maintenance/storage/compressOld.inc

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