3 * Job to update link tables for pages
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
25 * Job to update link tables for pages
27 * This job comes in a few variants:
28 * - a) Recursive jobs to update links for backlink pages for a given title.
29 * These jobs have have (recursive:true,table:<table>) set.
30 * - b) Jobs to update links for a set of pages (the job title is ignored).
31 * These jobs have have (pages:(<page ID>:(<namespace>,<title>),...) set.
32 * - c) Jobs to update links for a single page (the job title)
33 * These jobs need no extra fields set.
37 class RefreshLinksJob
extends Job
{
38 function __construct( $title, $params = '' ) {
39 parent
::__construct( 'refreshLinks', $title, $params );
40 // Base backlink update jobs and per-title update jobs can be de-duplicated.
41 // If template A changes twice before any jobs run, a clean queue will have:
43 // The second job is ignored by the queue on insertion.
44 // Suppose, many pages use template A, and that template itself uses template B.
45 // An edit to both will first create two base jobs. A clean FIFO queue will have:
47 // When these jobs run, the queue will have per-title and remnant partition jobs:
48 // (titleX,titleY,titleZ,...,A remnant,titleM,titleN,titleO,...,B remnant)
49 // Some these jobs will be the same, and will automatically be ignored by
50 // the queue upon insertion. Some title jobs will run before the duplicate is
51 // inserted, so the work will still be done twice in those cases. More titles
52 // can be de-duplicated as the remnant jobs continue to be broken down. This
53 // works best when $wgUpdateRowsPerJob, and either the pages have few backlinks
54 // and/or the backlink sets for pages A and B are almost identical.
55 $this->removeDuplicates
= !isset( $params['range'] )
56 && ( !isset( $params['pages'] ) ||
count( $params['pages'] ) == 1 );
60 global $wgUpdateRowsPerJob;
62 // Job to update all (or a range of) backlink pages for a page
63 if ( !empty( $this->params
['recursive'] ) ) {
64 // Carry over information for de-duplication
65 $extraParams = $this->getRootJobParams();
66 // Avoid slave lag when fetching templates.
67 // When the outermost job is run, we know that the caller that enqueued it must have
68 // committed the relevant changes to the DB by now. At that point, record the master
69 // position and pass it along as the job recursively breaks into smaller range jobs.
70 // Hopefully, when leaf jobs are popped, the slaves will have reached that position.
71 if ( isset( $this->params
['masterPos'] ) ) {
72 $extraParams['masterPos'] = $this->params
['masterPos'];
73 } elseif ( wfGetLB()->getServerCount() > 1 ) {
74 $extraParams['masterPos'] = wfGetLB()->getMasterPos();
76 $extraParams['masterPos'] = false;
78 // Convert this into no more than $wgUpdateRowsPerJob RefreshLinks per-title
79 // jobs and possibly a recursive RefreshLinks job for the rest of the backlinks
80 $jobs = BacklinkJobUtils
::partitionBacklinkJob(
84 array( 'params' => $extraParams )
86 JobQueueGroup
::singleton()->push( $jobs );
87 // Job to update link tables for for a set of titles
88 } elseif ( isset( $this->params
['pages'] ) ) {
89 foreach ( $this->params
['pages'] as $pageId => $nsAndKey ) {
90 list( $ns, $dbKey ) = $nsAndKey;
91 $this->runForTitle( Title
::makeTitleSafe( $ns, $dbKey ) );
93 // Job to update link tables for a given title
95 $this->runForTitle( $this->title
);
101 protected function runForTitle( Title
$title = null ) {
102 $linkCache = LinkCache
::singleton();
105 if ( is_null( $title ) ) {
106 $this->setLastError( "refreshLinks: Invalid title" );
110 // Wait for the DB of the current/next slave DB handle to catch up to the master.
111 // This way, we get the correct page_latest for templates or files that just changed
112 // milliseconds ago, having triggered this job to begin with.
113 if ( isset( $this->params
['masterPos'] ) && $this->params
['masterPos'] !== false ) {
114 wfGetLB()->waitFor( $this->params
['masterPos'] );
117 // Fetch the current revision...
118 $revision = Revision
::newFromTitle( $title, false, Revision
::READ_NORMAL
);
120 $this->setLastError( "refreshLinks: Article not found {$title->getPrefixedDBkey()}" );
121 return false; // XXX: what if it was just deleted?
123 $content = $revision->getContent( Revision
::RAW
);
125 // If there is no content, pretend the content is empty
126 $content = $revision->getContentHandler()->makeEmptyContent();
129 $parserOutput = false;
130 // If page_touched changed after this root job (with a good slave lag skew factor),
131 // then it is likely that any views of the pages already resulted in re-parses which
132 // are now in cache. This can be reused to avoid expensive parsing in some cases.
133 if ( isset( $this->params
['rootJobTimestamp'] ) ) {
134 $page = WikiPage
::factory( $title );
135 $skewedTimestamp = wfTimestamp( TS_UNIX
, $this->params
['rootJobTimestamp'] ) +
5;
136 if ( $page->getTouched() > wfTimestamp( TS_MW
, $skewedTimestamp ) ) {
137 $parserOptions = $page->makeParserOptions( 'canonical' );
138 $parserOutput = ParserCache
::singleton()->getDirty( $page, $parserOptions );
139 if ( $parserOutput && $parserOutput->getCacheTime() <= $skewedTimestamp ) {
140 $parserOutput = false; // too stale
144 // Fetch the current revision and parse it if necessary...
145 if ( $parserOutput == false ) {
146 // Revision ID must be passed to the parser output to get revision variables correct
147 $parserOutput = $content->getParserOutput( $title, $revision->getId(), null, false );
150 $updates = $content->getSecondaryDataUpdates( $title, null, false, $parserOutput );
151 DataUpdate
::runUpdates( $updates );
153 InfoAction
::invalidateCache( $title );
158 public function getDeduplicationInfo() {
159 $info = parent
::getDeduplicationInfo();
160 if ( is_array( $info['params'] ) ) {
161 // Don't let highly unique "masterPos" values ruin duplicate detection
162 unset( $info['params']['masterPos'] );
163 // For per-pages jobs, the job title is that of the template that changed
164 // (or similar), so remove that since it ruins duplicate detection
165 if ( isset( $info['pages'] ) ) {
166 unset( $info['namespace'] );
167 unset( $info['title'] );