maintenance/storage/compressOld.inc

   1 <?php
   2 /**
   3  * @addtogroup Maintenance
   4  */
   5
   6 /** */
   7 require_once( 'Revision.php' );
   8 require_once( 'ExternalStoreDB.php' );
   9
  10 /** @todo document */
  11 function compressOldPages( $start = 0, $extdb = '' ) {
  12         $fname = 'compressOldPages';
  13
  14         $chunksize = 50;
  15         print "Starting from old_id $start...\n";
  16         $dbw = wfGetDB( DB_MASTER );
  17         do {
  18                 $res = $dbw->select( 'text', array( 'old_id','old_flags','old_namespace','old_title','old_text' ),
  19                         "old_id>=$start", $fname, array( 'ORDER BY' => 'old_id', 'LIMIT' => $chunksize, 'FOR UPDATE' ) );
  20                 if( $dbw->numRows( $res ) == 0 ) {
  21                         break;
  22                 }
  23                 $last = $start;
  24                 while( $row = $dbw->fetchObject( $res ) ) {
  25                         # print "  {$row->old_id} - {$row->old_namespace}:{$row->old_title}\n";
  26                         compressPage( $row, $extdb );
  27                         $last = $row->old_id;
  28                 }
  29                 $dbw->freeResult( $res );
  30                 $start = $last + 1; # Deletion may leave long empty stretches
  31                 print "$start...\n";
  32         } while( true );
  33 }
  34
  35 /** @todo document */
  36 function compressPage( $row, $extdb ) {
  37         $fname = 'compressPage';
  38         if ( false !== strpos( $row->old_flags, 'gzip' ) || false !== strpos( $row->old_flags, 'object' ) ) {
  39                 #print "Already compressed row {$row->old_id}\n";
  40                 return false;
  41         }
  42         $dbw = wfGetDB( DB_MASTER );
  43         $flags = $row->old_flags ? "{$row->old_flags},gzip" : "gzip";
  44         $compress = gzdeflate( $row->old_text );
  45
  46         # Store in external storage if required
  47         if ( $extdb !== '' ) {
  48                 $storeObj = new ExternalStoreDB;
  49                 $compress = $storeObj->store( $extdb, $compress );
  50                 if ( $compress === false ) {
  51                         print "Unable to store object\n";
  52                         return false;
  53                 }
  54         }
  55
  56         # Update text row
  57         $dbw->update( 'text',
  58                 array( /* SET */
  59                         'old_flags' => $flags,
  60                         'old_text' => $compress
  61                 ), array( /* WHERE */
  62                         'old_id' => $row->old_id
  63                 ), $fname, 'LIMIT 1'
  64         );
  65         return true;
  66 }
  67
  68 define( 'LS_INDIVIDUAL', 0 );
  69 define( 'LS_CHUNKED', 1 );
  70
  71 /** @todo document */
  72 function compressWithConcat( $startId, $maxChunkSize, $maxChunkFactor, $factorThreshold, $beginDate,
  73         $endDate, $extdb="", $maxPageId = false )
  74 {
  75         $fname = 'compressWithConcat';
  76         $loadStyle = LS_CHUNKED;
  77
  78         $dbr = wfGetDB( DB_SLAVE );
  79         $dbw = wfGetDB( DB_MASTER );
  80
  81         # Set up external storage
  82         if ( $extdb != '' ) {
  83                 $storeObj = new ExternalStoreDB;
  84         }
  85
  86         # Get all articles by page_id
  87         if ( !$maxPageId ) {
  88                 $maxPageId = $dbr->selectField( 'page', 'max(page_id)', '', $fname );
  89         }
  90         print "Starting from $startId of $maxPageId\n";
  91         $pageConds = array();
  92
  93         /*
  94         if ( $exclude_ns0 ) {
  95                 print "Excluding main namespace\n";
  96                 $pageConds[] = 'page_namespace<>0';
  97         }
  98         if ( $queryExtra ) {
  99                 $pageConds[] = $queryExtra;
 100         }
 101          */
 102
 103         # For each article, get a list of revisions which fit the criteria
 104
 105         # No recompression, use a condition on old_flags
 106         # Don't compress object type entities, because that might produce data loss when
 107         # overwriting bulk storage concat rows. Don't compress external references, because
 108         # the script doesn't yet delete rows from external storage.
