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, $extdb="" )
  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         # Don't work with current revisions
 120         $tables[] = 'page';
 121         $conds[] = 'page_id=rev_page AND rev_id != page_latest';
 122
 123         $oldReadsSinceLastSlaveWait = 0;        #check slave lag periodically
 124         $totalMatchingRevisions = 0;
 125         $masterPos = false;
 126         for ( $pageId = $startId; $pageId <= $maxPageId; $pageId++ ) {
 127                 $pageRes = $dbr->select( 'page', array('page_id', 'page_namespace', 'page_title'),
 128                         $pageConds + array('page_id' => $pageId), $fname );
 129                 if ( $dbr->numRows( $pageRes ) == 0 ) {
 130                         continue;
 131                 }
 132                 $pageRow = $dbr->fetchObject( $pageRes );
 133
 134                 # Display progress
 135                 $titleObj = Title::makeTitle( $pageRow->page_namespace, $pageRow->page_title );
 136                 print "$pageId\t" . $titleObj->getPrefixedDBkey() . " ";
 137
 138                 # Load revisions
 139                 $revRes = $dbw->select( $tables, $fields,
 140                         array( 'rev_page' => $pageRow->page_id ) + $conds,
 141                         $fname,
 142                         $revLoadOptions
 143                 );
 144                 $revs = array();
 145                 while ( $revRow = $dbw->fetchObject( $revRes ) ) {
 146                         $revs[] = $revRow;
 147                 }
 148
 149                 if ( count( $revs ) < 2) {
 150                         # No revisions matching, no further processing
 151                         print "\n";
 152                         continue;
 153                 }
 154
 155                 # For each chunk
 156                 $i = 0;
 157                 while ( $i < count( $revs ) ) {
 158                         if ( $i < count( $revs ) - $maxChunkSize ) {
 159                                 $thisChunkSize = $maxChunkSize;
 160                         } else {
 161                                 $thisChunkSize = count( $revs ) - $i;
 162                         }
 163
 164                         $chunk = new ConcatenatedGzipHistoryBlob();
 165                         $stubs = array();
 166                         $dbw->begin();
 167                         $usedChunk = false;
 168                         $primaryOldid = $revs[$i]->rev_text_id;
 169
 170                         # Get the text of each revision and add it to the object
 171                         for ( $j = 0; $j < $thisChunkSize && $chunk->isHappy( $maxChunkFactor, $factorThreshold ); $j++ ) {
 172                                 $oldid = $revs[$i + $j]->rev_text_id;
 173
 174                                 # Get text
 175                                 if ( $loadStyle == LS_INDIVIDUAL ) {
 176                                         $textRow = $dbw->selectRow( 'text',
 177                                                 array( 'old_flags', 'old_text' ),
 178                                                 array( 'old_id' => $oldid ),
 179                                                 $fname,
 180                                                 'FOR UPDATE'
 181                                         );
 182                                         $text = Revision::getRevisionText( $textRow );
 183                                 } else {
 184                                         $text = Revision::getRevisionText( $revs[$i + $j] );
 185                                 }
 186
 187                                 if ( $text === false ) {
 188                                         print "\nError, unable to get text in old_id $oldid\n";
 189                                         #$dbw->delete( 'old', array( 'old_id' => $oldid ) );
 190                                 }
 191
 192                                 if ( $extdb == "" && $j == 0 ) {
 193                                         $chunk->setText( $text );
 194                                         print '.';
 195                                 } else {
 196                                         # Don't make a stub if it's going to be longer than the article
 197                                         # Stubs are typically about 100 bytes
 198                                         if ( strlen( $text ) < 120 ) {
 199                                                 $stub = false;
 200                                                 print 'x';
 201                                         } else {
 202                                                 $stub = $chunk->addItem( $text );
 203                                                 $stub->setLocation( $primaryOldid );
 204                                                 $stub->setReferrer( $oldid );
 205                                                 $hash = $stub->getHash();
 206                                                 print '.';
 207                                                 $usedChunk = true;
 208                                         }
 209                                         $stubs[$j] = $stub;
 210                                 }
 211                         }
 212                         $thisChunkSize = $j;
 213
 214                         # If we couldn't actually use any stubs because the pages were too small, do nothing
 215                         if ( $usedChunk ) {
 216                                 if ( $extdb != "" ) {
 217                                         # Move blob objects to External Storage
 218                                         $storeObj = new ExternalStoreDB;
 219                                         $stored = $storeObj->store( $extdb, serialize( $chunk ));
 220                                         if ($stored === false) {
 221                                                 print "Unable to store object\n";
 222                                                 return false;
 223                                         }
 224                                         # Store External Storage URLs instead of Stub placeholders
 225                                         foreach ($stubs as $stub) {
 226                                                 if ($stub===false)
 227                                                         continue;
 228                                                 # $stored should provide base path to a BLOB
 229                                                 $url = $stored."/".$stub->getHash();
 230                                                 $dbw->update( 'text',
 231                                                         array( /* SET */
 232                                                                 'old_text' => $url,
 233                                                                 'old_flags' => 'external,utf-8',
 234                                                         ), array ( /* WHERE */
 235                                                                 'old_id' => $stub->getReferrer(),
 236                                                         )
 237                                                 );
 238                                         }
 239                                 } else {
 240                                         # Store the main object locally
 241                                         $dbw->update( 'text',
 242                                                 array( /* SET */
 243                                                         'old_text' => serialize( $chunk ),
 244                                                         'old_flags' => 'object,utf-8',
 245                                                 ), array( /* WHERE */
 246                                                         'old_id' => $primaryOldid
 247                                                 )
 248                                         );
 249
 250                                         # Store the stub objects
 251                                         for ( $j = 1; $j < $thisChunkSize; $j++ ) {
 252                                                 # Skip if not compressing
 253                                                 if ( $stubs[$j] !== false ) {
 254                                                         $dbw->update( 'text',
 255                                                                 array( /* SET */
 256                                                                         'old_text' => serialize($stubs[$j]),
 257                                                                         'old_flags' => 'object,utf-8',
 258                                                                 ), array( /* WHERE */
 259                                                                         'old_id' => $revs[$i + $j]->rev_text_id
 260                                                                 )
 261                                                         );
 262                                                 }
 263                                         }
 264                                 }
 265                         }
 266                         # Done, next
 267                         print "/";
 268                         $dbw->commit();
 269                         $i += $thisChunkSize;
 270                 }
 271                 print "\n";
 272         }
 273         return true;
 274 }
 275 ?>