Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
0.00% |
0 / 123 |
|
0.00% |
0 / 3 |
CRAP | |
0.00% |
0 / 1 |
| PruneFRIncludeData | |
0.00% |
0 / 117 |
|
0.00% |
0 / 3 |
342 | |
0.00% |
0 / 1 |
| __construct | |
0.00% |
0 / 23 |
|
0.00% |
0 / 1 |
2 | |||
| execute | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
2 | |||
| pruneFlaggedRevs | |
0.00% |
0 / 91 |
|
0.00% |
0 / 1 |
272 | |||
| 1 | <?php |
| 2 | /** |
| 3 | * @ingroup Maintenance |
| 4 | */ |
| 5 | |
| 6 | if ( getenv( 'MW_INSTALL_PATH' ) ) { |
| 7 | $IP = getenv( 'MW_INSTALL_PATH' ); |
| 8 | } else { |
| 9 | $IP = __DIR__ . '/../../..'; |
| 10 | } |
| 11 | |
| 12 | require_once "$IP/maintenance/Maintenance.php"; |
| 13 | |
| 14 | class PruneFRIncludeData extends Maintenance { |
| 15 | |
| 16 | public function __construct() { |
| 17 | parent::__construct(); |
| 18 | $this->addDescription( "This script clears template data for reviewed versions" . |
| 19 | "that are 1+ hour old and have 50+ newer versions in page. By default," . |
| 20 | "it will just output how many rows can be deleted. Use the 'prune' option " . |
| 21 | "to actually delete them." |
| 22 | ); |
| 23 | $this->addOption( 'prune', 'Actually do a live run' ); |
| 24 | $this->addOption( 'start', 'The ID of the starting rev', false, true ); |
| 25 | $this->addOption( 'sleep', 'Extra sleep time between each batch', false, true ); |
| 26 | $this->setBatchSize( 500 ); |
| 27 | $this->addOption( |
| 28 | 'rev-age', |
| 29 | 'Revisions older than this age will be deleted in seconds (default: 3600)', |
| 30 | false, |
| 31 | true |
| 32 | ); |
| 33 | $this->addOption( |
| 34 | 'rev-num', |
| 35 | 'Revisions must have at least this number of reviewed revisions on top (default: 50)', |
| 36 | false, |
| 37 | true |
| 38 | ); |
| 39 | $this->requireExtension( 'FlaggedRevs' ); |
| 40 | } |
| 41 | |
| 42 | /** |
| 43 | * @inheritDoc |
| 44 | */ |
| 45 | public function execute() { |
| 46 | $start = $this->getOption( 'start' ); |
| 47 | $prune = $this->getOption( 'prune' ); |
| 48 | $this->pruneFlaggedRevs( $start, $prune ); |
| 49 | } |
| 50 | |
| 51 | /** |
| 52 | * @param int|null $start Revision ID |
| 53 | * @param bool $prune |
| 54 | */ |
| 55 | private function pruneFlaggedRevs( $start = null, $prune = false ) { |
| 56 | if ( $prune ) { |
| 57 | $this->output( "Pruning old flagged revision inclusion data...\n" ); |
| 58 | } else { |
| 59 | $this->output( "Running dry-run of old flagged revision inclusion data pruning...\n" ); |
| 60 | } |
| 61 | |
| 62 | $dbw = wfGetDB( DB_PRIMARY ); |
| 63 | $dbr = wfGetDB( DB_REPLICA ); |
| 64 | |
| 65 | if ( $start === null ) { |
| 66 | $start = $dbr->selectField( 'flaggedpages', 'MIN(fp_page_id)', false, __METHOD__ ); |
| 67 | } |
| 68 | $end = $dbr->selectField( 'flaggedpages', 'MAX(fp_page_id)', false, __METHOD__ ); |
| 69 | if ( $start === null || $end === null ) { |
| 70 | $this->output( "...flaggedpages table seems to be empty.\n" ); |
| 71 | return; |
| 72 | } |
| 73 | $end += $this->mBatchSize - 1; # Do remaining chunk |
| 74 | $blockStart = (int)$start; |
| 75 | $blockEnd = (int)$start + $this->mBatchSize - 1; |
| 76 | |
| 77 | // Tally |
| 78 | $tDeleted = 0; |
| 79 | |
| 80 | $newerRevs = (int)$this->getOption( 'rev-num', 50 ); |
| 81 | $sleep = (int)$this->getOption( 'sleep', 0 ); |
| 82 | $cutoff = $dbr->timestamp( time() - (int)$this->getOption( 'rev-age', 3600 ) ); |
| 83 | |
| 84 | // First clean up revisions that don't exist anymore. |
| 85 | while ( true ) { |
