includes/HistoryBlob.php

   1 <?php
   2
   3 /**
   4  * Base class for general text storage via the "object" flag in old_flags, or
   5  * two-part external storage URLs. Used for represent efficient concatenated
   6  * storage, and migration-related pointer objects.
   7  */
   8 interface HistoryBlob
   9 {
  10         /**
  11          * Adds an item of text, returns a stub object which points to the item.
  12          * You must call setLocation() on the stub object before storing it to the
  13          * database
  14          * Returns the key for getItem()
  15          */
  16         public function addItem( $text );
  17
  18         /**
  19          * Get item by key, or false if the key is not present
  20          */
  21         public function getItem( $key );
  22
  23         /**
  24          * Set the "default text"
  25          * This concept is an odd property of the current DB schema, whereby each text item has a revision
  26          * associated with it. The default text is the text of the associated revision. There may, however,
  27          * be other revisions in the same object.
  28          *
  29          * Default text is not required for two-part external storage URLs.
  30          */
  31         public function setText( $text );
  32
  33         /**
  34          * Get default text. This is called from Revision::getRevisionText()
  35          */
  36         function getText();
  37 }
  38
  39 /**
  40  * Concatenated gzip (CGZ) storage
  41  * Improves compression ratio by concatenating like objects before gzipping
  42  */
  43 class ConcatenatedGzipHistoryBlob implements HistoryBlob
  44 {
  45         public $mVersion = 0, $mCompressed = false, $mItems = array(), $mDefaultHash = '';
  46         public $mSize = 0;
  47         public $mMaxSize = 10000000;
  48         public $mMaxCount = 100;
  49
  50         /** Constructor */
  51         public function ConcatenatedGzipHistoryBlob() {
  52                 if ( !function_exists( 'gzdeflate' ) ) {
  53                         throw new MWException( "Need zlib support to read or write this kind of history object (ConcatenatedGzipHistoryBlob)\n" );
  54                 }
  55         }
  56
  57         public function addItem( $text ) {
  58                 $this->uncompress();
  59                 $hash = md5( $text );
  60                 if ( !isset( $this->mItems[$hash] ) ) {
  61                         $this->mItems[$hash] = $text;
  62                         $this->mSize += strlen( $text );
  63                 }
  64                 return $hash;
  65         }
  66
  67         public function getItem( $hash ) {
  68                 $this->uncompress();
  69                 if ( array_key_exists( $hash, $this->mItems ) ) {
  70                         return $this->mItems[$hash];
  71                 } else {
  72                         return false;
  73                 }
  74         }
  75
  76         public function setText( $text ) {
  77                 $this->uncompress();
  78                 $this->mDefaultHash = $this->addItem( $text );
  79         }
  80
  81         public function getText() {
  82                 $this->uncompress();
  83                 return $this->getItem( $this->mDefaultHash );
  84         }
  85
  86         /**
  87          * Remove an item
  88          */
  89         public function removeItem( $hash ) {
  90                 $this->mSize -= strlen( $this->mItems[$hash] );
  91                 unset( $this->mItems[$hash] );
  92         }
  93
  94         /**
  95          * Compress the bulk data in the object
  96          */
  97         public function compress() {
  98                 if ( !$this->mCompressed  ) {
  99                         $this->mItems = gzdeflate( serialize( $this->mItems ) );
 100                         $this->mCompressed = true;
 101                 }
 102         }
 103
 104         /**
 105          * Uncompress bulk data
 106          */
 107         public function uncompress() {
 108                 if ( $this->mCompressed ) {
 109                         $this->mItems = unserialize( gzinflate( $this->mItems ) );
 110                         $this->mCompressed = false;
 111                 }
 112         }
 113
 114
 115         function __sleep() {
 116                 $this->compress();
 117                 return array( 'mVersion', 'mCompressed', 'mItems', 'mDefaultHash' );
 118         }
 119
 120         function __wakeup() {
 121                 $this->uncompress();
 122         }
 123
 124         /**
 125          * Helper function for compression jobs
 126          * Returns true until the object is "full" and ready to be committed
 127          */
 128         public function isHappy() {
 129                 return $this->mSize < $this->mMaxSize
 130                         && count( $this->mItems ) < $this->mMaxCount;
 131         }
 132 }
 133
 134
 135 /**
 136  * One-step cache variable to hold base blobs; operations that
 137  * pull multiple revisions may often pull multiple times from
 138  * the same blob. By keeping the last-used one open, we avoid
 139  * redundant unserialization and decompression overhead.
