maintenance/storage/checkStorage.php

   1 <?php
   2
   3 /**
   4  * Fsck for MediaWiki
   5  */
   6
   7 define( 'CONCAT_HEADER', 'O:27:"concatenatedgziphistoryblob"' );
   8
   9 if ( !defined( 'MEDIAWIKI' ) ) {
  10         require_once( dirname(__FILE__) . '/../commandLine.inc' );
  11         require_once( 'ExternalStore.php' );
  12         require_once( 'ExternalStoreDB.php' );
  13         require_once( 'SpecialImport.php' );
  14
  15         $cs = new CheckStorage;
  16         $fix = isset( $options['fix'] );
  17         if ( isset( $args[0] ) ) {
  18                 $xml = $args[0];
  19         } else {
  20                 $xml = false;
  21         }
  22         $cs->check( $fix, $xml );
  23 }
  24
  25
  26 //----------------------------------------------------------------------------------
  27
  28 class CheckStorage
  29 {
  30         var $oldIdMap, $errors;
  31         var $dbStore = null;
  32
  33         var $errorDescriptions = array(
  34                 'restore text' => 'Damaged text, need to be restored from a backup',
  35                 'restore revision' => 'Damaged revision row, need to be restored from a backup',
  36                 'unfixable' => 'Unexpected errors with no automated fixing method',
  37                 'fixed' => 'Errors already fixed',
  38                 'fixable' => 'Errors which would already be fixed if --fix was specified',
  39         );
  40
  41         function check( $fix = false, $xml = '' ) {
  42                 $fname = 'checkStorage';
  43                 $dbr =& wfGetDB( DB_SLAVE );
  44                 if ( $fix ) {
  45                         $dbw =& wfGetDB( DB_MASTER );
  46                         print "Checking, will fix errors if possible...\n";
  47                 } else {
  48                         print "Checking...\n";
  49                 }
  50                 $maxRevId = $dbr->selectField( 'revision', 'MAX(rev_id)', false, $fname );
  51                 $chunkSize = 1000;
  52                 $flagStats = array();
  53                 $objectStats = array();
  54                 $knownFlags = array( 'external', 'gzip', 'object', 'utf-8' );
  55                 $this->errors = array(
  56                         'restore text' => array(),
  57                         'restore revision' => array(),
  58                         'unfixable' => array(),
  59                         'fixed' => array(),
  60                         'fixable' => array(),
  61                 );
  62
  63                 for ( $chunkStart = 1 ; $chunkStart < $maxRevId; $chunkStart += $chunkSize ) {
  64                         $chunkEnd = $chunkStart + $chunkSize - 1;
  65                         //print "$chunkStart of $maxRevId\n";
  66
  67                         // Fetch revision rows
  68                         $this->oldIdMap = array();
  69                         $dbr->ping();
  70                         $res = $dbr->select( 'revision', array( 'rev_id', 'rev_text_id' ),
  71                                 array( "rev_id BETWEEN $chunkStart AND $chunkEnd" ), $fname );
  72                         while ( $row = $dbr->fetchObject( $res ) ) {
  73                                 $this->oldIdMap[$row->rev_id] = $row->rev_text_id;
  74                         }
  75                         $dbr->freeResult( $res );
  76
  77                         if ( !count( $this->oldIdMap ) ) {
  78                                 continue;
  79                         }
  80
  81                         // Fetch old_flags
  82                         $missingTextRows = array_flip( $this->oldIdMap );
  83                         $externalRevs = array();
  84                         $objectRevs = array();
  85                         $res = $dbr->select( 'text', array( 'old_id', 'old_flags' ),
  86                                 'old_id IN (' . implode( ',', $this->oldIdMap ) . ')', $fname );
  87                         while ( $row = $dbr->fetchObject( $res ) ) {
  88                                 $flags = $row->old_flags;
  89                                 $id = $row->old_id;
  90
  91                                 // Create flagStats row if it doesn't exist
  92                                 $flagStats = $flagStats + array( $flags => 0 );
  93                                 // Increment counter
  94                                 $flagStats[$flags]++;
  95
  96                                 // Not missing
  97                                 unset( $missingTextRows[$row->old_id] );
  98
  99                                 // Check for external or object
 100                                 if ( $flags == '' ) {
 101                                         $flagArray = array();
 102                                 } else {
 103                                         $flagArray = explode( ',', $flags );
 104                                 }
 105                                 if ( in_array( 'external', $flagArray ) ) {
 106                                         $externalRevs[] = $id;
 107                                 } elseif ( in_array( 'object', $flagArray ) ) {
 108                                         $objectRevs[] = $id;
 109                                 }
