<?php
/**
- * @package MediaWiki
- * @subpackage Maintenance
+ * @file
+ * @ingroup Maintenance ExternalStorage
*/
-/** */
-require_once( 'Revision.php' );
-require_once( 'ExternalStoreDB.php' );
-
/** @todo document */
function compressOldPages( $start = 0, $extdb = '' ) {
$fname = 'compressOldPages';
$chunksize = 50;
print "Starting from old_id $start...\n";
- $dbw =& wfGetDB( DB_MASTER );
+ $dbw = wfGetDB( DB_MASTER );
do {
- $end = $start + $chunksize;
- $res = $dbw->select( 'text', array( 'old_id','old_flags','old_namespace','old_title','old_text' ),
+ $res = $dbw->select( 'text', array( 'old_id','old_flags','old_text' ),
"old_id>=$start", $fname, array( 'ORDER BY' => 'old_id', 'LIMIT' => $chunksize, 'FOR UPDATE' ) );
if( $dbw->numRows( $res ) == 0 ) {
break;
}
$last = $start;
- while( $row = $dbw->fetchObject( $res ) ) {
+ foreach ( $res as $row ) {
# print " {$row->old_id} - {$row->old_namespace}:{$row->old_title}\n";
compressPage( $row, $extdb );
$last = $row->old_id;
}
- $dbw->freeResult( $res );
$start = $last + 1; # Deletion may leave long empty stretches
print "$start...\n";
} while( true );
#print "Already compressed row {$row->old_id}\n";
return false;
}
- $dbw =& wfGetDB( DB_MASTER );
+ $dbw = wfGetDB( DB_MASTER );
$flags = $row->old_flags ? "{$row->old_flags},gzip" : "gzip";
$compress = gzdeflate( $row->old_text );
}
# Update text row
- $dbw->update( 'text',
+ $dbw->update( 'text',
array( /* SET */
'old_flags' => $flags,
'old_text' => $compress
), array( /* WHERE */
'old_id' => $row->old_id
- ), $fname, 'LIMIT 1'
+ ), $fname,
+ array( 'LIMIT' => 1 )
);
return true;
}
define( 'LS_CHUNKED', 1 );
/** @todo document */
-function compressWithConcat( $startId, $maxChunkSize, $maxChunkFactor, $factorThreshold, $beginDate, $endDate )
+function compressWithConcat( $startId, $maxChunkSize, $beginDate,
+ $endDate, $extdb="", $maxPageId = false )
{
$fname = 'compressWithConcat';
$loadStyle = LS_CHUNKED;
-
- $dbr =& wfGetDB( DB_SLAVE );
- $dbw =& wfGetDB( DB_MASTER );
+
+ $dbr = wfGetDB( DB_SLAVE );
+ $dbw = wfGetDB( DB_MASTER );
+
+ # Set up external storage
+ if ( $extdb != '' ) {
+ $storeObj = new ExternalStoreDB;
+ }
# Get all articles by page_id
- $maxPageId = $dbr->selectField( 'page', 'max(page_id)', '', $fname );
+ if ( !$maxPageId ) {
+ $maxPageId = $dbr->selectField( 'page', 'max(page_id)', '', $fname );
+ }
+ print "Starting from $startId of $maxPageId\n";
$pageConds = array();
/*
$pageConds[] = 'page_namespace<>0';
}
if ( $queryExtra ) {
- $pageConds[] = $queryExtra;
+ $pageConds[] = $queryExtra;
}
*/
# For each article, get a list of revisions which fit the criteria
+
# No recompression, use a condition on old_flags
+ # Don't compress object type entities, because that might produce data loss when
+ # overwriting bulk storage concat rows. Don't compress external references, because
+ # the script doesn't yet delete rows from external storage.
$conds = array(
- "old_flags NOT LIKE '%object%' " .
