| Index: trunk/phase3/maintenance/refreshLinks.inc |
| — | — | @@ -146,12 +146,13 @@ |
| 147 | 147 | * @author Merlijn van Deen <valhallasw@arctus.nl> |
| 148 | 148 | */ |
| 149 | 149 | function deleteLinksFromNonexistent( $maxLag = 0, $batchSize = 100 ) { |
| 150 | | - $fname = 'deleteLinksFromNonexistent'; |
| 151 | 150 | wfWaitForSlaves( $maxLag ); |
| 152 | 151 | |
| 153 | 152 | $dbw = wfGetDB( DB_MASTER ); |
| 154 | | - $dbr = wfGetDB( DB_SLAVE ); |
| 155 | | - $dbr->bufferResults(false); |
| | 153 | + |
| | 154 | + $lb = wfGetLBFactory()->newMainLB(); |
| | 155 | + $dbr = $lb->getConnection( DB_SLAVE ); |
| | 156 | + $dbr->bufferResults( false ); |
| 156 | 157 | |
| 157 | 158 | $linksTables = array( // table name => page_id field |
| 158 | 159 | 'pagelinks' => 'pl_from', |
| — | — | @@ -161,65 +162,41 @@ |
| 162 | 163 | 'externallinks' => 'el_from', |
| 163 | 164 | ); |
| 164 | 165 | |
| 165 | | - $readPage = $dbr->tableName( 'page' ); |
| 166 | | - |
| 167 | 166 | foreach ( $linksTables as $table => $field ) { |
| 168 | | - $readLinks = $dbr->tableName( $table ); |
| 169 | | - |
| 170 | 167 | print "Retrieving illegal entries from $table... "; |
| 171 | 168 | |
| 172 | | - $sql = "SELECT DISTINCT( $field ) FROM $readLinks LEFT JOIN $readPage ON $field=page_id WHERE page_id IS NULL;"; |
| 173 | | - $results = $dbr->query( $sql, $fname . ':' . $readLinks ); |
| | 169 | + // SELECT DISTINCT( $field ) FROM $table LEFT JOIN page ON $field=page_id WHERE page_id IS NULL; |
| | 170 | + $results = $dbr->select( array( $table, 'page' ), |
| | 171 | + $field, |
| | 172 | + array('page_id' => null ), |
| | 173 | + __METHOD__, |
| | 174 | + 'DISTINCT', |
| | 175 | + array( 'page' => array( 'LEFT JOIN', "$field=page_id")) |
| | 176 | + ); |
| 174 | 177 | |
| 175 | | - print $results->numRows() . " illegal " . $field. "s. "; |
| | 178 | + $counter = 0; |
| | 179 | + $list = array(); |
| | 180 | + print "0.."; |
| 176 | 181 | |
| 177 | | - if ( $results->numRows() > 0 ) { |
| 178 | | - $counter = 0; |
| 179 | | - $list = array(); |
| 180 | | - print "Removing illegal links: 1.."; |
| 181 | | - |
| 182 | | - foreach( $results as $row ) { |
| 183 | | - $counter++; |
| 184 | | - $list[] = $row->$field; |
| 185 | | - if ( ( $counter % $batchSize ) == 0 ) { |
| 186 | | - print $counter . ".."; |
| 187 | | - deleteBatch($dbw, $table, $field, $list); |
| 188 | | - $list = array(); |
| 189 | | - } |
| | 182 | + foreach( $results as $row ) { |
| | 183 | + $counter++; |
| | 184 | + $list[] = $row->$field; |
| | 185 | + if ( ( $counter % $batchSize ) == 0 ) { |
| | 186 | + wfWaitForSlaves(5); |
| | 187 | + $dbw->delete( $table, array( $field => $list ), __METHOD__ ); |
| | 188 | + |
| | 189 | + print $counter . ".."; |
| | 190 | + $list = array(); |
| 190 | 191 | } |
| 191 | | - print $counter; |
| 192 | | - deleteBatch($dbw, $table, $field, $list); |
| 193 | 192 | } |
| 194 | 193 | |
| 195 | | - print "\n"; |
| 196 | | - } |
| 197 | | -} |
| 198 | | - |
| 199 | | -/* Deletes a batch of items from a table. |
| 200 | | - * Runs the query: DELETE FROM <$table> WHERE <$field> IN (<$list>) |
| 201 | | - * |
| 202 | | - * @param $dbw Database Database object to run the DELETE query on |
| 203 | | - * @param $table table to work on; will be converted via $dbw->tableName. |
| 204 | | - * @param $field column to search in |
| 205 | | - * @param $list values to remove. Array with SQL-safe (!) values. |
| 206 | | - * |
| 207 | | - * @author Merlijn van Deen <valhallasw@arctus.nl> |
| 208 | | - */ |
| 209 | | -function deleteBatch($dbw, $table, $field, $list) { |
| 210 | | - if (count($list) == 0) return; |
| 211 | | - |
| 212 | | - $masterLinks = $dbw->tableName( $table ); |
| 213 | | - $fname = "deleteBatch:masterLinks"; |
| 214 | | - |
| 215 | | - if ( !$dbw->ping() ) { |
| 216 | | - print "\nDB disconnected, reconnecting..."; |
| 217 | | - while ( !$dbw->ping() ) { |
| 218 | | - print "."; |
| 219 | | - sleep(10); |
| | 194 | + print $counter; |
| | 195 | + if (count($list) > 0) { |
| | 196 | + $dbw->delete( $table, array( $field => $list ), __METHOD__ ); |
| 220 | 197 | } |
| | 198 | + |
| 221 | 199 | print "\n"; |
| 222 | 200 | } |
| 223 | | - |
| 224 | | - $sql = "DELETE FROM $masterLinks WHERE $field IN (" . join("," , $list) . ");"; |
| 225 | | - $dbw->query($sql, $fname); |
| | 201 | + |
| | 202 | + $lb->closeAll(); |
| 226 | 203 | } |
| Index: trunk/phase3/maintenance/refreshLinks.php |
| — | — | @@ -45,6 +45,10 @@ |
| 46 | 46 | refreshLinks( $start, $options['new-only'], $options['m'], $options['e'], $options['redirects-only'], $options['old-redirects-only'] ); |
| 47 | 47 | } |
| 48 | 48 | |
| | 49 | +if ( !isset( $options['batch-size'] ) ) { |
| | 50 | + $options['batch-size'] = 100; |
| | 51 | +} |
| | 52 | + |
| 49 | 53 | deleteLinksFromNonexistent($options['m'], $options['batch-size']); |
| 50 | 54 | |
| 51 | 55 | if ( $options['globals'] ) { |