 110
 111                                 // Check for unrecognised flags
 112                                 if ( $flags == '0' ) {
 113                                         // This is a known bug from 2004
 114                                         // It's safe to just erase the old_flags field
 115                                         if ( $fix ) {
 116                                                 $this->error( 'fixed', "Warning: old_flags set to 0", $id );
 117                                                 $dbw->ping();
 118                                                 $dbw->update( 'text', array( 'old_flags' => '' ),
 119                                                         array( 'old_id' => $id ), $fname );
 120                                                 echo "Fixed\n";
 121                                         } else {
 122                                                 $this->error( 'fixable', "Warning: old_flags set to 0", $id );
 123                                         }
 124                                 } elseif ( count( array_diff( $flagArray, $knownFlags ) ) ) {
 125                                         $this->error( 'unfixable', "Error: invalid flags field \"$flags\"", $id );
 126                                 }
 127                         }
 128                         $dbr->freeResult( $res );
 129
 130                         // Output errors for any missing text rows
 131                         foreach ( $missingTextRows as $oldId => $revId ) {
 132                                 $this->error( 'restore revision', "Error: missing text row", $oldId );
 133                         }
 134
 135                         // Verify external revisions
 136                         $externalConcatBlobs = array();
 137                         $externalNormalBlobs = array();
 138                         if ( count( $externalRevs ) ) {
 139                                 $res = $dbr->select( 'text', array( 'old_id', 'old_flags', 'old_text' ),
 140                                         array( 'old_id IN (' . implode( ',', $externalRevs ) . ')' ), $fname );
 141                                 while ( $row = $dbr->fetchObject( $res ) ) {
 142                                         $urlParts = explode( '://', $row->old_text, 2 );
 143                                         if ( count( $urlParts ) !== 2 || $urlParts[1] == '' ) {
 144                                                 $this->error( 'restore text', "Error: invalid URL \"{$row->old_text}\"", $row->old_id );
 145                                                 continue;
 146                                         }
 147                                         list( $proto, $path ) = $urlParts;
 148                                         if ( $proto != 'DB' ) {
 149                                                 $this->error( 'restore text', "Error: invalid external protocol \"$proto\"", $row->old_id );
 150                                                 continue;
 151                                         }
 152                                         $path = explode( '/', $row->old_text );
 153                                         $cluster = $path[2];
 154                                         $id = $path[3];
 155                                         if ( isset( $path[4] ) ) {
 156                                                 $externalConcatBlobs[$cluster][$id][] = $row->old_id;
 157                                         } else {
 158                                                 $externalNormalBlobs[$cluster][$id][] = $row->old_id;
 159                                         }
 160                                 }
 161                                 $dbr->freeResult( $res );
 162                         }
 163
 164                         // Check external concat blobs for the right header
 165                         $this->checkExternalConcatBlobs( $externalConcatBlobs );
 166
 167                         // Check external normal blobs for existence
 168                         if ( count( $externalNormalBlobs ) ) {
 169                                 if ( is_null( $this->dbStore ) ) {
 170                                         $this->dbStore = new ExternalStoreDB;
 171                                 }
 172                                 foreach ( $externalConcatBlobs as $cluster => $xBlobIds ) {
 173                                         $blobIds = array_keys( $xBlobIds );
 174                                         $extDb =& $this->dbStore->getSlave( $cluster );
 175                                         $blobsTable = $this->dbStore->getTable( $extDb );
 176                                         $res = $extDb->select( $blobsTable,
 177                                                 array( 'blob_id' ),
 178                                                 array( 'blob_id IN( ' . implode( ',', $blobIds ) . ')' ), $fname );
 179                                         while ( $row = $extDb->fetchObject( $res ) ) {
 180                                                 unset( $xBlobIds[$row->blob_id] );
 181                                         }
 182                                         $extDb->freeResult( $res );
