| Index: trunk/extensions/FlaggedRevs/maintenance/pruneRevData.php |
| — | — | @@ -0,0 +1,37 @@ |
| | 2 | +<?php |
| | 3 | + |
| | 4 | +if ( getenv( 'MW_INSTALL_PATH' ) ) { |
| | 5 | + $IP = getenv( 'MW_INSTALL_PATH' ); |
| | 6 | +} else { |
| | 7 | + $IP = dirname(__FILE__).'/../../..'; |
| | 8 | +} |
| | 9 | + |
| | 10 | +$options = array( 'prune', 'help', 'start' ); |
| | 11 | +require "$IP/maintenance/commandLine.inc"; |
| | 12 | +require dirname(__FILE__) . '/pruneRevData.inc'; |
| | 13 | + |
| | 14 | +if( isset($options['help']) ) { |
| | 15 | + echo <<<TEXT |
| | 16 | +Purpose: |
| | 17 | + This script clears template/image data for reviewed versions |
| | 18 | + that are 1+ month old and have 50+ newer versions in page. By |
| | 19 | + default, it will just output how many rows can be deleted. Use |
| | 20 | + the 'prune' option to actually delete them. |
| | 21 | +Usage: |
| | 22 | + php pruneData.php --help |
| | 23 | + php pruneData.php [--prune --start <ID> ] |
| | 24 | + |
| | 25 | + --help : This help message |
| | 26 | + --prune : Actually do a live run |
| | 27 | + --<ID> : The ID of the starting rev |
| | 28 | + |
| | 29 | +TEXT; |
| | 30 | + exit(0); |
| | 31 | +} |
| | 32 | + |
| | 33 | +error_reporting( E_ALL ); |
| | 34 | + |
| | 35 | +$start = isset($options['start']) ? $options['start'] : null; |
| | 36 | +$prune = isset($options['prune']) ? true : null; |
| | 37 | + |
| | 38 | +prune_flaggedrevs($start,$prune); |
| Property changes on: trunk/extensions/FlaggedRevs/maintenance/pruneRevData.php |
| ___________________________________________________________________ |
| Name: svn:eol-style |
| 1 | 39 | + native |
| Index: trunk/extensions/FlaggedRevs/maintenance/pruneRevData.inc |
| — | — | @@ -0,0 +1,119 @@ |
| | 2 | +<?php
|
| | 3 | +
|
| | 4 | +function prune_flaggedrevs( $start = null, $prune = false ) {
|
| | 5 | + if( $prune ) {
|
| | 6 | + echo "Pruning old flagged revision inclusion data...\n";
|
| | 7 | + } else {
|
| | 8 | + echo "Running dry-run of old flagged revision inclusion data pruning...\n";
|
| | 9 | + }
|
| | 10 | +
|
| | 11 | + $BATCH_SIZE = 500;
|
| | 12 | +
|
| | 13 | + $db = wfGetDB( DB_MASTER );
|
| | 14 | +
|
| | 15 | + if( $start === null ) {
|
| | 16 | + $start = $db->selectField( 'flaggedpages', 'MIN(fp_page_id)', false, __FUNCTION__ );
|
| | 17 | + }
|
| | 18 | + $end = $db->selectField( 'flaggedpages', 'MAX(fp_page_id)', false, __FUNCTION__ );
|
| | 19 | + if( is_null( $start ) || is_null( $end ) ) {
|
| | 20 | + echo "...flaggedpages table seems to be empty.\n";
|
| | 21 | + return;
|
| | 22 | + }
|
| | 23 | + $end += $BATCH_SIZE - 1; # Do remaining chunk
|
| | 24 | + $blockStart = $start;
|
| | 25 | + $blockEnd = $start + $BATCH_SIZE - 1;
|
| | 26 | +
|
| | 27 | + $tDeleted = $fDeleted = 0; // tallies
|
| | 28 | +
|
| | 29 | + $newerRevs = 50;
|
| | 30 | + $cutoff = $db->timestamp( time() - 30*24*3600 );
|
| | 31 | + // DEV
|
| | 32 | + $newerRevs = 1;
|
| | 33 | + $cutoff = $db->timestamp( time() - 3600 );
|
| | 34 | + while( $blockEnd <= $end ) {
|
| | 35 | + echo "...doing fp_page_id from $blockStart to $blockEnd\n";
|
| | 36 | + $cond = "fp_page_id BETWEEN $blockStart AND $blockEnd";
|
| | 37 | + $res = $db->select( 'flaggedpages', 'fp_page_id', $cond, __FUNCTION__ );
|
| | 38 | + $batchCount = 0; // rows deleted without slave lag check
|
| | 39 | + // Go through a chunk of flagged pages...
