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,
  75         $endDate, $extdb="", $maxPageId = false )
  76 {
  77         $fname = 'compressWithConcat';
  78         $loadStyle = LS_CHUNKED;
  79
  80         $dbr =& wfGetDB( DB_SLAVE );
  81         $dbw =& wfGetDB( DB_MASTER );
  82
  83         # Set up external storage
  84         if ( $extdb != '' ) {
  85                 $storeObj = new ExternalStoreDB;
  86         }
  87
  88         # Get all articles by page_id
  89         if ( !$maxPageId ) {
  90                 $maxPageId = $dbr->selectField( 'page', 'max(page_id)', '', $fname );
  91         }
  92         print "Starting from $startId of $maxPageId\n";
  93         $pageConds = array();
  94
  95         /*
  96         if ( $exclude_ns0 ) {
  97                 print "Excluding main namespace\n";
  98                 $pageConds[] = 'page_namespace<>0';
  99         }
 100         if ( $queryExtra ) {
 101                 $pageConds[] = $queryExtra;
 102         }
 103          */
 104
 105         # For each article, get a list of revisions which fit the criteria
 106
 107         # No recompression, use a condition on old_flags
 108         # Don't compress object type entities, because that might produce data loss when
 109         # overwriting bulk storage concat rows. Don't compress external references, because
 110         # the script doesn't yet delete rows from external storage.
 111         $conds = array(
 112                 "old_flags NOT LIKE '%object%' AND old_flags NOT LIKE '%external%'");
 113
 114         if ( $beginDate ) {
 115                 $conds[] = "rev_timestamp>'" . $beginDate . "'";
 116         }
 117         if ( $endDate )  {
 118                 $conds[] = "rev_timestamp<'" . $endDate . "'";
 119         }
 120         if ( $loadStyle == LS_CHUNKED ) {
 121                 $tables = array( 'revision', 'text' );
 122                 $fields = array( 'rev_id', 'rev_text_id', 'old_flags', 'old_text' );
 123                 $conds[] = 'rev_text_id=old_id';
 124                 $revLoadOptions = 'FOR UPDATE';
 125         } else {
 126                 $tables = array( 'revision' );
 127                 $fields = array( 'rev_id', 'rev_text_id' );
 128                 $revLoadOptions = array();
 129         }
 130
 131         # Don't work with current revisions
 132         # Don't lock the page table for update either -- TS 2006-04-04
 133         #$tables[] = 'page';
 134         #$conds[] = 'page_id=rev_page AND rev_id != page_latest';
 135
 136         $oldReadsSinceLastSlaveWait = 0;        #check slave lag periodically
 137         $totalMatchingRevisions = 0;
 138         $masterPos = false;
 139         for ( $pageId = $startId; $pageId <= $maxPageId; $pageId++ ) {
 140                 wfWaitForSlaves( 5 );
 141
 142                 # Wake up
 143                 $dbr->ping();
 144
 145                 # Get the page row
 146                 $pageRes = $dbr->select( 'page',
 147                         array('page_id', 'page_namespace', 'page_title','page_latest'),
 148                         $pageConds + array('page_id' => $pageId), $fname );
 149                 if ( $dbr->numRows( $pageRes ) == 0 ) {
 150                         continue;
 151                 }
 152                 $pageRow = $dbr->fetchObject( $pageRes );
 153
 154                 # Display progress
 155                 $titleObj = Title::makeTitle( $pageRow->page_namespace, $pageRow->page_title );
 156                 print "$pageId\t" . $titleObj->getPrefixedDBkey() . " ";
 157
 158                 print_r(
 159                         array(
 160                                 'rev_page' => $pageRow->page_id,
 161                                 # Don't operate on the current revision
 162                                 # Use < instead of <> in case the current revision has changed
 163                                 # since the page select, which wasn't locking
 164                                 'rev_id < ' . $pageRow->page_latest
 165                         ) + $conds
 166                 );
 167                 exit;
 168
 169                 # Load revisions
 170                 $revRes = $dbw->select( $tables, $fields,
 171                         array(
 172                                 'rev_page' => $pageRow->page_id,
 173                                 # Don't operate on the current revision
 174                                 # Use < instead of <> in case the current revision has changed
 175                                 # since the page select, which wasn't locking
 176                                 'rev_id < ' . $pageRow->page_latest
 177                         ) + $conds,
 178                         $fname,
 179                         $revLoadOptions
 180                 );
 181                 $revs = array();
 182                 while ( $revRow = $dbw->fetchObject( $revRes ) ) {
 183                         $revs[] = $revRow;
 184                 }
 185
 186                 if ( count( $revs ) < 2) {
 187                         # No revisions matching, no further processing
 188                         print "\n";
 189                         continue;
 190                 }
 191
 192                 # For each chunk
 193                 $i = 0;
 194                 while ( $i < count( $revs ) ) {
 195                         if ( $i < count( $revs ) - $maxChunkSize ) {
 196                                 $thisChunkSize = $maxChunkSize;
 197                         } else {
 198                                 $thisChunkSize = count( $revs ) - $i;
 199                         }
 200
 201                         $chunk = new ConcatenatedGzipHistoryBlob();
 202                         $stubs = array();
 203                         $dbw->begin();
 204                         $usedChunk = false;
 205                         $primaryOldid = $revs[$i]->rev_text_id;
 206
 207                         # Get the text of each revision and add it to the object
 208                         for ( $j = 0; $j < $thisChunkSize && $chunk->isHappy( $maxChunkFactor, $factorThreshold ); $j++ ) {
 209                                 $oldid = $revs[$i + $j]->rev_text_id;
 210
 211                                 # Get text
 212                                 if ( $loadStyle == LS_INDIVIDUAL ) {
 213                                         $textRow = $dbw->selectRow( 'text',
