maintenance/storage/fixBug20757.php

   1 <?php
   2
   3 require_once( dirname( __FILE__ ) . '/../Maintenance.php' );
   4
   5 class FixBug20757 extends Maintenance {
   6         var $batchSize = 10000;
   7         var $mapCache = array();
   8         var $mapCacheSize = 0;
   9         var $maxMapCacheSize = 1000000;
  10
  11         function __construct() {
  12                 parent::__construct();
  13                 $this->mDescription = 'Script to fix bug 20757 assuming that blob_tracking is intact';
  14                 $this->addOption( 'dry-run', 'Report only' );
  15                 $this->addOption( 'start', 'old_id to start at', false, true );
  16         }
  17
  18         function execute() {
  19                 $dbr = wfGetDB( DB_SLAVE );
  20                 $dbw = wfGetDB( DB_MASTER );
  21
  22                 $dryRun = $this->getOption( 'dry-run' );
  23                 if ( $dryRun ) {
  24                         print "Dry run only.\n";
  25                 }
  26
  27                 $startId = $this->getOption( 'start', 0 );
  28                 $numGood = 0;
  29                 $numFixed = 0;
  30                 $numBad = 0;
  31
  32                 $totalRevs = $dbr->selectField( 'text', 'MAX(old_id)', false, __METHOD__ );
  33
  34                 if ( $dbr->getType() == 'mysql'
  35                         && version_compare( $dbr->getServerVersion(), '4.1.0', '>=' ) )
  36                 {
  37                         // In MySQL 4.1+, the binary field old_text has a non-working LOWER() function
  38                         $lowerLeft = 'LOWER(CONVERT(LEFT(old_text,22) USING latin1))';
  39                 } else {
  40                         // No CONVERT() in MySQL 4.0
  41                         $lowerLeft = 'LOWER(LEFT(old_text,22))';
  42                 }
  43
  44                 while ( true ) {
  45                         print "ID: $startId / $totalRevs\r";
  46
  47                         $res = $dbr->select(
  48                                 'text',
  49                                 array( 'old_id', 'old_flags', 'old_text' ),
  50                                 array(
  51                                         'old_id > ' . intval( $startId ),
  52                                         'old_flags LIKE \'%object%\' AND old_flags NOT LIKE \'%external%\'',
  53                                         "$lowerLeft = 'o:15:\"historyblobstub\"'",
  54                                 ),
  55                                 __METHOD__,
  56                                 array(
  57                                         'ORDER BY' => 'old_id',
  58                                         'LIMIT' => $this->batchSize,
  59                                 )
  60                         );
  61
  62                         if ( !$res->numRows() ) {
  63                                 break;
  64                         }
  65
  66                         $secondaryIds = array();
  67                         $stubs = array();
  68
  69                         foreach ( $res as $row ) {
  70                                 $startId = $row->old_id;
  71
  72                                 // Basic sanity checks
  73                                 $obj = unserialize( $row->old_text );
  74                                 if ( $obj === false ) {
  75                                         print "{$row->old_id}: unrecoverable: cannot unserialize\n";
  76                                         ++$numBad;
  77                                         continue;
  78                                 }
  79
  80                                 if ( !is_object( $obj ) ) {
  81                                         print "{$row->old_id}: unrecoverable: unserialized to type " .
  82                                                 gettype( $obj ) . ", possible double-serialization\n";
  83                                         ++$numBad;
  84                                         continue;
  85                                 }
  86
  87                                 if ( strtolower( get_class( $obj ) ) !== 'historyblobstub' ) {
  88                                         print "{$row->old_id}: unrecoverable: unexpected object class " .
