maintenance/storage/compressOld.inc

   1 <?php
   2 /**
   3  * @package MediaWiki
   4  * @subpackage Maintenance
   5  */
   6
   7 /** */
   8 require_once( 'Revision.php' );
   9 require_once( 'ExternalStoreDB.php' );
  10
  11 /** @todo document */
  12 function compressOldPages( $start = 0, $extdb = '' ) {
  13         $fname = 'compressOldPages';
  14
  15         $chunksize = 50;
  16         print "Starting from old_id $start...\n";
  17         $dbw =& wfGetDB( DB_MASTER );
  18         do {
  19                 $end = $start + $chunksize;
  20                 $res = $dbw->select( 'text', array( 'old_id','old_flags','old_namespace','old_title','old_text' ),
  21                         "old_id>=$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, $extdb );
  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, $extdb ) {
  39         $fname = 'compressPage';
  40         if ( false !== strpos( $row->old_flags, 'gzip' ) || false !== strpos( $row->old_flags, 'object' ) ) {
  41                 #print "Already compressed row {$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
  48         # Store in external storage if required
  49         if ( $extdb !== '' ) {
  50                 $storeObj = new ExternalStoreDB;
  51                 $compress = $storeObj->store( $extdb, $compress );
  52                 if ( $compress === false ) {
  53                         print "Unable to store object\n";
  54                         return false;
  55                 }
  56         }
  57
  58         # Update text row
  59         $dbw->update( 'text',
  60                 array( /* SET */
  61                         'old_flags' => $flags,
  62                         'old_text' => $compress
  63                 ), array( /* WHERE */
  64                         'old_id' => $row->old_id
  65                 ), $fname, 'LIMIT 1'
  66         );
  67         return true;
  68 }
  69
  70 define( 'LS_INDIVIDUAL', 0 );
  71 define( 'LS_CHUNKED', 1 );
  72
  73 /** @todo document */
  74 function compressWithConcat( $startId, $maxChunkSize, $maxChunkFactor, $factorThreshold, $beginDate, $endDate )
  75 {
  76         $fname = 'compressWithConcat';
  77         $loadStyle = LS_CHUNKED;
  78
  79         $dbr =& wfGetDB( DB_SLAVE );
  80         $dbw =& wfGetDB( DB_MASTER );
  81
  82         # Get all articles by page_id
  83         $maxPageId = $dbr->selectField( 'page', 'max(page_id)', '', $fname );
  84         $pageConds = array();
  85
  86         /*
  87         if ( $exclude_ns0 ) {
  88                 print "Excluding main namespace\n";
  89                 $pageConds[] = 'page_namespace<>0';
  90         }
  91         if ( $queryExtra ) {
  92                 $pageConds[] = $queryExtra;
  93         }
  94          */
  95
  96         # For each article, get a list of revisions which fit the criteria
  97         # No recompression, use a condition on old_flags
  98         $conds = array(
  99                 "old_flags NOT LIKE '%object%' " .
 100                 " AND (old_flags NOT LIKE '%external%' OR old_text NOT LIKE 'DB://%/%/%')");
 101
 102         if ( $beginDate ) {
 103                 $conds[] = "rev_timestamp>'" . $beginDate . "'";
 104         }
 105         if ( $endDate )  {
 106                 $conds[] = "rev_timestamp<'" . $endDate . "'";
 107         }
 108         if ( $loadStyle == LS_CHUNKED ) {
 109                 $tables = array( 'revision', 'text' );
 110                 $fields = array( 'rev_id', 'rev_text_id', 'old_flags', 'old_text' );
 111                 $conds[] = 'rev_text_id=old_id';
 112                 $revLoadOptions = 'FOR UPDATE';
 113         } else {
 114                 $tables = array( 'revision' );
 115                 $fields = array( 'rev_id', 'rev_text_id' );
 116                 $revLoadOptions = array();
 117         }
 118
 119         $oldReadsSinceLastSlaveWait = 0;        #check slave lag periodically
 120         $totalMatchingRevisions = 0;
 121         $masterPos = false;
 122         for ( $pageId = $startId; $pageId <= $maxPageId; $pageId++ ) {
 123                 $pageRes = $dbr->select( 'page', array('page_id', 'page_namespace', 'page_title'),
 124                         $pageConds + array('page_id' => $pageId), $fname );
 125                 if ( $dbr->numRows( $pageRes ) == 0 ) {
 126                         continue;
 127                 }
 128                 $pageRow = $dbr->fetchObject( $pageRes );
 129
 130                 # Display progress
 131                 $titleObj = Title::makeTitle( $pageRow->page_namespace, $pageRow->page_title );
 132                 print "$pageId\t" . $titleObj->getPrefixedDBkey() . " ";
 133
 134                 # Load revisions
 135                 $revRes = $dbw->select( $tables, $fields,
 136                         array( 'rev_page' => $pageRow->page_id ) + $conds,
 137                         $fname,
 138                         $revLoadOptions
 139                 );
 140                 $revs = array();
 141                 while ( $revRow = $dbw->fetchObject( $revRes ) ) {
 142                         $revs[] = $revRow;
 143                 }
 144
 145                 if ( count( $revs ) < 2) {
 146                         # No revisions matching, no further processing
 147                         print "\n";
 148                         continue;
 149                 }
 150
 151                 # For each chunk
 152                 $i = 0;
 153                 while ( $i < count( $revs ) ) {