 214                                                 array( 'old_flags', 'old_text' ),
 215                                                 array( 'old_id' => $oldid ),
 216                                                 $fname,
 217                                                 'FOR UPDATE'
 218                                         );
 219                                         $text = Revision::getRevisionText( $textRow );
 220                                 } else {
 221                                         $text = Revision::getRevisionText( $revs[$i + $j] );
 222                                 }
 223
 224                                 if ( $text === false ) {
 225                                         print "\nError, unable to get text in old_id $oldid\n";
 226                                         #$dbw->delete( 'old', array( 'old_id' => $oldid ) );
 227                                 }
 228
 229                                 if ( $extdb == "" && $j == 0 ) {
 230                                         $chunk->setText( $text );
 231                                         print '.';
 232                                 } else {
 233                                         # Don't make a stub if it's going to be longer than the article
 234                                         # Stubs are typically about 100 bytes
 235                                         if ( strlen( $text ) < 120 ) {
 236                                                 $stub = false;
 237                                                 print 'x';
 238                                         } else {
 239                                                 $stub = $chunk->addItem( $text );
 240                                                 $stub->setLocation( $primaryOldid );
 241                                                 $stub->setReferrer( $oldid );
 242                                                 print '.';
 243                                                 $usedChunk = true;
 244                                         }
 245                                         $stubs[$j] = $stub;
 246                                 }
 247                         }
 248                         $thisChunkSize = $j;
 249
 250                         # If we couldn't actually use any stubs because the pages were too small, do nothing
 251                         if ( $usedChunk ) {
 252                                 if ( $extdb != "" ) {
 253                                         # Move blob objects to External Storage
 254                                         $stored = $storeObj->store( $extdb, serialize( $chunk ));
 255                                         if ($stored === false) {
 256                                                 print "Unable to store object\n";
 257                                                 return false;
 258                                         }
 259                                         # Store External Storage URLs instead of Stub placeholders
 260                                         foreach ($stubs as $stub) {
 261                                                 if ($stub===false)
 262                                                         continue;
 263                                                 # $stored should provide base path to a BLOB
 264                                                 $url = $stored."/".$stub->getHash();
 265                                                 $dbw->update( 'text',
 266                                                         array( /* SET */
 267                                                                 'old_text' => $url,
 268                                                                 'old_flags' => 'external,utf-8',
 269                                                         ), array ( /* WHERE */
 270                                                                 'old_id' => $stub->getReferrer(),
 271                                                         )
 272                                                 );
 273                                         }
 274                                 } else {
 275                                         # Store the main object locally
 276                                         $dbw->update( 'text',
 277                                                 array( /* SET */
 278                                                         'old_text' => serialize( $chunk ),
 279                                                         'old_flags' => 'object,utf-8',
 280                                                 ), array( /* WHERE */
 281                                                         'old_id' => $primaryOldid
 282                                                 )
 283                                         );
 284
 285                                         # Store the stub objects
 286                                         for ( $j = 1; $j < $thisChunkSize; $j++ ) {
 287                                                 # Skip if not compressing
 288                                                 if ( $stubs[$j] !== false ) {
 289                                                         $dbw->update( 'text',
 290                                                                 array( /* SET */
 291                                                                         'old_text' => serialize($stubs[$j]),
 292                                                                         'old_flags' => 'object,utf-8',
 293                                                                 ), array( /* WHERE */
 294                                                                         'old_id' => $revs[$i + $j]->rev_text_id
 295                                                                 )
 296                                                         );
 297                                                 }
 298                                         }
 299                                 }
 300                         }
 301                         # Done, next
 302                         print "/";
 303                         $dbw->commit();
 304                         $i += $thisChunkSize;
 305                         wfWaitForSlaves( 5 );
 306                 }
 307                 print "\n";
 308         }
 309         return true;
 310 }
 311 ?>