 183                                         // Print errors for missing blobs rows
 184                                         foreach ( $xBlobIds as $blobId => $oldId ) {
 185                                                 $this->error( 'restore text', "Error: missing target $blobId for one-part ES URL", $oldId );
 186                                         }
 187                                 }
 188                         }
 189
 190                         // Check local objects
 191                         $dbr->ping();
 192                         $concatBlobs = array();
 193                         $curIds = array();
 194                         if ( count( $objectRevs ) ) {
 195                                 $headerLength = 300;
 196                                 $res = $dbr->select( 'text', array( 'old_id', 'old_flags', "LEFT(old_text, $headerLength) AS header" ),
 197                                         array( 'old_id IN (' . implode( ',', $objectRevs ) . ')' ), $fname );
 198                                 while ( $row = $dbr->fetchObject( $res ) ) {
 199                                         $oldId = $row->old_id;
 200                                         if ( !preg_match( '/^O:(\d+):"(\w+)"/', $row->header, $matches ) ) {
 201                                                 $this->error( 'restore text', "Error: invalid object header", $oldId );
 202                                                 continue;
 203                                         }
 204
 205                                         $className = strtolower( $matches[2] );
 206                                         if ( strlen( $className ) != $matches[1] ) {
 207                                                 $this->error( 'restore text', "Error: invalid object header, wrong class name length", $oldId );
 208                                                 continue;
 209                                         }
 210
 211                                         $objectStats = $objectStats + array( $className => 0 );
 212                                         $objectStats[$className]++;
 213
 214                                         switch ( $className ) {
 215                                                 case 'concatenatedgziphistoryblob':
 216                                                         // Good
 217                                                         break;
 218                                                 case 'historyblobstub':
 219                                                 case 'historyblobcurstub':
 220                                                         if ( strlen( $row->header ) == $headerLength ) {
 221                                                                 $this->error( 'unfixable', "Error: overlong stub header", $oldId );
 222                                                                 continue;
 223                                                         }
 224                                                         $stubObj = unserialize( $row->header );
 225                                                         if ( !is_object( $stubObj ) ) {
 226                                                                 $this->error( 'restore text', "Error: unable to unserialize stub object", $oldId );
 227                                                                 continue;
 228                                                         }
 229                                                         if ( $className == 'historyblobstub' ) {
 230                                                                 $concatBlobs[$stubObj->mOldId][] = $oldId;
 231                                                         } else {
 232                                                                 $curIds[$stubObj->mCurId][] = $oldId;
 233                                                         }
 234                                                         break;
 235                                                 default:
 236                                                         $this->error( 'unfixable', "Error: unrecognised object class \"$className\"", $oldId );
 237                                         }
 238                                 }
 239                                 $dbr->freeResult( $res );
 240                         }
 241
 242                         // Check local concat blob validity
 243                         $externalConcatBlobs = array();
 244                         if ( count( $concatBlobs ) ) {
 245                                 $headerLength = 300;
 246                                 $res = $dbr->select( 'text', array( 'old_id', 'old_flags', "LEFT(old_text, $headerLength) AS header" ),
 247                                         array( 'old_id IN (' . implode( ',', array_keys( $concatBlobs ) ) . ')' ), $fname );
 248                                 while ( $row = $dbr->fetchObject( $res ) ) {
 249                                         $flags = explode( ',', $row->old_flags );
 250                                         if ( in_array( 'external', $flags ) ) {
 251                                                 // Concat blob is in external storage?
