X-Git-Url: https://git.heureux-cyclage.org/?a=blobdiff_plain;f=maintenance%2FpopulateParentId.php;h=01730033426373058ae56c8da46112ed3268e154;hb=674f6f9bd50868d57b639d0c9627914a654997ce;hp=863d7edae6369da8d25bc498d086c3ca1395a402;hpb=41ea588beff62d9f900a1950f105ae4cb474956e;p=lhc%2Fweb%2Fwiklou.git diff --git a/maintenance/populateParentId.php b/maintenance/populateParentId.php index 863d7edae6..0173003342 100644 --- a/maintenance/populateParentId.php +++ b/maintenance/populateParentId.php @@ -6,9 +6,8 @@ * and to find new page edits by users. */ -define( 'BATCH_SIZE', 200 ); - require_once 'commandLine.inc'; +require_once 'populateParentId.inc'; $db =& wfGetDB( DB_MASTER ); if ( !$db->tableExists( 'revision' ) ) { @@ -17,76 +16,3 @@ if ( !$db->tableExists( 'revision' ) ) { } populate_rev_parent_id( $db ); - -function populate_rev_parent_id( $db ) { - echo "Populating rev_parent_id column\n"; - $start = $db->selectField( 'revision', 'MIN(rev_id)', false, __FUNCTION__ ); - $end = $db->selectField( 'revision', 'MAX(rev_id)', false, __FUNCTION__ ); - # Do remaining chunk - $end += BATCH_SIZE - 1; - $blockStart = $start; - $blockEnd = $start + BATCH_SIZE - 1; - $count = 0; - $changed = 0; - while( $blockEnd <= $end ) { - echo "...doing rev_id from $blockStart to $blockEnd\n"; - $cond = "rev_id BETWEEN $blockStart AND $blockEnd"; - $res = $db->select( 'revision', - array('rev_id','rev_page','rev_timestamp','rev_parent_id'), - $cond, __FUNCTION__ ); - # Go through and update rev_parent_id from these rows. - # Assume that the previous revision of the title was - # the original previous revision of the title when the - # edit was made... - foreach( $res as $row ) { - # First, check rows with the same timestamp other than this one - # with a smaller rev ID. The highest ID "wins". This avoids loops - # as either id or timestamp need to be going from parent_id to parent_id - $previousID = $db->selectField( 'revision', 'rev_id', - array( 'rev_page' => $row->rev_page, 'rev_timestamp' => $row->rev_timestamp, - "rev_id < {$row->rev_id}" ), - __FUNCTION__, - array( 'ORDER BY' => 'rev_id DESC' ) ); - # If there are none, check the the highest ID with a lower timestamp - if( !$previousID ) { - # Get the highest older timestamp - $lastTimestamp = $db->selectField( 'revision', 'rev_timestamp', - array( 'rev_page' => $row->rev_page, "rev_timestamp < '{$row->rev_timestamp}'" ), - __FUNCTION__, - array( 'ORDER BY' => 'rev_timestamp DESC' ) ); - # If there is one, let the highest rev ID win - if( $lastTimestamp ) { - $previousID = $db->selectField( 'revision', 'rev_id', - array( 'rev_page' => $row->rev_page, 'rev_timestamp' => $lastTimestamp ), - __FUNCTION__, - array( 'ORDER BY' => 'rev_id DESC' ) ); - } - } - $previousID = intval($previousID); - if( $previousID != $row->rev_parent_id ) - $changed++; - # Update the row... - $db->update( 'revision', - array( 'rev_parent_id' => $previousID ), - array( 'rev_id' => $row->rev_id ), - __FUNCTION__ ); - $count++; - } - $blockStart += BATCH_SIZE - 1; - $blockEnd += BATCH_SIZE - 1; - wfWaitForSlaves( 5 ); - } - $logged = $db->insert( 'updatelog', - array( 'ul_key' => 'populate rev_parent_id' ), - __FUNCTION__, - 'IGNORE' ); - if( $logged ) { - echo "rev_parent_id population complete ... {$count} rows [{$changed} changed]\n"; - return true; - } else { - echo "Could not insert rev_parent_id population row.\n"; - return false; - } -} - -