- " AND (old_flags NOT LIKE '%external%' OR old_text NOT LIKE 'DB://%/%/%')");
-
+ 'old_flags NOT ' . $dbr->buildLike( $dbr->anyString(), 'object', $dbr->anyString() ) . ' AND old_flags NOT '
+ . $dbr->buildLike( $dbr->anyString(), 'external', $dbr->anyString() ) );
+
if ( $beginDate ) {
+ if ( !preg_match( '/^\d{14}$/', $beginDate ) ) {
+ print "Invalid begin date \"$beginDate\"\n";
+ return false;
+ }
$conds[] = "rev_timestamp>'" . $beginDate . "'";
- }
+ }
if ( $endDate ) {
+ if ( !preg_match( '/^\d{14}$/', $endDate ) ) {
+ print "Invalid end date \"$endDate\"\n";
+ return false;
+ }
$conds[] = "rev_timestamp<'" . $endDate . "'";
}
if ( $loadStyle == LS_CHUNKED ) {
$revLoadOptions = array();
}
- $oldReadsSinceLastSlaveWait = 0; #check slave lag periodically
- $totalMatchingRevisions = 0;
- $masterPos = false;
+ # Don't work with current revisions
+ # Don't lock the page table for update either -- TS 2006-04-04
+ #$tables[] = 'page';
+ #$conds[] = 'page_id=rev_page AND rev_id != page_latest';
+
for ( $pageId = $startId; $pageId <= $maxPageId; $pageId++ ) {
- $pageRes = $dbr->select( 'page', array('page_id', 'page_namespace', 'page_title'),
+ wfWaitForSlaves( 5 );
+
+ # Wake up
+ $dbr->ping();
+
+ # Get the page row
+ $pageRes = $dbr->select( 'page',
+ array('page_id', 'page_namespace', 'page_title','page_latest'),
$pageConds + array('page_id' => $pageId), $fname );
if ( $dbr->numRows( $pageRes ) == 0 ) {
continue;
# Load revisions
$revRes = $dbw->select( $tables, $fields,
- array( 'rev_page' => $pageRow->page_id ) + $conds,
+ array_merge( array(
+ 'rev_page' => $pageRow->page_id,
+ # Don't operate on the current revision
+ # Use < instead of <> in case the current revision has changed
+ # since the page select, which wasn't locking
+ 'rev_id < ' . $pageRow->page_latest
+ ), $conds ),
$fname,
$revLoadOptions
);
$revs = array();
- while ( $revRow = $dbw->fetchObject( $revRes ) ) {
+ foreach ( $revRes as $revRow ) {
$revs[] = $revRow;
}
-
+
if ( count( $revs ) < 2) {
# No revisions matching, no further processing
print "\n";
$dbw->begin();
$usedChunk = false;
$primaryOldid = $revs[$i]->rev_text_id;
-
+
# Get the text of each revision and add it to the object
- for ( $j = 0; $j < $thisChunkSize && $chunk->isHappy( $maxChunkFactor, $factorThreshold ); $j++ ) {
+ for ( $j = 0; $j < $thisChunkSize && $chunk->isHappy(); $j++ ) {
$oldid = $revs[$i + $j]->rev_text_id;
-
+
# Get text
if ( $loadStyle == LS_INDIVIDUAL ) {
- $textRow = $dbw->selectRow( 'text',
+ $textRow = $dbw->selectRow( 'text',
array( 'old_flags', 'old_text' ),
array( 'old_id' => $oldid ),
$fname,
#$dbw->delete( 'old', array( 'old_id' => $oldid ) );
}
- if ( $j == 0 ) {
+ if ( $extdb == "" && $j == 0 ) {
$chunk->setText( $text );
print '.';
} else {
$stub = false;
print 'x';
} else {
- $stub = $chunk->addItem( $text );
+ $stub = new HistoryBlobStub( $chunk->addItem( $text ) );
$stub->setLocation( $primaryOldid );
- $hash = $stub->getHash();
- $stub = serialize( $stub );
+ $stub->setReferrer( $oldid );
print '.';
$usedChunk = true;
}
}
}
$thisChunkSize = $j;
-
+
# If we couldn't actually use any stubs because the pages were too small, do nothing
if ( $usedChunk ) {
- # Store the main object
- $dbw->update( 'text',
- array( /* SET */
- 'old_text' => serialize( $chunk ),
- 'old_flags' => 'object,utf-8',
- ), array( /* WHERE */
- 'old_id' => $primaryOldid
- )
- );
-
- # Store the stub objects
- for ( $j = 1; $j < $thisChunkSize; $j++ ) {
- # Skip if not compressing
- if ( $stubs[$j] !== false ) {
+ if ( $extdb != "" ) {
+ # Move blob objects to External Storage
+ $stored = $storeObj->store( $extdb, serialize( $chunk ));
+ if ($stored === false) {
+ print "Unable to store object\n";
+ return false;
+ }
+ # Store External Storage URLs instead of Stub placeholders
+ foreach ($stubs as $stub) {
+ if ($stub===false)
+ continue;
+ # $stored should provide base path to a BLOB
+ $url = $stored."/".$stub->getHash();
$dbw->update( 'text',
array( /* SET */
- 'old_text' => $stubs[$j],
- 'old_flags' => 'object,utf-8',
- ), array( /* WHERE */
- 'old_id' => $revs[$i + $j]->rev_text_id
+ 'old_text' => $url,
+ 'old_flags' => 'external,utf-8',
+ ), array ( /* WHERE */
+ 'old_id' => $stub->getReferrer(),
)
);
}
+ } else {
+ # Store the main object locally
+ $dbw->update( 'text',
+ array( /* SET */
+ 'old_text' => serialize( $chunk ),
+ 'old_flags' => 'object,utf-8',
+ ), array( /* WHERE */
+ 'old_id' => $primaryOldid
+ )
+ );
+
+ # Store the stub objects
+ for ( $j = 1; $j < $thisChunkSize; $j++ ) {
+ # Skip if not compressing and don't overwrite the first revision
+ if ( $stubs[$j] !== false && $revs[$i + $j]->rev_text_id != $primaryOldid ) {
+ $dbw->update( 'text',
+ array( /* SET */
+ 'old_text' => serialize($stubs[$j]),
+ 'old_flags' => 'object,utf-8',
+ ), array( /* WHERE */
+ 'old_id' => $revs[$i + $j]->rev_text_id
+ )
+ );
+ }
+ }
}
}
# Done, next
print "/";
$dbw->commit();
$i += $thisChunkSize;
+ wfWaitForSlaves( 5 );
}
print "\n";
}
return true;
}
-?>