 140  */
 141 global $wgBlobCache;
 142 $wgBlobCache = array();
 143
 144
 145 /**
 146  * Pointer object for an item within a CGZ blob stored in the text table.
 147  */
 148 class HistoryBlobStub {
 149         var $mOldId, $mHash, $mRef;
 150
 151         /**
 152          * @param string $hash The content hash of the text
 153          * @param integer $oldid The old_id for the CGZ object
 154          */
 155         function HistoryBlobStub( $hash = '', $oldid = 0 ) {
 156                 $this->mHash = $hash;
 157         }
 158
 159         /**
 160          * Sets the location (old_id) of the main object to which this object
 161          * points
 162          */
 163         function setLocation( $id ) {
 164                 $this->mOldId = $id;
 165         }
 166
 167         /**
 168          * Sets the location (old_id) of the referring object
 169          */
 170         function setReferrer( $id ) {
 171                 $this->mRef = $id;
 172         }
 173
 174         /**
 175          * Gets the location of the referring object
 176          */
 177         function getReferrer() {
 178                 return $this->mRef;
 179         }
 180
 181         function getText() {
 182                 $fname = 'HistoryBlobStub::getText';
 183                 global $wgBlobCache;
 184                 if( isset( $wgBlobCache[$this->mOldId] ) ) {
 185                         $obj = $wgBlobCache[$this->mOldId];
 186                 } else {
 187                         $dbr = wfGetDB( DB_SLAVE );
 188                         $row = $dbr->selectRow( 'text', array( 'old_flags', 'old_text' ), array( 'old_id' => $this->mOldId ) );
 189                         if( !$row ) {
 190                                 return false;
 191                         }
 192                         $flags = explode( ',', $row->old_flags );
 193                         if( in_array( 'external', $flags ) ) {
 194                                 $url=$row->old_text;
 195                                 @list( /* $proto */ ,$path)=explode('://',$url,2);
 196                                 if ($path=="") {
 197                                         wfProfileOut( $fname );
 198                                         return false;
 199                                 }
 200                                 $row->old_text=ExternalStore::fetchFromUrl($url);
 201
 202                         }
 203                         if( !in_array( 'object', $flags ) ) {
 204                                 return false;
 205                         }
 206
 207                         if( in_array( 'gzip', $flags ) ) {
 208                                 // This shouldn't happen, but a bug in the compress script
 209                                 // may at times gzip-compress a HistoryBlob object row.
 210                                 $obj = unserialize( gzinflate( $row->old_text ) );
 211                         } else {
 212                                 $obj = unserialize( $row->old_text );
 213                         }
 214
 215                         if( !is_object( $obj ) ) {
 216                                 // Correct for old double-serialization bug.
 217                                 $obj = unserialize( $obj );
 218                         }
 219
 220                         // Save this item for reference; if pulling many
 221                         // items in a row we'll likely use it again.
 222                         $obj->uncompress();
 223                         $wgBlobCache = array( $this->mOldId => $obj );
 224                 }
 225                 return $obj->getItem( $this->mHash );
 226         }
 227
 228         /**
 229          * Get the content hash
 230          */
 231         function getHash() {
 232                 return $this->mHash;
 233         }
 234 }
 235
 236
 237 /**
 238  * To speed up conversion from 1.4 to 1.5 schema, text rows can refer to the
 239  * leftover cur table as the backend. This avoids expensively copying hundreds
 240  * of megabytes of data during the conversion downtime.
 241  *
 242  * Serialized HistoryBlobCurStub objects will be inserted into the text table
 243  * on conversion if $wgFastSchemaUpgrades is set to true.
