require_once( dirname( __FILE__ ) . '/Maintenance.php' );
class UpdateCollation extends Maintenance {
- const BATCH_SIZE = 1000;
+ const BATCH_SIZE = 50; // Number of rows to process in one batch
+ const SYNC_INTERVAL = 20; // Wait for slaves after this many batches
+
+ var $sizeHistogram = array();
public function __construct() {
parent::__construct();
up-to-date, it will do nothing.
TEXT;
- $this->addOption( 'force', 'Run on all rows, even if the collation is ' .
+ $this->addOption( 'force', 'Run on all rows, even if the collation is ' .
'supposed to be up-to-date.' );
- }
-
- public function syncDBs() {
- $lb = wfGetLB();
- $dbw = $lb->getConnection( DB_MASTER );
- $pos = $dbw->getMasterPos();
- $lb->waitForAll( $pos );
+ $this->addOption( 'previous-collation', 'Set the previous value of ' .
+ '$wgCategoryCollation here to speed up this script, especially if your ' .
+ 'categorylinks table is large. This will only update rows with that ' .
+ 'collation, though, so it may miss out-of-date rows with a different, ' .
+ 'even older collation.', false, true );
+ $this->addOption( 'target-collation', 'Set this to the new collation type to ' .
+ 'use instead of $wgCategoryCollation. Usually you should not use this, ' .
+ 'you should just update $wgCategoryCollation in LocalSettings.php.',
+ false, true );
+ $this->addOption( 'dry-run', 'Don\'t actually change the collations, just ' .
+ 'compile statistics.' );
+ $this->addOption( 'verbose-stats', 'Show more statistics.' );
}
public function execute() {
global $wgCategoryCollation, $wgMiserMode;
- $dbw = wfGetDB( DB_MASTER );
+ $dbw = $this->getDB( DB_MASTER );
$force = $this->getOption( 'force' );
+ $dryRun = $this->getOption( 'dry-run' );
+ $verboseStats = $this->getOption( 'verbose-stats' );
+ if ( $this->hasOption( 'target-collation' ) ) {
+ $collationName = $this->getOption( 'target-collation' );
+ $collation = Collation::factory( $collationName );
+ } else {
+ $collationName = $wgCategoryCollation;
+ $collation = Collation::singleton();
+ }
- $options = array( 'LIMIT' => self::BATCH_SIZE );
+ $options = array( 'LIMIT' => self::BATCH_SIZE, 'STRAIGHT_JOIN' );
- if ( $force ) {
+ if ( $force || $dryRun ) {
$options['ORDER BY'] = 'cl_from, cl_to';
+ $collationConds = array();
} else {
- $collationConds = array( 0 =>
- 'cl_collation != ' . $dbw->addQuotes( $wgCategoryCollation ) );
+ if ( $this->hasOption( 'previous-collation' ) ) {
+ $collationConds['cl_collation'] = $this->getOption( 'previous-collation' );
+ } else {
+ $collationConds = array( 0 =>
+ 'cl_collation != ' . $dbw->addQuotes( $collationName )
+ );
+ }
if ( !$wgMiserMode ) {
$count = $dbw->selectField(
$collationConds,
__METHOD__
);
-
- if ( $count == 0 ) {
- $this->output( "Collations up-to-date.\n" );
- return;
- }
- $this->output( "Fixing collation for $count rows.\n" );
+ } else {
+ $count = $dbw->estimateRowCount(
+ 'categorylinks',
+ '*',
+ $collationConds,
+ __METHOD__
+ );
+ }
+ if ( $count == 0 ) {
+ $this->output( "Collations up-to-date.\n" );
+ return;
}
+ $this->output( "Fixing collation for $count rows.\n" );
}
$count = 0;
- $row = false;
+ $batchCount = 0;
$batchConds = array();
do {
- $this->output( 'Processing next ' . self::BATCH_SIZE . ' rows... ');
+ $this->output( "Selecting next " . self::BATCH_SIZE . " rows..." );
$res = $dbw->select(
array( 'categorylinks', 'page' ),
array( 'cl_from', 'cl_to', 'cl_sortkey_prefix', 'cl_collation',
__METHOD__,
$options
);
+ $this->output( " processing..." );
- $dbw->begin();
+ if ( !$dryRun ) {
+ $dbw->begin( __METHOD__ );
+ }
foreach ( $res as $row ) {
$title = Title::newFromRow( $row );