 252                                                 if ( in_array( 'object', $flags ) ) {
 253                                                         $urlParts = explode( '/', $row->header );
 254                                                         if ( $urlParts[0] != 'DB:' ) {
 255                                                                 $this->error( 'unfixable', "Error: unrecognised external storage type \"{$urlParts[0]}", $row->old_id );
 256                                                         } else {
 257                                                                 $cluster = $urlParts[2];
 258                                                                 $id = $urlParts[3];
 259                                                                 if ( !isset( $externalConcatBlobs[$cluster][$id] ) ) {
 260                                                                         $externalConcatBlobs[$cluster][$id] = array();
 261                                                                 }
 262                                                                 $externalConcatBlobs[$cluster][$id] = array_merge(
 263                                                                         $externalConcatBlobs[$cluster][$id], $concatBlobs[$row->old_id]
 264                                                                 );
 265                                                         }
 266                                                 } else {
 267                                                         $this->error( 'unfixable', "Error: invalid flags \"{$row->old_flags}\" on concat bulk row {$row->old_id}",
 268                                                                 $concatBlobs[$row->old_id] );
 269                                                 }
 270                                         } elseif ( strcasecmp( substr( $row->header, 0, strlen( CONCAT_HEADER ) ), CONCAT_HEADER ) ) {
 271                                                 $this->error( 'restore text', "Error: Incorrect object header for concat bulk row {$row->old_id}",
 272                                                         $concatBlobs[$row->old_id] );
 273                                         } # else good
 274
 275                                         unset( $concatBlobs[$row->old_id] );
 276                                 }
 277                                 $dbr->freeResult( $res );
 278                         }
 279
 280                         // Check targets of unresolved stubs
 281                         $this->checkExternalConcatBlobs( $externalConcatBlobs );
 282
 283                         // next chunk
 284                 }
 285
 286                 print "\n\nErrors:\n";
 287                 foreach( $this->errors as $name => $errors ) {
 288                         if ( count( $errors ) ) {
 289                                 $description = $this->errorDescriptions[$name];
 290                                 echo "$description: " . implode( ',', array_keys( $errors ) ) . "\n";
 291                         }
 292                 }
 293
 294                 if ( count( $this->errors['restore text'] ) && $fix ) {
 295                         if ( (string)$xml !== '' ) {
 296                                 $this->restoreText( array_keys( $this->errors['restore text'] ), $xml );
 297                         } else {
 298                                 echo "Can't fix text, no XML backup specified\n";
 299                         }
 300                 }
 301
 302                 print "\nFlag statistics:\n";
 303                 $total = array_sum( $flagStats );
 304                 foreach ( $flagStats as $flag => $count ) {
 305                         printf( "%-30s %10d %5.2f%%\n", $flag, $count, $count / $total * 100 );
 306                 }
 307                 print "\nLocal object statistics:\n";
 308                 $total = array_sum( $objectStats );
 309                 foreach ( $objectStats as $className => $count ) {
 310                         printf( "%-30s %10d %5.2f%%\n", $className, $count, $count / $total * 100 );
 311                 }
 312         }
 313
 314
 315         function error( $type, $msg, $ids ) {
 316                 if ( is_array( $ids ) && count( $ids ) == 1 ) {
 317                         $ids = reset( $ids );
 318                 }
 319                 if ( is_array( $ids ) ) {
 320                         $revIds = array();
 321                         foreach ( $ids as $id ) {
 322                                 $revIds = array_merge( $revIds, array_keys( $this->oldIdMap, $id ) );
 323                         }
 324                         print "$msg in text rows " . implode( ', ', $ids ) .
 325                                 ", revisions " . implode( ', ', $revIds ) . "\n";
 326                 } else {
 327                         $id = $ids;
 328                         $revIds = array_keys( $this->oldIdMap, $id );
 329                         if ( count( $revIds ) == 1 ) {
 330                                 print "$msg in old_id $id, rev_id {$revIds[0]}\n";
 331                         } else {
 332                                 print "$msg in old_id $id, revisions " . implode( ', ', $revIds ) . "\n";
 333                         }
 334                 }
 335                 $this->errors[$type] = $this->errors[$type] + array_flip( $revIds );
 336         }
 337
 338         function checkExternalConcatBlobs( $externalConcatBlobs ) {
 339                 $fname = 'CheckStorage::checkExternalConcatBlobs';
 340                 if ( !count( $externalConcatBlobs ) ) {
 341                         return;
 342                 }
 343
 344                 if ( is_null( $this->dbStore ) ) {
 345                         $this->dbStore = new ExternalStoreDB;
 346                 }
