maintenance/compressOld.inc

   1 <?php
   2 /**
   3  * @package MediaWiki
   4  * @subpackage Maintenance
   5  */
   6
   7 /** */
   8 require_once( 'Revision.php' );
   9
  10 /** @todo document */
  11 function compressOldPages( $start = 0 ) {
  12         $fname = 'compressOldPages';
  13
  14         $chunksize = 50;
  15         print "Starting from old_id $start...\n";
  16         $dbw =& wfGetDB( DB_MASTER );
  17         $old = $dbw->tableName( 'old' );
  18         do {
  19                 $end = $start + $chunksize;
  20                 $res = $dbw->select( 'old', array( 'old_id','old_flags','old_namespace','old_title','old_text' ),
  21                         "old_id>=$start", $fname, array( 'ORDER BY' => 'old_id', 'LIMIT' => $chunksize, 'FOR UPDATE' ) );
  22                 if( $dbw->numRows( $res ) == 0 ) {
  23                         break;
  24                 }
  25                 $last = $start;
  26                 while( $row = $dbw->fetchObject( $res ) ) {
  27                         # print "  {$row->old_id} - {$row->old_namespace}:{$row->old_title}\n";
  28                         compressPage( $row );
  29                         $last = $row->old_id;
  30                 }
  31                 $dbw->freeResult( $res );
  32                 $start = $last + 1; # Deletion may leave long empty stretches
  33                 print "$start...\n";
  34         } while( true );
  35 }
  36
  37 /** @todo document */
  38 function compressPage( $row ) {
  39         $fname = 'compressPage';
  40         if( false !== strpos( $row->old_flags, "gzip" ) ) {
  41                 print "Already compressed row {$row->old_id}?\n";
  42                 return false;
  43         }
  44         $dbw =& wfGetDB( DB_MASTER );
  45         $flags = $row->old_flags ? "{$row->old_flags},gzip" : "gzip";
  46         $compress = gzdeflate( $row->old_text );
  47         $dbw->update( 'old',
  48                 array( /* SET */
  49                         'old_flags' => $flags,
  50                         'old_text' => $compress
  51                 ), array( /* WHERE */
  52                         'old_id' => $row->old_id
  53                 ), $fname, 'LIMIT 1'
  54         );
  55         return true;
  56 }
  57
  58 define( 'LS_INDIVIDUAL', 0 );
  59 define( 'LS_CHUNKED', 1 );
  60
  61 /** @todo document */
  62 function compressWithConcat( $startId, $maxChunkSize, $maxChunkFactor, $factorThreshold, $beginDate, $endDate )
  63 {
  64         $fname = 'compressWithConcat';
  65         $loadStyle = LS_CHUNKED;
  66
  67         $dbr =& wfGetDB( DB_SLAVE );
  68         $dbw =& wfGetDB( DB_MASTER );
  69
  70         # Get all articles by page_id
  71         $maxPageId = $dbr->selectField( 'page', 'max(page_id)', '', $fname );
  72         $pageConds = array();
  73
  74         if ( $exclude_ns0 ) {
  75                 print "Excluding main namespace\n";
  76                 $pageConds[] = 'page_namespace<>0';
  77         }
  78         if ( $queryExtra ) {
  79                 $pageConds[] = $queryExtra;
  80         }
  81
  82         # For each article, get a list of revisions which fit the criteria
  83         # No recompression, use a condition on old_flags
  84         $conds = array("old_flags NOT LIKE '%object%'");
  85
  86         if ( $beginDate ) {
  87                 $conds[] = "rev_timestamp>'" . $beginDate . "'";
  88         }
  89         if ( $endDate )  {
  90                 $conds[] = "rev_timestamp<'" . $endDate . "'";
  91         }
  92         if ( $loadStyle == LS_CHUNKED ) {
  93                 $tables = array( 'revision', 'text' );
  94                 $fields = array( 'rev_id', 'rev_text_id', 'old_flags', 'old_text' );
  95                 $conds[] = 'rev_text_id=old_id';
  96                 $revLoadOptions = 'FOR UPDATE';
  97         } else {
  98                 $tables = array( 'revision' );
  99                 $fields = array( 'rev_id', 'rev_text_id' );
 100                 $revLoadOptions = array();
 101         }
 102
 103         $oldReadsSinceLastSlaveWait = 0;        #check slave lag periodically
 104         $totalMatchingRevisions = 0;
 105         $masterPos = false;
 106         for ( $pageId = $startId; $pageId <= $maxPageId; $pageId++ ) {
 107                 $pageRes = $dbr->select( 'page', array('page_id', 'page_namespace', 'page_title'),
 108                         $pageConds + array('page_id' => $pageId), $fname );
 109                 if ( $dbr->numRows( $pageRes ) == 0 ) {
 110                         continue;
 111                 }
 112                 $pageRow = $dbr->fetchObject( $pageRes );
 113
 114                 # Display progress
 115                 $titleObj = Title::makeTitle( $pageRow->page_namespace, $pageRow->page_title );
 116                 print "$pageId\t" . $titleObj->getPrefixedDBkey() . " ";
 117
 118                 # Load revisions
 119                 $revRes = $dbw->select( $tables, $fields,
 120                         array( 'rev_page' => $pageRow->page_id ) + $conds,
 121                         $fname,
 122                         $revLoadOptions
 123                 );
 124                 $revs = array();
 125                 while ( $revRow = $dbw->fetchObject( $revRes ) ) {
 126                         $revs[] = $revRow;
 127                 }
 128
 129                 if ( count( $revs ) < 2) {
 130                         # No revisions matching, no further processing
 131                         print "\n";
 132                         continue;
 133                 }
 134
