @param array $existing
354 private function getLinkInsertions( $existing = array() ) {
356 foreach ( $this->mLinks
as $ns => $dbkeys ) {
357 $diffs = isset( $existing[$ns] )
358 ?
array_diff_key( $dbkeys, $existing[$ns] )
360 foreach ( $diffs as $dbk => $id ) {
362 'pl_from' => $this->mId
,
363 'pl_from_namespace' => $this->mTitle
->getNamespace(),
364 'pl_namespace' => $ns,
374 * Get an array of template insertions. Like getLinkInsertions()
375 * @param array $existing
378 private 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_from_namespace' => $this->mTitle
->getNamespace(),
386 'tl_namespace' => $ns,
396 * Get an array of image insertions
397 * Skips the names specified in $existing
398 * @param array $existing
401 private function getImageInsertions( $existing = array() ) {
403 $diffs = array_diff_key( $this->mImages
, $existing );
404 foreach ( $diffs as $iname => $dummy ) {
406 'il_from' => $this->mId
,
407 'il_from_namespace' => $this->mTitle
->getNamespace(),
416 * Get an array of externallinks insertions. Skips the names specified in $existing
417 * @param array $existing
420 private function getExternalInsertions( $existing = array() ) {
422 $diffs = array_diff_key( $this->mExternals
, $existing );
423 foreach ( $diffs as $url => $dummy ) {
424 foreach ( wfMakeUrlIndexes( $url ) as $index ) {
426 'el_id' => $this->mDb
->nextSequenceValue( 'externallinks_el_id_seq' ),
427 'el_from' => $this->mId
,
429 'el_index' => $index,
438 * Get an array of category insertions
440 * @param array $existing Mapping existing category names to sort keys. If both
441 * match a link in $this, the link will be omitted from the output
445 private function getCategoryInsertions( $existing = array() ) {
446 global $wgContLang, $wgCategoryCollation;
447 $diffs = array_diff_assoc( $this->mCategories
, $existing );
449 foreach ( $diffs as $name => $prefix ) {
450 $nt = Title
::makeTitleSafe( NS_CATEGORY
, $name );
451 $wgContLang->findVariantLink( $name, $nt, true );
453 if ( $this->mTitle
->getNamespace() == NS_CATEGORY
) {
455 } elseif ( $this->mTitle
->getNamespace() == NS_FILE
) {
461 # Treat custom sortkeys as a prefix, so that if multiple
462 # things are forced to sort as '*' or something, they'll
463 # sort properly in the category rather than in page_id
465 $sortkey = Collation
::singleton()->getSortKey(
466 $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,
483 * Get an array of interlanguage link insertions
485 * @param array $existing Mapping existing language codes to titles
489 private function getInterlangInsertions( $existing = array() ) {
490 $diffs = array_diff_assoc( $this->mInterlangs
, $existing );
492 foreach ( $diffs as $lang => $title ) {
494 'll_from' => $this->mId
,
504 * Get an array of page property insertions
505 * @param array $existing
508 function getPropertyInsertions( $existing = array() ) {
509 $diffs = array_diff_assoc( $this->mProperties
, $existing );
512 foreach ( array_keys( $diffs ) as $name ) {
513 $arr[] = $this->getPagePropRowData( $name );
520 * Returns an associative array to be used for inserting a row into
521 * the page_props table. Besides the given property name, this will
522 * include the page id from $this->mId and any property value from
523 * $this->mProperties.
525 * The array returned will include the pp_sortkey field if this
526 * is present in the database (as indicated by $wgPagePropsHaveSortkey).
527 * The sortkey value is currently determined by getPropertySortKeyValue().
529 * @note this assumes that $this->mProperties[$prop] is defined.
531 * @param string $prop The name of the property.
535 private function getPagePropRowData( $prop ) {
536 global $wgPagePropsHaveSortkey;
538 $value = $this->mProperties
[$prop];
541 'pp_page' => $this->mId
,
542 'pp_propname' => $prop,
543 'pp_value' => $value,
546 if ( $wgPagePropsHaveSortkey ) {
547 $row['pp_sortkey'] = $this->getPropertySortKeyValue( $value );
554 * Determines the sort key for the given property value.
555 * This will return $value if it is a float or int,
556 * 1 or resp. 0 if it is a bool, and null otherwise.
558 * @note In the future, we may allow the sortkey to be specified explicitly
559 * in ParserOutput::setProperty.