 109         $conds = array(
 110                 "old_flags NOT LIKE '%object%' AND old_flags NOT LIKE '%external%'");
 111
 112         if ( $beginDate ) {
 113                 if ( !preg_match( '/^\d{14}$/', $beginDate ) ) {
 114                         print "Invalid begin date \"$beginDate\"\n";
 115                         return false;
 116                 }
 117                 $conds[] = "rev_timestamp>'" . $beginDate . "'";
 118         }
 119         if ( $endDate )  {
 120                 if ( !preg_match( '/^\d{14}$/', $endDate ) ) {
 121                         print "Invalid end date \"$endDate\"\n";
 122                         return false;
 123                 }
 124                 $conds[] = "rev_timestamp<'" . $endDate . "'";
 125         }
 126         if ( $loadStyle == LS_CHUNKED ) {
 127                 $tables = array( 'revision', 'text' );
 128                 $fields = array( 'rev_id', 'rev_text_id', 'old_flags', 'old_text' );
 129                 $conds[] = 'rev_text_id=old_id';
 130                 $revLoadOptions = 'FOR UPDATE';
 131         } else {
 132                 $tables = array( 'revision' );
 133                 $fields = array( 'rev_id', 'rev_text_id' );
 134                 $revLoadOptions = array();
 135         }
 136
 137         # Don't work with current revisions
 138         # Don't lock the page table for update either -- TS 2006-04-04
 139         #$tables[] = 'page';
 140         #$conds[] = 'page_id=rev_page AND rev_id != page_latest';
 141
 142         for ( $pageId = $startId; $pageId <= $maxPageId; $pageId++ ) {
 143                 wfWaitForSlaves( 5 );
 144
 145                 # Wake up
 146                 $dbr->ping();
 147
 148                 # Get the page row
 149                 $pageRes = $dbr->select( 'page',
 150                         array('page_id', 'page_namespace', 'page_title','page_latest'),
 151                         $pageConds + array('page_id' => $pageId), $fname );
 152                 if ( $dbr->numRows( $pageRes ) == 0 ) {
 153                         continue;
 154                 }
 155                 $pageRow = $dbr->fetchObject( $pageRes );
 156
 157                 # Display progress
 158                 $titleObj = Title::makeTitle( $pageRow->page_namespace, $pageRow->page_title );
 159                 print "$pageId\t" . $titleObj->getPrefixedDBkey() . " ";
 160
 161                 # Load revisions
 162                 $revRes = $dbw->select( $tables, $fields,
 163                         array_merge( array(
 164                                 'rev_page' => $pageRow->page_id,
 165                                 # Don't operate on the current revision
 166                                 # Use < instead of <> in case the current revision has changed
 167                                 # since the page select, which wasn't locking
 168                                 'rev_id < ' . $pageRow->page_latest
 169                         ), $conds ),
 170                         $fname,
 171                         $revLoadOptions
 172                 );
 173                 $revs = array();
 174                 while ( $revRow = $dbw->fetchObject( $revRes ) ) {
 175                         $revs[] = $revRow;
 176                 }
 177
 178                 if ( count( $revs ) < 2) {
 179                         # No revisions matching, no further processing
 180                         print "\n";
 181                         continue;
 182                 }
 183
 184                 # For each chunk
 185                 $i = 0;
 186                 while ( $i < count( $revs ) ) {
 187                         if ( $i < count( $revs ) - $maxChunkSize ) {
 188                                 $thisChunkSize = $maxChunkSize;
 189                         } else {
 190                                 $thisChunkSize = count( $revs ) - $i;
 191                         }
 192
 193                         $chunk = new ConcatenatedGzipHistoryBlob();
 194                         $stubs = array();
 195                         $dbw->begin();
 196                         $usedChunk = false;
 197                         $primaryOldid = $revs[$i]->rev_text_id;
 198
 199                         # Get the text of each revision and add it to the object
 200                         for ( $j = 0; $j < $thisChunkSize && $chunk->isHappy( $maxChunkFactor, $factorThreshold ); $j++ ) {
 201                                 $oldid = $revs[$i + $j]->rev_text_id;
 202
 203                                 # Get text
 204                                 if ( $loadStyle == LS_INDIVIDUAL ) {
 205                                         $textRow = $dbw->selectRow( 'text',
 206                                                 array( 'old_flags', 'old_text' ),
 207                                                 array( 'old_id' => $oldid ),
 208                                                 $fname,
 209                                                 'FOR UPDATE'
 210                                         );
 211                                         $text = Revision::getRevisionText( $textRow );
 212                                 } else {
 213                                         $text = Revision::getRevisionText( $revs[$i + $j] );
 214                                 }
 215
 216                                 if ( $text === false ) {
 217                                         print "\nError, unable to get text in old_id $oldid\n";
 218                                         #$dbw->delete( 'old', array( 'old_id' => $oldid ) );
