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 if ( isset( $options['limit'] ) ) {
  19         $limit = $options['limit'];
  20         $untilHappy = false;
  21 } else {
  22         $limit = 1000;
  23         $untilHappy = true;
  24 }
  25 $type = isset( $options['type'] ) ? $options['type'] : 'ConcatenatedGzipHistoryBlob';
  26
  27
  28 $dbr = wfGetDB( DB_SLAVE );
  29 $res = $dbr->select(
  30         array( 'page', 'revision', 'text' ),
  31         '*',
  32         array(
  33                 'page_namespace' => $title->getNamespace(),
  34                 'page_title' => $title->getDBkey(),
  35                 'page_id=rev_page',
  36                 'rev_timestamp > ' . $dbr->addQuotes( $dbr->timestamp( $start ) ),
  37                 'rev_text_id=old_id'
  38         ), __FILE__, array( 'LIMIT' => $limit )
  39 );
  40
  41 $blob = new $type;
  42 $hashes = array();
  43 $keys = array();
  44 $uncompressedSize = 0;
  45 $t = -microtime( true );
  46 foreach ( $res as $row ) {
  47         $revision = new Revision( $row );
  48         $text = $revision->getText();
  49         $uncompressedSize += strlen( $text );
  50         $hashes[$row->rev_id] = md5( $text );
  51         $keys[$row->rev_id] = $blob->addItem( $text );
  52         if ( $untilHappy && !$blob->isHappy() ) {
  53                 break;
  54         }
  55 }
  56
  57 $serialized = serialize( $blob );
  58 $t += microtime( true );
  59 # print_r( $blob->mDiffMap );
  60
  61 printf( "%s\nCompression ratio for %d revisions: %5.2f, %s -> %d\n",
  62         $type,
  63         count( $hashes ),
  64         $uncompressedSize / strlen( $serialized ),
  65         $wgLang->formatSize( $uncompressedSize ),
  66         strlen( $serialized )
  67 );
  68 printf( "Compression time: %5.2f ms\n", $t * 1000 );
  69
  70 $t = -microtime( true );
  71 $blob = unserialize( $serialized );
  72 foreach ( $keys as $id => $key ) {
  73         $text = $blob->getItem( $key );
  74         if ( md5( $text ) != $hashes[$id] ) {
  75                 echo "Content hash mismatch for rev_id $id\n";
  76                 # var_dump( $text );
  77         }
  78 }
  79 $t += microtime( true );
  80 printf( "Decompression time: %5.2f ms\n", $t * 1000 );
  81