  89                                                 get_class( $obj ) . "\n";
  90                                         ++$numBad;
  91                                         continue;
  92                                 }
  93
  94                                 // Process flags
  95                                 $flags = explode( ',', $row->old_flags );
  96                                 if ( in_array( 'utf-8', $flags ) || in_array( 'utf8', $flags ) ) {
  97                                         $legacyEncoding = false;
  98                                 } else {
  99                                         $legacyEncoding = true;
 100                                 }
 101
 102                                 // Queue the stub for future batch processing
 103                                 $id = intval( $obj->mOldId );
 104                                 $secondaryIds[] = $id;
 105                                 $stubs[$row->old_id] = array(
 106                                         'legacyEncoding' => $legacyEncoding,
 107                                         'secondaryId' => $id,
 108                                         'hash' => $obj->mHash,
 109                                 );
 110                         }
 111
 112                         $secondaryIds = array_unique( $secondaryIds );
 113
 114                         if ( !count( $secondaryIds ) ) {
 115                                 continue;
 116                         }
 117
 118                         // Run the batch query on blob_tracking
 119                         $res = $dbr->select(
 120                                 'blob_tracking',
 121                                 '*',
 122                                 array(
 123                                         'bt_text_id' => $secondaryIds,
 124                                 ),
 125                                 __METHOD__
 126                         );
 127                         $trackedBlobs = array();
 128                         foreach ( $res as $row ) {
 129                                 $trackedBlobs[$row->bt_text_id] = $row;
 130                         }
 131
 132                         // Process the stubs
 133                         foreach ( $stubs as $primaryId => $stub ) {
 134                                 $secondaryId = $stub['secondaryId'];
 135                                 if ( !isset( $trackedBlobs[$secondaryId] ) ) {
 136                                         // No tracked blob. Work out what went wrong
 137                                         $secondaryRow = $dbr->selectRow(
 138                                                 'text',
 139                                                 array( 'old_flags', 'old_text' ),
 140                                                 array( 'old_id' => $secondaryId ),
 141                                                 __METHOD__
 142                                         );
 143                                         if ( !$secondaryRow ) {
 144                                                 print "$primaryId: unrecoverable: secondary row is missing\n";
 145                                                 ++$numBad;
 146                                         } elseif ( $this->isUnbrokenStub( $stub, $secondaryRow ) ) {
 147                                                 // Not broken yet, and not in the tracked clusters so it won't get
 148                                                 // broken by the current RCT run.
 149                                                 ++$numGood;
 150                                         } elseif ( strpos( $secondaryRow->old_flags, 'external' ) !== false ) {
 151                                                 print "$primaryId: unrecoverable: secondary gone to {$secondaryRow->old_text}\n";
 152                                                 ++$numBad;
 153                                         } else {
 154                                                 print "$primaryId: unrecoverable: miscellaneous corruption of secondary row\n";
 155                                                 ++$numBad;
 156                                         }
 157                                         unset( $stubs[$primaryId] );
 158                                         continue;
 159                                 }
 160                                 $trackRow = $trackedBlobs[$secondaryId];
 161
 162                                 // Check that the specified text really is available in the tracked source row
 163                                 $url = "DB://{$trackRow->bt_cluster}/{$trackRow->bt_blob_id}/{$stub['hash']}";
 164                                 $text = ExternalStore::fetchFromURL( $url );
 165                                 if ( $text === false ) {
 166                                         print "$primaryId: unrecoverable: source text missing\n";
 167                                         ++$numBad;
 168                                         unset( $stubs[$primaryId] );
 169                                         continue;
 170                                 }
 171                                 if ( md5( $text ) !== $stub['hash'] ) {
 172                                         print "$primaryId: unrecoverable: content hashes do not match\n";
 173                                         ++$numBad;
 174                                         unset( $stubs[$primaryId] );
 175                                         continue;
 176                                 }
 177
 178                                 // Find the page_id and rev_id
 179                                 // The page is probably the same as the page of the secondary row
 180                                 $pageId = intval( $trackRow->bt_page );
 181                                 if ( !$pageId ) {
 182                                         $revId = $pageId = 0;
 183                                 } else {
 184                                         $revId = $this->findTextIdInPage( $pageId, $primaryId );
 185                                         if ( !$revId ) {
 186                                                 // Actually an orphan
 187                                                 $pageId = $revId = 0;
 188                                         }
 189                                 }
 190
 191                                 $newFlags = $stub['legacyEncoding'] ? 'external' : 'external,utf-8';
