diff options
author | Pierre Schmitz <pierre@archlinux.de> | 2008-08-15 01:29:47 +0200 |
---|---|---|
committer | Pierre Schmitz <pierre@archlinux.de> | 2008-08-15 01:29:47 +0200 |
commit | 370e83bb0dfd0c70de268c93bf07ad5ee0897192 (patch) | |
tree | 491674f4c242e4d6ba0d04eafa305174c35a3391 /maintenance/populateParentId.inc | |
parent | f4debf0f12d0524d2b2427c55ea3f16b680fad97 (diff) |
Update auf 1.13.0
Diffstat (limited to 'maintenance/populateParentId.inc')
-rw-r--r-- | maintenance/populateParentId.inc | 83 |
1 files changed, 83 insertions, 0 deletions
diff --git a/maintenance/populateParentId.inc b/maintenance/populateParentId.inc new file mode 100644 index 00000000..3fecc63b --- /dev/null +++ b/maintenance/populateParentId.inc @@ -0,0 +1,83 @@ +<?php + +define( 'BATCH_SIZE', 200 ); + +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__ ); + if( is_null( $start ) || is_null( $end ) ){ + echo "...revision table seems to be empty.\n"; + $db->insert( 'updatelog', + array( 'ul_key' => 'populate rev_parent_id' ), + __FUNCTION__, + 'IGNORE' ); + return; + } + # 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 timestamp can only decrease and never loops with IDs (from parent to parent) + $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; + } +} + |