 347
 348                 foreach ( $externalConcatBlobs as $cluster => $oldIds ) {
 349                         $blobIds = array_keys( $oldIds );
 350                         $extDb =& $this->dbStore->getSlave( $cluster );
 351                         $blobsTable = $this->dbStore->getTable( $extDb );
 352                         $headerLength = strlen( CONCAT_HEADER );
 353                         $res = $extDb->select( $blobsTable,
 354                                 array( 'blob_id', "LEFT(blob_text, $headerLength) AS header" ),
 355                                 array( 'blob_id IN( ' . implode( ',', $blobIds ) . ')' ), $fname );
 356                         while ( $row = $extDb->fetchObject( $res ) ) {
 357                                 if ( strcasecmp( $row->header, CONCAT_HEADER ) ) {
 358                                         $this->error( 'restore text', "Error: invalid header on target $cluster/{$row->blob_id} of two-part ES URL",
 359                                                 $oldIds[$row->blob_id] );
 360                                 }
 361                                 unset( $oldIds[$row->blob_id] );
 362
 363                         }
 364                         $extDb->freeResult( $res );
 365
 366                         // Print errors for missing blobs rows
 367                         foreach ( $oldIds as $blobId => $oldIds ) {
 368                                 $this->error( 'restore text', "Error: missing target $cluster/$blobId for two-part ES URL", $oldIds );
 369                         }
 370                 }
 371         }
 372
 373         function restoreText( $revIds, $xml ) {
 374                 global $wgTmpDirectory, $wgDBname;
 375
 376                 if ( !count( $revIds ) ) {
 377                         return;
 378                 }
 379
 380                 print "Restoring text from XML backup...\n";
 381
 382                 $revFileName = "$wgTmpDirectory/broken-revlist-$wgDBname";
 383                 $filteredXmlFileName = "$wgTmpDirectory/filtered-$wgDBname.xml";
 384
 385                 // Write revision list
 386                 if ( !file_put_contents( $revFileName, implode( "\n", $revIds ) ) ) {
 387                         echo "Error writing revision list, can't restore text\n";
 388                         return;
 389                 }
 390
 391                 // Run mwdumper
 392                 echo "Filtering XML dump...\n";
 393                 $exitStatus = 0;
 394                 passthru( 'mwdumper ' .
 395                         wfEscapeShellArg(
 396                                 "--output=file:$filteredXmlFileName",
 397                                 "--filter=revlist:$revFileName",
 398                                 $xml
 399                         ), $exitStatus
 400                 );
 401
 402                 if ( $exitStatus ) {
 403                         echo "mwdumper died with exit status $exitStatus\n";
 404                         return;
 405                 }
 406
 407                 $file = fopen( $filteredXmlFileName, 'r' );
 408                 if ( !$file ) {
 409                         echo "Unable to open filtered XML file\n";
 410                         return;
 411                 }
 412
 413                 $dbr =& wfGetDB( DB_SLAVE );
 414                 $dbw =& wfGetDB( DB_MASTER );
 415                 $dbr->ping();
 416                 $dbw->ping();
 417
 418                 $source = new ImportStreamSource( $file );
 419                 $importer = new WikiImporter( $source );
 420                 $importer->setRevisionCallback( array( &$this, 'importRevision' ) );
 421                 $importer->doImport();
 422         }
 423
 424         function importRevision( &$revision, &$importer ) {
 425                 $fname = 'CheckStorage::importRevision';
 426
 427                 $id = $revision->getID();
 428                 $text = $revision->getText();
 429                 if ( $text === '' ) {
 430                         // This is what happens if the revision was broken at the time the
 431                         // dump was made. Unfortunately, it also happens if the revision was
 432                         // legitimately blank, so there's no way to tell the difference. To
 433                         // be safe, we'll skip it and leave it broken
 434                         $id = $id ? $id : '';
 435                         echo "Revision $id is blank in the dump, may have been broken before export\n";
 436                         return;
 437                 }
 438
 439                 if ( !$id )  {
 440                         // No ID, can't import
 441                         echo "No id tag in revision, can't import\n";
 442                         return;
 443                 }
 444
 445                 // Find text row again
 446                 $dbr =& wfGetDB( DB_SLAVE );
 447                 $oldId = $dbr->selectField( 'revision', 'rev_text_id', array( 'rev_id' => $id ), $fname );
 448                 if ( !$oldId ) {
 449                         echo "Missing revision row for rev_id $id\n";
 450                         return;
 451                 }
 452
 453                 // Compress the text
 454                 $flags = Revision::compressRevisionText( $text );
 455
 456                 // Update the text row
 457                 $dbw->update( 'text',
 458                         array( 'old_flags' => $flags, 'old_text' => $text ),
 459                         array( 'old_id' => $oldId ),
 460                         $fname, array( 'LIMIT' => 1 )
 461                 );
 462
 463                 // Remove it from the unfixed list and add it to the fixed list
 464                 unset( $this->errors['restore text'][$id] );
 465                 $this->errors['fixed'][$id] = true;
 466         }
 467 }
 468 ?>