 192
 193                                 if ( !$dryRun ) {
 194                                         // Reset the text row to point to the original copy
 195                                         $dbw->begin();
 196                                         $dbw->update(
 197                                                 'text',
 198                                                 // SET
 199                                                 array(
 200                                                         'old_flags' => $newFlags,
 201                                                         'old_text' => $url
 202                                                 ),
 203                                                 // WHERE
 204                                                 array( 'old_id' => $primaryId ),
 205                                                 __METHOD__
 206                                         );
 207
 208                                         // Add a blob_tracking row so that the new reference can be recompressed
 209                                         // without needing to run trackBlobs.php again
 210                                         $dbw->insert( 'blob_tracking',
 211                                                 array(
 212                                                         'bt_page' => $pageId,
 213                                                         'bt_rev_id' => $revId,
 214                                                         'bt_text_id' => $primaryId,
 215                                                         'bt_cluster' => $trackRow->bt_cluster,
 216                                                         'bt_blob_id' => $trackRow->bt_blob_id,
 217                                                         'bt_cgz_hash' => $stub['hash'],
 218                                                         'bt_new_url' => null,
 219                                                         'bt_moved' => 0,
 220                                                 ),
 221                                                 __METHOD__
 222                                         );
 223                                         $dbw->commit();
 224                                         $this->waitForSlaves();
 225                                 }
 226
 227                                 print "$primaryId: resolved to $url\n";
 228                                 ++$numFixed;
 229                         }
 230                 }
 231
 232                 print "\n";
 233                 print "Fixed: $numFixed\n";
 234                 print "Unrecoverable: $numBad\n";
 235                 print "Good stubs: $numGood\n";
 236         }
 237
 238         function waitForSlaves() {
 239                 static $iteration = 0;
 240                 ++$iteration;
 241                 if ( ++$iteration > 50 == 0 ) {
 242                         wfWaitForSlaves( 5 );
 243                         $iteration = 0;
 244                 }
 245         }
 246
 247         function findTextIdInPage( $pageId, $textId ) {
 248                 $ids = $this->getRevTextMap( $pageId );
 249                 if ( !isset( $ids[$textId] ) ) {
 250                         return null;
 251                 } else {
 252                         return $ids[$textId];
 253                 }
 254         }
 255
 256         function getRevTextMap( $pageId ) {
 257                 if ( !isset( $this->mapCache[$pageId] ) ) {
 258                         // Limit cache size
 259                         while ( $this->mapCacheSize > $this->maxMapCacheSize ) {
 260                                 $key = key( $this->mapCache );
 261                                 $this->mapCacheSize -= count( $this->mapCache[$key] );
 262                                 unset( $this->mapCache[$key] );
 263                         }
 264
 265                         $dbr = wfGetDB( DB_SLAVE );
 266                         $map = array();
 267                         $res = $dbr->select( 'revision',
 268                                 array( 'rev_id', 'rev_text_id' ),
 269                                 array( 'rev_page' => $pageId ),
 270                                 __METHOD__
 271                         );
 272                         foreach ( $res as $row ) {
 273                                 $map[$row->rev_text_id] = $row->rev_id;
 274                         }
 275                         $this->mapCache[$pageId] = $map;
 276                         $this->mapCacheSize += count( $map );
 277                 }
 278                 return $this->mapCache[$pageId];
 279         }
 280
 281         /**
 282          * This is based on part of HistoryBlobStub::getText().
 283          * Determine if the text can be retrieved from the row in the normal way.
 284          */
 285         function isUnbrokenStub( $stub, $secondaryRow ) {
 286                 $flags = explode( ',', $secondaryRow->old_flags );
 287                 $text = $secondaryRow->old_text;
 288                 if ( in_array( 'external', $flags ) ) {
 289                         $url = $text;
 290                         @list( /* $proto */ , $path ) = explode( '://', $url, 2 );
 291                         if ( $path == "" ) {
 292                                 return false;
 293                         }
 294                         $text = ExternalStore::fetchFromUrl( $url );
 295                 }
 296                 if ( !in_array( 'object', $flags ) ) {
 297                         return false;
 298                 }
 299
 300                 if ( in_array( 'gzip', $flags ) ) {
 301                         $obj = unserialize( gzinflate( $text ) );
 302                 } else {
 303                         $obj = unserialize( $text );
 304                 }
 305
 306                 if ( !is_object( $obj ) ) {
 307                         // Correct for old double-serialization bug.
 308                         $obj = unserialize( $obj );
 309                 }
 310
 311                 if ( !is_object( $obj ) ) {
 312                         return false;
 313                 }
 314
 315                 $obj->uncompress();
 316                 $text = $obj->getItem( $stub['hash'] );
 317                 return $text !== false;
 318         }
 319 }
 320
 321 $maintClass = 'FixBug20757';
 322 require_once( DO_MAINTENANCE );
 323