| 86 | if ( !$prune ) { |
| 87 | break; |
| 88 | } |
| 89 | $sres = $dbr->selectFieldValues( |
| 90 | [ 'flaggedtemplates', 'revision' ], |
| 91 | 'ft_rev_id', |
| 92 | [ |
| 93 | 'rev_id' => null, |
| 94 | ], |
| 95 | __METHOD__, |
| 96 | [ 'LIMIT' => $this->mBatchSize ], |
| 97 | [ 'revision' => [ 'LEFT JOIN', 'rev_id = ft_rev_id' ] ] |
| 98 | ); |
| 99 | |
| 100 | if ( !$sres ) { |
| 101 | break; |
| 102 | } |
| 103 | $dbw->delete( 'flaggedtemplates', |
| 104 | [ 'ft_rev_id' => $sres ], |
| 105 | __METHOD__ |
| 106 | ); |
| 107 | $rowsCount = $dbw->affectedRows(); |
| 108 | $this->output( "...deleted $rowsCount rows\n" ); |
| 109 | $tDeleted += $rowsCount; |
| 110 | $this->waitForReplication(); |
| 111 | sleep( $sleep ); |
| 112 | } |
| 113 | |
| 114 | while ( $blockEnd <= $end ) { |
| 115 | $this->output( "...doing fp_page_id from $blockStart to $blockEnd\n" ); |
| 116 | $cond = "fp_page_id BETWEEN $blockStart AND $blockEnd"; |
| 117 | $res = $dbr->select( 'flaggedpages', 'fp_page_id', $cond, __METHOD__ ); |
| 118 | // Go through a chunk of flagged pages... |
| 119 | $revsClearIncludes = []; |
| 120 | foreach ( $res as $row ) { |
| 121 | // Get the newest X ($newerRevs) flagged revs for this page |
| 122 | $sres = $dbr->selectFieldValues( 'flaggedrevs', |
| 123 | 'fr_rev_id', |
| 124 | [ 'fr_page_id' => $row->fp_page_id ], |
| 125 | __METHOD__, |
| 126 | [ 'ORDER BY' => 'fr_rev_id DESC', 'LIMIT' => $newerRevs ] |
| 127 | ); |
| 128 | // See if there are older revs that can be pruned... |
| 129 | if ( count( $sres ) !== $newerRevs ) { |
| 130 | continue; |
| 131 | } |
| 132 | // Get the oldest of the top X revisions |
| 133 | $oldestId = (int)$sres[ $newerRevs - 1 ]; |
| 134 | // Get revs not in the top X that were not reviewed recently |
| 135 | $sres = $dbr->select( 'flaggedrevs', |
| 136 | 'fr_rev_id', |
| 137 | [ |
| 138 | 'fr_page_id' => $row->fp_page_id, |
| 139 | 'fr_rev_id < ' . $oldestId, // not in the newest X |
| 140 | 'fr_timestamp < ' . $dbr->addQuotes( $cutoff ) // not reviewed recently |
| 141 | ], |
| 142 | __METHOD__, |
| 143 | [ 'ORDER BY' => 'fr_rev_id ASC', 'LIMIT' => 5000 ] |
| 144 | ); |
| 145 | // Build an array of these rev Ids |
| 146 | foreach ( $sres as $srow ) { |
| 147 | $revsClearIncludes[] = $srow->fr_rev_id; |
| 148 | } |
| 149 | } |
| 150 | $blockStart += $this->mBatchSize; |
| 151 | $blockEnd += $this->mBatchSize; |
| 152 | if ( !$revsClearIncludes ) { |
| 153 | continue; |
| 154 | } elseif ( $prune ) { |
| 155 | foreach ( array_chunk( $revsClearIncludes, $this->mBatchSize ) as $batch ) { |
| 156 | $dbw->delete( 'flaggedtemplates', |
| 157 | [ 'ft_rev_id' => $batch ], |
| 158 | __METHOD__ |
| 159 | ); |
| 160 | $tDeleted += $dbw->affectedRows(); |
| 161 | $this->waitForReplication(); |
| 162 | sleep( $sleep ); |
| 163 | } |
| 164 | } else { |
| 165 | // Dry run: say how many includes rows would have been cleared |
| 166 | $tDeleted += $dbr->selectField( 'flaggedtemplates', |
| 167 | 'COUNT(*)', |
| 168 | [ 'ft_rev_id' => $revsClearIncludes ], |
| 169 | __METHOD__ |
| 170 | ); |
| 171 | } |
| 172 | } |
| 173 | if ( $prune ) { |
| 174 | $this->output( "Flagged revision inclusion prunning complete ...\n" ); |
| 175 | } else { |
| 176 | $this->output( "Flagged revision inclusion prune test complete ...\n" ); |
| 177 | } |
| 178 | $this->output( "Rows: \tflaggedtemplates:$tDeleted\n" ); |
| 179 | } |
| 180 | } |
| 181 | |
| 182 | $maintClass = PruneFRIncludeData::class; |
| 183 | require_once RUN_MAINTENANCE_IF_MAIN; |