 135                 # For each chunk
 136                 $i = 0;
 137                 while ( $i < count( $revs ) ) {
 138                         if ( $i < count( $revs ) - $maxChunkSize ) {
 139                                 $thisChunkSize = $maxChunkSize;
 140                         } else {
 141                                 $thisChunkSize = count( $revs ) - $i;
 142                         }
 143
 144                         $chunk = new ConcatenatedGzipHistoryBlob();
 145                         $stubs = array();
 146                         $dbw->begin();
 147                         $usedChunk = false;
 148                         $primaryOldid = $revs[$i]->rev_text_id;
 149
 150                         # Get the text of each revision and add it to the object
 151                         for ( $j = 0; $j < $thisChunkSize && $chunk->isHappy( $maxChunkFactor, $factorThreshold ); $j++ ) {
 152                                 $oldid = $revs[$i + $j]->rev_text_id;
 153
 154                                 # Get text
 155                                 if ( $loadStyle == LS_INDIVIDUAL ) {
 156                                         $textRow = $dbw->selectRow( 'text',
 157                                                 array( 'old_flags', 'old_text' ),
 158                                                 array( 'old_id' => $oldid ),
 159                                                 $fname,
 160                                                 'FOR UPDATE'
 161                                         );
 162                                         $text = Revision::getRevisionText( $textRow );
 163                                 } else {
 164                                         $text = Revision::getRevisionText( $revs[$i + $j] );
 165                                 }
 166
 167                                 if ( $text === false ) {
 168                                         print "\nError, unable to get text in old_id $oldid\n";
 169                                         #$dbw->delete( 'old', array( 'old_id' => $oldid ) );
 170                                 }
 171
 172                                 if ( $j == 0 ) {
 173                                         $chunk->setText( $text );
 174                                         print '.';
 175                                 } else {
 176                                         # Don't make a stub if it's going to be longer than the article
 177                                         # Stubs are typically about 100 bytes
 178                                         if ( strlen( $text ) < 120 ) {
 179                                                 $stub = false;
 180                                                 print 'x';
 181                                         } else {
 182                                                 $stub = $chunk->addItem( $text );
 183                                                 $stub->setLocation( $primaryOldid );
 184                                                 $hash = $stub->getHash();
 185                                                 $stub = serialize( $stub );
 186                                                 print '.';
 187                                                 $usedChunk = true;
 188                                         }
 189                                         $stubs[$j] = $stub;
 190                                 }
 191                         }
 192                         $thisChunkSize = $j;
 193
 194                         # If we couldn't actually use any stubs because the pages were too small, do nothing
 195                         if ( $usedChunk ) {
 196                                 # Store the main object
 197                                 $dbw->update( 'text',
 198                                         array( /* SET */
 199                                                 'old_text' => serialize( $chunk ),
 200                                                 'old_flags' => 'object',
 201                                         ), array( /* WHERE */
 202                                                 'old_id' => $primaryOldid
 203                                         )
 204                                 );
 205
 206                                 # Store the stub objects
 207                                 for ( $j = 1; $j < $thisChunkSize; $j++ ) {
 208                                         # Skip if not compressing
 209                                         if ( $stubs[$j] !== false ) {
 210                                                 $dbw->update( 'text',
 211                                                         array( /* SET */
 212                                                                 'old_text' => $stubs[$j],
 213                                                                 'old_flags' => 'object',
 214                                                         ), array( /* WHERE */
 215                                                                 'old_id' => $revs[$i + $j]->rev_text_id
 216                                                         )
 217                                                 );
 218                                         }
 219                                 }
 220                         }
 221                         # Done, next
 222                         print "/";
 223                         $dbw->commit();
 224                         $i += $thisChunkSize;
 225                 }
 226                 print "\n";
 227         }
 228         return true;
 229 }
 230 ?>