|
| | 40 | + while( $row = $db->fetchObject($res) ) {
|
| | 41 | + // Get the newest X ($newerRevs) flagged revs for this page
|
| | 42 | + $sres = $db->select( 'flaggedrevs',
|
| | 43 | + 'fr_rev_id',
|
| | 44 | + array( 'fr_page_id' => $row->fp_page_id ),
|
| | 45 | + __METHOD__,
|
| | 46 | + array( 'ORDER BY' => 'fr_rev_id DESC', 'LIMIT' => $newerRevs )
|
| | 47 | + );
|
| | 48 | + // See if there are older revs that can be pruned...
|
| | 49 | + if( $db->numRows( $sres ) == $newerRevs ) {
|
| | 50 | + // Get the oldest of the top X revisions
|
| | 51 | + $sres->seek( $newerRevs - 1 );
|
| | 52 | + $lrow = $db->fetchObject( $sres );
|
| | 53 | + $oldestId = (int)$lrow->fr_rev_id; // oldest revision Id
|
| | 54 | + // Get revs not in the top X that were not reviewed recently
|
| | 55 | + $db->freeResult( $sres );
|
| | 56 | + $sres = $db->select( 'flaggedrevs',
|
| | 57 | + 'fr_rev_id',
|
| | 58 | + array(
|
| | 59 | + 'fr_page_id' => $row->fp_page_id,
|
| | 60 | + 'fr_rev_id < '.$oldestId, // not in the newest X
|
| | 61 | + 'fr_timestamp < '.$db->addQuotes( $cutoff ) // not reviewed recently
|
| | 62 | + ),
|
| | 63 | + __METHOD__,
|
| | 64 | + // Sanity check (start with the oldest)
|
| | 65 | + array( 'ORDER BY' => 'fr_rev_id ASC', 'LIMIT' => 5000 )
|
| | 66 | + );
|
| | 67 | + // Build an array of these rev Ids
|
| | 68 | + $revsClearIncludes = array();
|
| | 69 | + foreach( $sres as $srow ) {
|
| | 70 | + $revsClearIncludes[] = $srow->fr_rev_id;
|
| | 71 | + }
|
| | 72 | + $batchCount += count($revsClearIncludes); // # of revs to prune
|
| | 73 | + $db->freeResult( $sres );
|
| | 74 | + // Write run: clear the include data for these old revs
|
| | 75 | + if( $prune ) {
|
| | 76 | + $db->begin();
|
| | 77 | + $db->delete( 'flaggedtemplates',
|
| | 78 | + array('ft_rev_id' => $revsClearIncludes),
|
| | 79 | + __METHOD__
|
| | 80 | + );
|
| | 81 | + $tDeleted += $db->affectedRows();
|
| | 82 | + $db->delete( 'flaggedimages',
|
| | 83 | + array('fi_rev_id' => $revsClearIncludes),
|
| | 84 | + __METHOD__
|
| | 85 | + );
|
| | 86 | + $fDeleted += $db->affectedRows();
|
| | 87 | + $db->commit();
|
| | 88 | + // Dry run: say how many includes rows would have been cleared
|
| | 89 | + } else if( count($revsClearIncludes) ) {
|
| | 90 | + $tDeleted += $db->selectField( 'flaggedtemplates',
|
| | 91 | + 'COUNT(*)',
|
| | 92 | + array('ft_rev_id' => $revsClearIncludes),
|
| | 93 | + __METHOD__
|
| | 94 | + );
|
| | 95 | + $fDeleted += $db->selectField( 'flaggedimages',
|
| | 96 | + 'COUNT(*)',
|
| | 97 | + array('fi_rev_id' => $revsClearIncludes),
|
| | 98 | + __METHOD__
|
| | 99 | + );
|
| | 100 | + }
|
| | 101 | + // Check slave lag...
|
| | 102 | + if( $batchCount >= $BATCH_SIZE ) {
|
| | 103 | + $batchCount = 0;
|
| | 104 | + wfWaitForSlaves( 5 );
|
| | 105 | + }
|
| | 106 | + } else {
|
| | 107 | + $db->freeResult( $sres );
|
| | 108 | + }
|
| | 109 | + }
|
| | 110 | + $db->freeResult( $res );
|
| | 111 | + $blockStart += $BATCH_SIZE;
|
| | 112 | + $blockEnd += $BATCH_SIZE;
|
| | 113 | + }
|
| | 114 | + if( $prune ) {
|
| | 115 | + echo "...flagged revision inclusion prunning complete ...\n";
|
| | 116 | + } else {
|
| | 117 | + echo "...flagged revision inclusion prune test complete ...\n";
|
| | 118 | + }
|
| | 119 | + echo "Rows: \tflaggedtemplates:$tDeleted\t\tflaggedimages:$fDeleted\n";
|
| | 120 | +}
|