maintenance/storage/compressOld.inc

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