561 * @param mixed $value
565 private function getPropertySortKeyValue( $value ) {
566 if ( is_int( $value ) ||
is_float( $value ) ||
is_bool( $value ) ) {
567 return floatval( $value );
574 * Get an array of interwiki insertions for passing to the DB
575 * Skips the titles specified by the 2-D array $existing
576 * @param array $existing
579 private function getInterwikiInsertions( $existing = array() ) {
581 foreach ( $this->mInterwikis
as $prefix => $dbkeys ) {
582 $diffs = isset( $existing[$prefix] )
583 ?
array_diff_key( $dbkeys, $existing[$prefix] )
586 foreach ( $diffs as $dbk => $id ) {
588 'iwl_from' => $this->mId
,
589 'iwl_prefix' => $prefix,
599 * Given an array of existing links, returns those links which are not in $this
600 * and thus should be deleted.
601 * @param array $existing
604 private function getLinkDeletions( $existing ) {
606 foreach ( $existing as $ns => $dbkeys ) {
607 if ( isset( $this->mLinks
[$ns] ) ) {
608 $del[$ns] = array_diff_key( $existing[$ns], $this->mLinks
[$ns] );
610 $del[$ns] = $existing[$ns];
618 * Given an array of existing templates, returns those templates which are not in $this
619 * and thus should be deleted.
620 * @param array $existing
623 private function getTemplateDeletions( $existing ) {
625 foreach ( $existing as $ns => $dbkeys ) {
626 if ( isset( $this->mTemplates
[$ns] ) ) {
627 $del[$ns] = array_diff_key( $existing[$ns], $this->mTemplates
[$ns] );
629 $del[$ns] = $existing[$ns];
637 * Given an array of existing images, returns those images which are not in $this
638 * and thus should be deleted.
639 * @param array $existing
642 private function getImageDeletions( $existing ) {
643 return array_diff_key( $existing, $this->mImages
);
647 * Given an array of existing external links, returns those links which are not
648 * in $this and thus should be deleted.
649 * @param array $existing
652 private function getExternalDeletions( $existing ) {
653 return array_diff_key( $existing, $this->mExternals
);
657 * Given an array of existing categories, returns those categories which are not in $this
658 * and thus should be deleted.
659 * @param array $existing
662 private function getCategoryDeletions( $existing ) {
663 return array_diff_assoc( $existing, $this->mCategories
);
667 * Given an array of existing interlanguage links, returns those links which are not
668 * in $this and thus should be deleted.
669 * @param array $existing
672 private function getInterlangDeletions( $existing ) {
673 return array_diff_assoc( $existing, $this->mInterlangs
);
677 * Get array of properties which should be deleted.
678 * @param array $existing
681 function getPropertyDeletions( $existing ) {
682 return array_diff_assoc( $existing, $this->mProperties
);
686 * Given an array of existing interwiki links, returns those links which are not in $this
687 * and thus should be deleted.
688 * @param array $existing
691 private function getInterwikiDeletions( $existing ) {
693 foreach ( $existing as $prefix => $dbkeys ) {
694 if ( isset( $this->mInterwikis
[$prefix] ) ) {
695 $del[$prefix] = array_diff_key( $existing[$prefix], $this->mInterwikis
[$prefix] );
697 $del[$prefix] = $existing[$prefix];
705 * Get an array of existing links, as a 2-D array
709 private function getExistingLinks() {
710 $res = $this->mDb
->select( 'pagelinks', array( 'pl_namespace', 'pl_title' ),
711 array( 'pl_from' => $this->mId
), __METHOD__
, $this->mOptions
);
713 foreach ( $res as $row ) {
714 if ( !isset( $arr[$row->pl_namespace
] ) ) {
715 $arr[$row->pl_namespace
] = array();
717 $arr[$row->pl_namespace
][$row->pl_title
] = 1;
724 * Get an array of existing templates, as a 2-D array
728 private function getExistingTemplates() {
729 $res = $this->mDb
->select( 'templatelinks', array( 'tl_namespace', 'tl_title' ),
730 array( 'tl_from' => $this->mId
), __METHOD__
, $this->mOptions
);
732 foreach ( $res as $row ) {
733 if ( !isset( $arr[$row->tl_namespace
] ) ) {
734 $arr[$row->tl_namespace
] = array();
736 $arr[$row->tl_namespace
][$row->tl_title
] = 1;
743 * Get an array of existing images, image names in the keys
747 private function getExistingImages() {
748 $res = $this->mDb
->select( 'imagelinks', array( 'il_to' ),
749 array( 'il_from' => $this->mId
), __METHOD__
, $this->mOptions
);
751 foreach ( $res as $row ) {
752 $arr[$row->il_to
] = 1;
759 * Get an array of existing external links, URLs in the keys
763 private function getExistingExternals() {
764 $res = $this->mDb
->select( 'externallinks', array( 'el_to' ),
765 array( 'el_from' => $this->mId
), __METHOD__
, $this->mOptions
);
767 foreach ( $res as $row ) {
768 $arr[$row->el_to
] = 1;
775 * Get an array of existing categories, with the name in the key and sort key in the value.
