<?php
+/**
+ * @todo document
+ * @addtogroup Maintenance
+ */
-define( "REPORTING_INTERVAL", 50 );
-define( "PAUSE_INTERVAL", 50 );
+/** */
+define( "REPORTING_INTERVAL", 100 );
+#define( "REPORTING_INTERVAL", 1 );
-function refreshLinks( $start ) {
- global $wgUser, $wgTitle, $wgArticle, $wgEnablePersistentLC, $wgLinkCache, $wgOut;
+function refreshLinks( $start, $newOnly = false, $maxLag = false, $end = 0, $redirectsOnly = false, $oldRedirectsOnly = false ) {
+ global $wgUser, $wgParser, $wgUseTidy;
- $res = wfQuery("SELECT max(cur_id) as m FROM cur", DB_READ);
- $row = wfFetchObject( $res );
- $end = $row->m;
-
- print("Refreshing link table. Starting from cur_id $start of $end.\n");
+ $fname = 'refreshLinks';
+ $dbr = wfGetDB( DB_SLAVE );
+ $start = intval( $start );
# Don't generate TeX PNGs (lack of a sensible current directory causes errors anyway)
- $wgUser->setOption("math", 3);
+ $wgUser->setOption('math', MW_MATH_SOURCE);
+
+ # Don't generate extension images (e.g. Timeline)
+ $wgParser->clearTagHooks();
+
+ # Don't use HTML tidy
+ $wgUseTidy = false;
+
+ $what = $redirectsOnly ? "redirects" : "links";
+
+ if( $oldRedirectsOnly ) {
+ # This entire code path is cut-and-pasted from below. Hurrah.
+ $res = $dbr->query(
+ "SELECT page_id ".
+ "FROM page ".
+ "LEFT JOIN redirect ON page_id=rd_from ".
+ "WHERE page_is_redirect=1 AND rd_from IS NULL AND ".
+ ($end == 0 ? "page_id >= $start"
+ : "page_id BETWEEN $start AND $end"),
+ $fname
+ );
+ $num = $dbr->numRows( $res );
+ print "Refreshing $num old redirects from $start...\n";
- for ($id = $start; $id <= $end; $id++) {
- if ( !($id % REPORTING_INTERVAL) ) {
- print "$id\n";
+ while( $row = $dbr->fetchObject( $res ) ) {
+ if ( !( ++$i % REPORTING_INTERVAL ) ) {
+ print "$i\n";
+ wfWaitForSlaves( $maxLag );
+ }
+ fixRedirect( $row->page_id );
}
+ } elseif( $newOnly ) {
+ print "Refreshing $what from ";
+ $res = $dbr->select( 'page',
+ array( 'page_id' ),
+ array(
+ 'page_is_new' => 1,
+ "page_id >= $start" ),
+ $fname
+ );
+ $num = $dbr->numRows( $res );
+ print "$num new articles...\n";
- if ( !($id % PAUSE_INTERVAL) ) {
- sleep(1);
+ $i = 0;
+ while ( $row = $dbr->fetchObject( $res ) ) {
+ if ( !( ++$i % REPORTING_INTERVAL ) ) {
+ print "$i\n";
+ wfWaitForSlaves( $maxLag );
+ }
+ if($redirectsOnly)
+ fixRedirect( $row->page_id );
+ else
+ fixLinksFromArticle( $row->page_id );
}
-
- $wgTitle = Title::newFromID( $id );
- if ( is_null( $wgTitle ) ) {
- continue;
+ } else {
+ print "Refreshing $what table.\n";
+ if ( !$end ) {
+ $end = $dbr->selectField( 'page', 'max(page_id)', false );
}
-
- $wgArticle = new Article( $wgTitle );
- $text = $wgArticle->getContent( true );
- $wgLinkCache = new LinkCache;
- $wgOut->addWikiText( $text );
-
- if ( $wgEnablePersistentLC ) {
- $wgLinkCache->saveToLinkscc( $id, wfStrencode( $wgTitle->getPrefixedDBkey() ) );
+ print("Starting from page_id $start of $end.\n");
+
+ for ($id = $start; $id <= $end; $id++) {
+
+ if ( !($id % REPORTING_INTERVAL) ) {
+ print "$id\n";
+ wfWaitForSlaves( $maxLag );
+ }
+ if($redirectsOnly)
+ fixRedirect( $id );
+ else
+ fixLinksFromArticle( $id );
+ }
+ }
+}
+
+function fixRedirect( $id ){
+ global $wgTitle, $wgArticle;
+
+ $wgTitle = Title::newFromID( $id );
+ $dbw = wfGetDB( DB_MASTER );
+
+ if ( is_null( $wgTitle ) ) {
+ return;
+ }
+ $wgArticle = new Article($wgTitle);
+
+ $rt = $wgArticle->followRedirect();
+
+ if($rt == false || !is_object($rt))
+ return;
+
+ $wgArticle->updateRedirectOn($dbw,$rt);
+}
+
+function fixLinksFromArticle( $id ) {
+ global $wgTitle, $wgParser;
+
+ $wgTitle = Title::newFromID( $id );
+ $dbw = wfGetDB( DB_MASTER );
+
+ $linkCache =& LinkCache::singleton();
+ $linkCache->clear();
+
+ if ( is_null( $wgTitle ) ) {
+ return;
+ }
+ $dbw->begin();
+
+ $revision = Revision::newFromTitle( $wgTitle );
+ if ( !$revision ) {
+ return;
+ }
+
+ $options = new ParserOptions;
+ $parserOutput = $wgParser->parse( $revision->getText(), $wgTitle, $options, true, true, $revision->getId() );
+ $update = new LinksUpdate( $wgTitle, $parserOutput, false );
+ $update->doUpdate();
+ $dbw->immediateCommit();
+}
+
+function deleteLinksFromNonexistent( $maxLag = 0 ) {
+ $fname = 'deleteLinksFromNonexistent';
+
+ wfWaitForSlaves( $maxLag );
+
+ $dbw = wfGetDB( DB_MASTER );
+
+ $linksTables = array(
+ 'pagelinks' => 'pl_from',
+ 'imagelinks' => 'il_from',
+ 'categorylinks' => 'cl_from',
+ 'templatelinks' => 'tl_from',
+ 'externallinks' => 'el_from',
+ );
+
+ $page = $dbw->tableName( 'page' );
+
+
+ foreach ( $linksTables as $table => $field ) {
+ if ( !$dbw->ping() ) {
+ print "DB disconnected, reconnecting...";
+ while ( !$dbw->ping() ) {
+ print ".";
+ sleep(10);
+ }
+ print "\n";
}
- $linksUpdate = new LinksUpdate( $id, $wgTitle->getPrefixedDBkey() );
- $linksUpdate->doDumbUpdate();
- $linksUpdate->fixBrokenLinks();
+ $pTable = $dbw->tableName( $table );
+ $sql = "DELETE $pTable FROM $pTable LEFT JOIN $page ON page_id=$field WHERE page_id IS NULL";
+
+ print "Deleting $table from non-existent articles...";
+ $dbw->query( $sql, $fname );
+ print " fixed " .$dbw->affectedRows() . " row(s)\n";
}
}
+
?>
dépôts / lhc / web / wiklou.git / blobdiff