 219                                 }
 220
 221                                 if ( $extdb == "" && $j == 0 ) {
 222                                         $chunk->setText( $text );
 223                                         print '.';
 224                                 } else {
 225                                         # Don't make a stub if it's going to be longer than the article
 226                                         # Stubs are typically about 100 bytes
 227                                         if ( strlen( $text ) < 120 ) {
 228                                                 $stub = false;
 229                                                 print 'x';
 230                                         } else {
 231                                                 $stub = $chunk->addItem( $text );
 232                                                 $stub->setLocation( $primaryOldid );
 233                                                 $stub->setReferrer( $oldid );
 234                                                 print '.';
 235                                                 $usedChunk = true;
 236                                         }
 237                                         $stubs[$j] = $stub;
 238                                 }
 239                         }
 240                         $thisChunkSize = $j;
 241
 242                         # If we couldn't actually use any stubs because the pages were too small, do nothing
 243                         if ( $usedChunk ) {
 244                                 if ( $extdb != "" ) {
 245                                         # Move blob objects to External Storage
 246                                         $stored = $storeObj->store( $extdb, serialize( $chunk ));
 247                                         if ($stored === false) {
 248                                                 print "Unable to store object\n";
 249                                                 return false;
 250                                         }
 251                                         # Store External Storage URLs instead of Stub placeholders
 252                                         foreach ($stubs as $stub) {
 253                                                 if ($stub===false)
 254                                                         continue;
 255                                                 # $stored should provide base path to a BLOB
 256                                                 $url = $stored."/".$stub->getHash();
 257                                                 $dbw->update( 'text',
 258                                                         array( /* SET */
 259                                                                 'old_text' => $url,
 260                                                                 'old_flags' => 'external,utf-8',
 261                                                         ), array ( /* WHERE */
 262                                                                 'old_id' => $stub->getReferrer(),
 263                                                         )
 264                                                 );
 265                                         }
 266                                 } else {
 267                                         # Store the main object locally
 268                                         $dbw->update( 'text',
 269                                                 array( /* SET */
 270                                                         'old_text' => serialize( $chunk ),
 271                                                         'old_flags' => 'object,utf-8',
 272                                                 ), array( /* WHERE */
 273                                                         'old_id' => $primaryOldid
 274                                                 )
 275                                         );
 276
 277                                         # Store the stub objects
 278                                         for ( $j = 1; $j < $thisChunkSize; $j++ ) {
 279                                                 # Skip if not compressing and don't overwrite the first revision
 280                                                 if ( $stubs[$j] !== false && $revs[$i + $j]->rev_text_id != $primaryOldid ) {
 281                                                         $dbw->update( 'text',
 282                                                                 array( /* SET */
 283                                                                         'old_text' => serialize($stubs[$j]),
 284                                                                         'old_flags' => 'object,utf-8',
 285                                                                 ), array( /* WHERE */
 286                                                                         'old_id' => $revs[$i + $j]->rev_text_id
 287                                                                 )
 288                                                         );
 289                                                 }
 290                                         }
 291                                 }
 292                         }
 293                         # Done, next
 294                         print "/";
 295                         $dbw->commit();
 296                         $i += $thisChunkSize;
 297                         wfWaitForSlaves( 5 );
 298                 }
 299                 print "\n";
 300         }
 301         return true;
 302 }
 303 ?>