 244  */
 245 class HistoryBlobCurStub {
 246         var $mCurId;
 247
 248         /**
 249          * @param integer $curid The cur_id pointed to
 250          */
 251         function HistoryBlobCurStub( $curid = 0 ) {
 252                 $this->mCurId = $curid;
 253         }
 254
 255         /**
 256          * Sets the location (cur_id) of the main object to which this object
 257          * points
 258          */
 259         function setLocation( $id ) {
 260                 $this->mCurId = $id;
 261         }
 262
 263         function getText() {
 264                 $dbr = wfGetDB( DB_SLAVE );
 265                 $row = $dbr->selectRow( 'cur', array( 'cur_text' ), array( 'cur_id' => $this->mCurId ) );
 266                 if( !$row ) {
 267                         return false;
 268                 }
 269                 return $row->cur_text;
 270         }
 271 }
 272
 273 /**
 274  * Diff-based history compression
 275  * Requires xdiff 1.5+ and zlib
 276  */
 277 class DiffHistoryBlob implements HistoryBlob {
 278         /** Uncompressed item cache */
 279         var $mItems = array();
 280
 281         /** Total uncompressed size */
 282         var $mSize = 0;
 283
 284         /**
 285          * Array of diffs. If a diff D from A to B is notated D = B - A, and Z is
 286          * an empty string:
 287          *
 288          *              { item[map[i]] - item[map[i-1]]   where i > 0
 289          *    diff[i] = {
 290          *              { item[map[i]] - Z                where i = 0
 291          */
 292         var $mDiffs;
 293
 294         /** The diff map, see above */
 295         var $mDiffMap;
 296
 297         /**
 298          * The key for getText()
 299          */
 300         var $mDefaultKey;
 301
 302         /**
 303          * Compressed storage
 304          */
 305         var $mCompressed;
 306
 307         /**
 308          * True if the object is locked against further writes
 309          */
 310         var $mFrozen = false;
 311
 312         /**
 313          * The maximum uncompressed size before the object becomes sad
 314          * Should be less than max_allowed_packet
 315          */
 316         var $mMaxSize = 10000000;
 317
 318         /**
 319          * The maximum number of text items before the object becomes sad
 320          */
 321         var $mMaxCount = 100;
 322
 323         /** Constants from xdiff.h */
 324         const XDL_BDOP_INS = 1;
 325         const XDL_BDOP_CPY = 2;
 326         const XDL_BDOP_INSB = 3;
 327
 328         function __construct() {
 329                 if ( !function_exists( 'gzdeflate' ) ) {
 330                         throw new MWException( "Need zlib support to read or write DiffHistoryBlob\n" );
 331                 }
 332         }
 333
 334         function addItem( $text ) {
 335                 if ( $this->mFrozen ) {
 336                         throw new MWException( __METHOD__.": Cannot add more items after sleep/wakeup" );
 337                 }
 338
 339                 $this->mItems[] = $text;
 340                 $this->mSize += strlen( $text );
 341                 $this->mDiffs = null; // later
 342                 return count( $this->mItems ) - 1;
 343         }
 344
 345         function getItem( $key ) {
 346                 return $this->mItems[$key];
 347         }
 348
 349         function setText( $text ) {
 350                 $this->mDefaultKey = $this->addItem( $text );
 351         }
 352
 353         function getText() {
 354                 return $this->getItem( $this->mDefaultKey );
 355         }
 356
 357         function compress() {
 358                 if ( !function_exists( 'xdiff_string_rabdiff' ) ){
 359                         throw new MWException( "Need xdiff 1.5+ support to write DiffHistoryBlob\n" );
 360                 }
 361                 if ( isset( $this->mDiffs ) ) {
 362                         // Already compressed
 363                         return;
 364                 }
 365                 if ( !count( $this->mItems ) ) {
 366                         // Empty
 367                         return;
 368                 }
 369
 370                 // Create two diff sequences: one for main text and one for small text
 371                 $sequences = array(
 372                         'small' => array(
 373                                 'tail' => '',
 374                                 'diffs' => array(),
 375                                 'map' => array(),
 376                         ),
 377                         'main' => array(
 378                                 'tail' => '',
 379                                 'diffs' => array(),