779 private function getExistingCategories() {
780 $res = $this->mDb
->select( 'categorylinks', array( 'cl_to', 'cl_sortkey_prefix' ),
781 array( 'cl_from' => $this->mId
), __METHOD__
, $this->mOptions
);
783 foreach ( $res as $row ) {
784 $arr[$row->cl_to
] = $row->cl_sortkey_prefix
;
791 * Get an array of existing interlanguage links, with the language code in the key and the
792 * title in the value.
796 private function getExistingInterlangs() {
797 $res = $this->mDb
->select( 'langlinks', array( 'll_lang', 'll_title' ),
798 array( 'll_from' => $this->mId
), __METHOD__
, $this->mOptions
);
800 foreach ( $res as $row ) {
801 $arr[$row->ll_lang
] = $row->ll_title
;
808 * Get an array of existing inline interwiki links, as a 2-D array
809 * @return array (prefix => array(dbkey => 1))
811 protected function getExistingInterwikis() {
812 $res = $this->mDb
->select( 'iwlinks', array( 'iwl_prefix', 'iwl_title' ),
813 array( 'iwl_from' => $this->mId
), __METHOD__
, $this->mOptions
);
815 foreach ( $res as $row ) {
816 if ( !isset( $arr[$row->iwl_prefix
] ) ) {
817 $arr[$row->iwl_prefix
] = array();
819 $arr[$row->iwl_prefix
][$row->iwl_title
] = 1;
826 * Get an array of existing categories, with the name in the key and sort key in the value.
828 * @return array Array of property names and values
830 private function getExistingProperties() {
831 $res = $this->mDb
->select( 'page_props', array( 'pp_propname', 'pp_value' ),
832 array( 'pp_page' => $this->mId
), __METHOD__
, $this->mOptions
);
834 foreach ( $res as $row ) {
835 $arr[$row->pp_propname
] = $row->pp_value
;
842 * Return the title object of the page being updated
845 public function getTitle() {
846 return $this->mTitle
;
850 * Returns parser output
852 * @return ParserOutput
854 public function getParserOutput() {
855 return $this->mParserOutput
;
859 * Return the list of images used as generated by the parser
862 public function getImages() {
863 return $this->mImages
;
867 * Invalidate any necessary link lists related to page property changes
868 * @param array $changed
870 private function invalidateProperties( $changed ) {
871 global $wgPagePropLinkInvalidations;
873 foreach ( $changed as $name => $value ) {
874 if ( isset( $wgPagePropLinkInvalidations[$name] ) ) {
875 $inv = $wgPagePropLinkInvalidations[$name];
876 if ( !is_array( $inv ) ) {
877 $inv = array( $inv );
879 foreach ( $inv as $table ) {
880 $update = new HTMLCacheUpdate( $this->mTitle
, $table );
888 * Fetch page links added by this LinksUpdate. Only available after the update is complete.
890 * @return null|array Array of Titles
892 public function getAddedLinks() {
893 if ( $this->linkInsertions
=== null ) {
897 foreach ( $this->linkInsertions
as $insertion ) {
898 $result[] = Title
::makeTitle( $insertion['pl_namespace'], $insertion['pl_title'] );
905 * Fetch page links removed by this LinksUpdate. Only available after the update is complete.
907 * @return null|array Array of Titles
909 public function getRemovedLinks() {
910 if ( $this->linkDeletions
=== null ) {
914 foreach ( $this->linkDeletions
as $ns => $titles ) {
915 foreach ( $titles as $title => $unused ) {
916 $result[] = Title
::makeTitle( $ns, $title );
924 * Update links table freshness
926 protected function updateLinksTimestamp() {
928 // The link updates made here only reflect the freshness of the parser output
929 $timestamp = $this->mParserOutput
->getCacheTime();
930 $this->mDb
->update( 'page',
931 array( 'page_links_updated' => $this->mDb
->timestamp( $timestamp ) ),
932 array( 'page_id' => $this->mId
),