maintenance/storage/testCompression.php

   1 <?php
   2
   3 $optionsWithArgs = array( 'start', 'limit', 'type' );
   4 require( dirname(__FILE__).'/../commandLine.inc' );
   5
   6 if ( !isset( $args[0] )  ) {
   7         echo "Usage: php testCompression.php [--type=<type>] [--start=<start-date>] [--limit=<num-revs>] <page-title>\n";
   8         exit( 1 );
   9 }
  10
  11 $title = Title::newFromText( $args[0] );
  12 if ( isset( $options['start'] ) ) {
  13         $start = wfTimestamp( TS_MW, strtotime( $options['start'] ) );
  14         echo "Starting from " . $wgLang->timeanddate( $start ) . "\n";
  15 } else {
  16         $start = '19700101000000';
  17 }
  18 $limit = isset( $options['limit'] ) ? $options['limit'] : 10;
  19 $type = isset( $options['type'] ) ? $options['type'] : 'ConcatenatedGzipHistoryBlob';
  20
  21
  22 $dbr = wfGetDB( DB_SLAVE );
  23 $res = $dbr->select(
  24         array( 'page', 'revision', 'text' ),
  25         '*',
  26         array(
  27                 'page_namespace' => $title->getNamespace(),
  28                 'page_title' => $title->getDBkey(),
  29                 'page_id=rev_page',
  30                 'rev_timestamp > ' . $dbr->addQuotes( $dbr->timestamp( $start ) ),
  31                 'rev_text_id=old_id'
  32         ), __FILE__, array( 'LIMIT' => $limit )
  33 );
  34
  35 $blob = new $type;
  36 $hashes = array();
  37 $keys = array();
  38 $uncompressedSize = 0;
  39 $t = -microtime( true );
  40 foreach ( $res as $row ) {
  41         $revision = new Revision( $row );
  42         $text = $revision->getText();
  43         $uncompressedSize += strlen( $text );
  44         $hashes[$row->rev_id] = md5( $text );
  45         $keys[$row->rev_id] = $blob->addItem( $text );
  46 }
  47
  48 $serialized = serialize( $blob );
  49 $t += microtime( true );
  50
  51 printf( "Compression ratio for %d revisions: %5.2f, %s -> %s\n",
  52         $res->numRows(),
  53         $uncompressedSize / strlen( $serialized ),
  54         $wgLang->formatSize( $uncompressedSize ),
  55         $wgLang->formatSize( strlen( $serialized ) )
  56 );
  57 printf( "Compression time: %5.2f ms\n", $t * 1000 );
  58
  59 $t = -microtime( true );
  60 $blob = unserialize( $serialized );
  61 foreach ( $keys as $id => $key ) {
  62         $text = $blob->getItem( $key );
  63         if ( md5( $text ) != $hashes[$id] ) {
  64                 echo "Content hash mismatch for rev_id $id\n";
  65                 #var_dump( $text );
  66         }
  67 }
  68 $t += microtime( true );
  69 printf( "Decompression time: %5.2f ms\n", $t * 1000 );
  70