 380                                 'map' => array(),
 381                         ),
 382                 );
 383                 $smallFactor = 0.5;
 384
 385                 for ( $i = 0; $i < count( $this->mItems ); $i++ ) {
 386                         $text = $this->mItems[$i];
 387                         if ( $i == 0 ) {
 388                                 $seqName = 'main';
 389                         } else {
 390                                 $mainTail = $sequences['main']['tail'];
 391                                 if ( strlen( $text ) < strlen( $mainTail ) * $smallFactor ) {
 392                                         $seqName = 'small';
 393                                 } else {
 394                                         $seqName = 'main';
 395                                 }
 396                         }
 397                         $seq =& $sequences[$seqName];
 398                         $tail = $seq['tail'];
 399                         $diff = $this->diff( $tail, $text );
 400                         $seq['diffs'][] = $diff;
 401                         $seq['map'][] = $i;
 402                         $seq['tail'] = $text;
 403                 }
 404                 unset( $seq ); // unlink dangerous alias
 405
 406                 // Knit the sequences together
 407                 $tail = '';
 408                 $this->mDiffs = array();
 409                 $this->mDiffMap = array();
 410                 foreach ( $sequences as $seq ) {
 411                         if ( !count( $seq['diffs'] ) ) {
 412                                 continue;
 413                         }
 414                         if ( $tail === '' ) {
 415                                 $this->mDiffs[] = $seq['diffs'][0];
 416                         } else {
 417                                 $head = $this->patch( '', $seq['diffs'][0] );
 418                                 $this->mDiffs[] = $this->diff( $tail, $head );
 419                         }
 420                         $this->mDiffMap[] = $seq['map'][0];
 421                         for ( $i = 1; $i < count( $seq['diffs'] ); $i++ ) {
 422                                 $this->mDiffs[] = $seq['diffs'][$i];
 423                                 $this->mDiffMap[] = $seq['map'][$i];
 424                         }
 425                         $tail = $seq['tail'];
 426                 }
 427         }
 428
 429         function diff( $t1, $t2 ) {
 430                 # Need to do a null concatenation with warnings off, due to bugs in the current version of xdiff
 431                 # "String is not zero-terminated"
 432                 wfSuppressWarnings();
 433                 $diff = xdiff_string_rabdiff( $t1, $t2 ) . '';
 434                 wfRestoreWarnings();
 435                 return $diff;
 436         }
 437
 438         function patch( $base, $diff ) {
 439                 if ( function_exists( 'xdiff_string_bpatch' ) ) {
 440                         wfSuppressWarnings();
 441                         $text = xdiff_string_bpatch( $base, $diff ) . '';
 442                         wfRestoreWarnings();
 443                         return $text;
 444                 }
 445
 446                 # Pure PHP implementation
 447
 448                 $header = unpack( 'Vofp/Vcsize', substr( $diff, 0, 8 ) );
 449
 450                 # Check the checksum if mhash is available
 451                 if ( extension_loaded( 'mhash' ) ) {
 452                         $ofp = mhash( MHASH_ADLER32, $base );
 453                         if ( $ofp !== substr( $diff, 0, 4 ) ) {
 454                                 wfDebug( __METHOD__. ": incorrect base checksum\n" );
 455                                 return false;
 456                         }
 457                 }
 458                 if ( $header['csize'] != strlen( $base ) ) {
 459                         wfDebug( __METHOD__. ": incorrect base length\n" );
 460                         return false;
 461                 }
 462
 463                 $p = 8;
 464                 $out = '';
 465                 while ( $p < strlen( $diff ) ) {
 466                         $x = unpack( 'Cop', substr( $diff, $p, 1 ) );
 467                         $op = $x['op'];
 468                         ++$p;
 469                         switch ( $op ) {
 470                         case self::XDL_BDOP_INS:
 471                                 $x = unpack( 'Csize', substr( $diff, $p, 1 ) );
 472                                 $p++;
 473                                 $out .= substr( $diff, $p, $x['size'] );
 474                                 $p += $x['size'];
 475                                 break;
 476                         case self::XDL_BDOP_INSB:
 477                                 $x = unpack( 'Vcsize', substr( $diff, $p, 4 ) );
 478                                 $p += 4;
 479                                 $out .= substr( $diff, $p, $x['csize'] );
 480                                 $p += $x['csize'];
 481                                 break;
 482                         case self::XDL_BDOP_CPY:
 483                                 $x = unpack( 'Voff/Vcsize', substr( $diff, $p, 8 ) );
 484                                 $p += 8;
 485                                 $out .= substr( $base, $x['off'], $x['csize'] );
 486                                 break;
 487                         default:
 488                                 wfDebug( __METHOD__.": invalid op\n" );
 489                                 return false;
 490                         }
 491                 }
 492                 return $out;
 493         }
 494
 495         function uncompress() {
 496                 if ( !$this->mDiffs ) {
 497                         return;
 498                 }
 499                 $tail = '';
 500                 for ( $diffKey = 0; $diffKey < count( $this->mDiffs ); $diffKey++ ) {
 501                         $textKey = $this->mDiffMap[$diffKey];
 502                         $text = $this->patch( $tail, $this->mDiffs[$diffKey] );
 503                         $this->mItems[$textKey] = $text;
 504                         $tail = $text;
 505                 }
 506         }
 507
 508         function __sleep() {
 509                 $this->compress();
 510                 if ( !count( $this->mItems ) ) {
 511                         // Empty object
 512                         $info = false;
 513                 } else {
 514                         // Take forward differences to improve the compression ratio for sequences
 515                         $map = '';
 516                         $prev = 0;
 517                         foreach ( $this->mDiffMap as $i ) {
 518                                 if ( $map !== '' ) {
 519                                         $map .= ',';
 520                                 }
 521                                 $map .= $i - $prev;
 522                                 $prev = $i;
 523                         }
 524                         $info = array(
 525                                 'diffs' => $this->mDiffs,
 526                                 'map' => $map
 527                         );
 528                 }
 529                 if ( isset( $this->mDefaultKey ) ) {
 530                         $info['default'] = $this->mDefaultKey;
 531                 }
 532                 $this->mCompressed = gzdeflate( serialize( $info ) );
 533                 return array( 'mCompressed' );
 534         }
 535
 536         function __wakeup() {
 537                 // addItem() doesn't work if mItems is partially filled from mDiffs
 538                 $this->mFrozen = true;
 539                 $info = unserialize( gzinflate( $this->mCompressed ) );
 540                 unset( $this->mCompressed );
 541
 542                 if ( !$info ) {
 543                         // Empty object
 544                         return;
 545                 }
 546
 547                 if ( isset( $info['default'] ) ) {
 548                         $this->mDefaultKey = $info['default'];
 549                 }
 550                 $this->mDiffs = $info['diffs'];
 551                 if ( isset( $info['base'] ) ) {
 552                         // Old format
 553                         $this->mDiffMap = range( 0, count( $this->mDiffs ) - 1 );
 554                         array_unshift( $this->mDiffs,
 555                                 pack( 'VVCV', 0, 0, self::XDL_BDOP_INSB, strlen( $info['base'] ) ) .
 556                                 $info['base'] );
 557                 } else {
 558                         // New format
 559                         $map = explode( ',', $info['map'] );
 560                         $cur = 0;
 561                         $this->mDiffMap = array();
 562                         foreach ( $map as $i ) {
 563                                 $cur += $i;
 564                                 $this->mDiffMap[] = $cur;
 565                         }
 566                 }
 567                 $this->uncompress();
 568         }
 569
 570         /**
 571          * Helper function for compression jobs
 572          * Returns true until the object is "full" and ready to be committed
 573          */
 574         function isHappy() {
 575                 return $this->mSize < $this->mMaxSize
 576                         && count( $this->mItems ) < $this->mMaxCount;
 577         }
 578
 579 }