if ( !$row->cl_collation ) {
# This is an old-style row, so the sortkey needs to be
# converted.
if ( $row->cl_sortkey == $title->getText()
- || $row->cl_sortkey == $title->getPrefixedText() ) {
+ || $row->cl_sortkey == $title->getPrefixedText() ) {
$prefix = '';
} else {
# Custom sortkey, use it as a prefix
} else {
$type = 'page';
}
- $dbw->update(
- 'categorylinks',
- array(
- 'cl_sortkey' => Collation::singleton()->getSortKey(
- $title->getCategorySortkey( $prefix ) ),
- 'cl_sortkey_prefix' => $prefix,
- 'cl_collation' => $wgCategoryCollation,
- 'cl_type' => $type,
- 'cl_timestamp = cl_timestamp',
- ),
- array( 'cl_from' => $row->cl_from, 'cl_to' => $row->cl_to ),
- __METHOD__
- );
+ $newSortKey = $collation->getSortKey(
+ $title->getCategorySortkey( $prefix ) );
+ if ( $verboseStats ) {
+ $this->updateSortKeySizeHistogram( $newSortKey );
+ }
+
+ if ( !$dryRun ) {
+ $dbw->update(
+ 'categorylinks',
+ array(
+ 'cl_sortkey' => $newSortKey,
+ 'cl_sortkey_prefix' => $prefix,
+ 'cl_collation' => $collationName,
+ 'cl_type' => $type,
+ 'cl_timestamp = cl_timestamp',
+ ),
+ array( 'cl_from' => $row->cl_from, 'cl_to' => $row->cl_to ),
+ __METHOD__
+ );
+ }
+ }
+ if ( !$dryRun ) {
+ $dbw->commit( __METHOD__ );
}
- $dbw->commit();
- if ( $force && $row ) {
+ if ( ( $force || $dryRun ) && $row ) {
$encFrom = $dbw->addQuotes( $row->cl_from );
$encTo = $dbw->addQuotes( $row->cl_to );
- $batchConds = array(
+ $batchConds = array(
"(cl_from = $encFrom AND cl_to > $encTo) " .
" OR cl_from > $encFrom" );
}
$count += $res->numRows();
$this->output( "$count done.\n" );
-
- $this->syncDBs();
+
+ if ( !$dryRun && ++$batchCount % self::SYNC_INTERVAL == 0 ) {
+ $this->output( "Waiting for slaves ... " );
+ wfWaitForSlaves();
+ $this->output( "done\n" );
+ }
} while ( $res->numRows() == self::BATCH_SIZE );
+
+ $this->output( "$count rows processed\n" );
+
+ if ( $verboseStats ) {
+ $this->output( "\n" );
+ $this->showSortKeySizeHistogram();
+ }
+ }
+
+ function updateSortKeySizeHistogram( $key ) {
+ $length = strlen( $key );
+ if ( !isset( $this->sizeHistogram[$length] ) ) {
+ $this->sizeHistogram[$length] = 0;
+ }
+ $this->sizeHistogram[$length]++;
+ }
+
+ function showSortKeySizeHistogram() {
+ $maxLength = max( array_keys( $this->sizeHistogram ) );
+ if ( $maxLength == 0 ) {
+ return;
+ }
+ $numBins = 20;
+ $coarseHistogram = array_fill( 0, $numBins, 0 );
+ $coarseBoundaries = array();
+ $boundary = 0;
+ for ( $i = 0; $i < $numBins - 1; $i++ ) {
+ $boundary += $maxLength / $numBins;
+ $coarseBoundaries[$i] = round( $boundary );
+ }
+ $coarseBoundaries[$numBins - 1] = $maxLength + 1;
+ $raw = '';
+ for ( $i = 0; $i <= $maxLength; $i++ ) {
+ if ( $raw !== '' ) {
+ $raw .= ', ';
+ }
+ if ( !isset( $this->sizeHistogram[$i] ) ) {
+ $val = 0;
+ } else {
+ $val = $this->sizeHistogram[$i];
+ }
+ for ( $coarseIndex = 0; $coarseIndex < $numBins - 1; $coarseIndex++ ) {
+ if ( $coarseBoundaries[$coarseIndex] > $i ) {
+ $coarseHistogram[$coarseIndex] += $val;
+ break;
+ }
+ }
+ if ( $coarseIndex == $numBins - 1 ) {
+ $coarseHistogram[$coarseIndex] += $val;
+ }
+ $raw .= $val;
+ }
+
+ $this->output( "Sort key size histogram\nRaw data: $raw\n\n" );
+
+ $maxBinVal = max( $coarseHistogram );
+ $scale = 60 / $maxBinVal;
+ $prevBoundary = 0;
+ for ( $coarseIndex = 0; $coarseIndex < $numBins; $coarseIndex++ ) {
+ if ( !isset( $coarseHistogram[$coarseIndex] ) ) {
+ $val = 0;
+ } else {
+ $val = $coarseHistogram[$coarseIndex];
+ }
+ $boundary = $coarseBoundaries[$coarseIndex];
+ $this->output( sprintf( "%-10s %-10d |%s\n",
+ $prevBoundary . '-' . ( $boundary - 1 ) . ': ',
+ $val,
+ str_repeat( '*', $scale * $val ) ) );
+ $prevBoundary = $boundary;
+ }
}
}
dépôts / lhc / web / wiklou.git / blobdiff