 154                         if ( $i < count( $revs ) - $maxChunkSize ) {
 155                                 $thisChunkSize = $maxChunkSize;
 156                         } else {
 157                                 $thisChunkSize = count( $revs ) - $i;
 158                         }
 159
 160                         $chunk = new ConcatenatedGzipHistoryBlob();
 161                         $stubs = array();
 162                         $dbw->begin();
 163                         $usedChunk = false;
 164                         $primaryOldid = $revs[$i]->rev_text_id;
 165
 166                         # Get the text of each revision and add it to the object
 167                         for ( $j = 0; $j < $thisChunkSize && $chunk->isHappy( $maxChunkFactor, $factorThreshold ); $j++ ) {
 168                                 $oldid = $revs[$i + $j]->rev_text_id;
 169
 170                                 # Get text
 171                                 if ( $loadStyle == LS_INDIVIDUAL ) {
 172                                         $textRow = $dbw->selectRow( 'text',
 173                                                 array( 'old_flags', 'old_text' ),
 174                                                 array( 'old_id' => $oldid ),
 175                                                 $fname,
 176                                                 'FOR UPDATE'
 177                                         );
 178                                         $text = Revision::getRevisionText( $textRow );
 179                                 } else {
 180                                         $text = Revision::getRevisionText( $revs[$i + $j] );
 181                                 }
 182
 183                                 if ( $text === false ) {
 184                                         print "\nError, unable to get text in old_id $oldid\n";
 185                                         #$dbw->delete( 'old', array( 'old_id' => $oldid ) );
 186                                 }
 187
 188                                 if ( $j == 0 ) {
 189                                         $chunk->setText( $text );
 190                                         print '.';
 191                                 } else {
 192                                         # Don't make a stub if it's going to be longer than the article
 193                                         # Stubs are typically about 100 bytes
 194                                         if ( strlen( $text ) < 120 ) {
 195                                                 $stub = false;
 196                                                 print 'x';
 197                                         } else {
 198                                                 $stub = $chunk->addItem( $text );
 199                                                 $stub->setLocation( $primaryOldid );
 200                                                 $hash = $stub->getHash();
 201                                                 $stub = serialize( $stub );
 202                                                 print '.';
 203                                                 $usedChunk = true;
 204                                         }
 205                                         $stubs[$j] = $stub;
 206                                 }
 207                         }
 208                         $thisChunkSize = $j;
 209
 210                         # If we couldn't actually use any stubs because the pages were too small, do nothing
 211                         if ( $usedChunk ) {
 212                                 # Store the main object
 213                                 $dbw->update( 'text',
 214                                         array( /* SET */
 215                                                 'old_text' => serialize( $chunk ),
 216                                                 'old_flags' => 'object,utf-8',
 217                                         ), array( /* WHERE */
 218                                                 'old_id' => $primaryOldid
 219                                         )
 220                                 );
 221
 222                                 # Store the stub objects
 223                                 for ( $j = 1; $j < $thisChunkSize; $j++ ) {
 224                                         # Skip if not compressing
 225                                         if ( $stubs[$j] !== false ) {
 226                                                 $dbw->update( 'text',
 227                                                         array( /* SET */
 228                                                                 'old_text' => $stubs[$j],
 229                                                                 'old_flags' => 'object,utf-8',
 230                                                         ), array( /* WHERE */
 231                                                                 'old_id' => $revs[$i + $j]->rev_text_id
 232                                                         )
 233                                                 );
 234                                         }
 235                                 }
 236                         }
 237                         # Done, next
 238                         print "/";
 239                         $dbw->commit();
 240                         $i += $thisChunkSize;
 241                 }
 242                 print "\n";
 243         }
 244         return true;
 245 }
 246 ?>