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          *
  15          * @param $text string
  16          *
  17          * @return String: the key for getItem()
  18          */
  19         function addItem( $text );
  20
  21         /**
  22          * Get item by key, or false if the key is not present
  23          *
  24          * @param $key string
  25          *
  26          * @return String or false
  27          */
  28         function getItem( $key );
  29
  30         /**
  31          * Set the "default text"
  32          * This concept is an odd property of the current DB schema, whereby each text item has a revision
  33          * associated with it. The default text is the text of the associated revision. There may, however,
  34          * be other revisions in the same object.
  35          *
  36          * Default text is not required for two-part external storage URLs.
  37          *
  38          * @param $text string
  39          */
  40         function setText( $text );
  41
  42         /**
  43          * Get default text. This is called from Revision::getRevisionText()
  44          *
  45          * @return String
  46          */
  47         function getText();
  48 }
  49
  50 /**
  51  * Concatenated gzip (CGZ) storage
  52  * Improves compression ratio by concatenating like objects before gzipping
  53  */
  54 class ConcatenatedGzipHistoryBlob implements HistoryBlob
  55 {
  56         public $mVersion = 0, $mCompressed = false, $mItems = array(), $mDefaultHash = '';
  57         public $mSize = 0;
  58         public $mMaxSize = 10000000;
  59         public $mMaxCount = 100;
  60
  61         /** Constructor */
  62         public function __construct() {
  63                 if ( !function_exists( 'gzdeflate' ) ) {
  64                         throw new MWException( "Need zlib support to read or write this kind of history object (ConcatenatedGzipHistoryBlob)\n" );
  65                 }
  66         }
  67
  68         /**
  69          * @param $text string
  70          * @return string
  71          */
  72         public function addItem( $text ) {
  73                 $this->uncompress();
  74                 $hash = md5( $text );
  75                 if ( !isset( $this->mItems[$hash] ) ) {
  76                         $this->mItems[$hash] = $text;
  77                         $this->mSize += strlen( $text );
  78                 }
  79                 return $hash;
  80         }
  81
  82         /**
  83          * @param $hash string
  84          * @return array|bool
  85          */
  86         public function getItem( $hash ) {
  87                 $this->uncompress();
  88                 if ( array_key_exists( $hash, $this->mItems ) ) {
  89                         return $this->mItems[$hash];
  90                 } else {
  91                         return false;
  92                 }
  93         }
  94
  95         /**
  96          * @param $text string
  97          * @return void
  98          */
  99         public function setText( $text ) {
 100                 $this->uncompress();
 101                 $this->mDefaultHash = $this->addItem( $text );
 102         }
 103
 104         /**
 105          * @return array|bool
 106          */
 107         public function getText() {
 108                 $this->uncompress();
 109                 return $this->getItem( $this->mDefaultHash );
 110         }
 111
 112         /**
 113          * Remove an item
 114          *
 115          * @param $hash string
 116          */
 117         public function removeItem( $hash ) {
 118                 $this->mSize -= strlen( $this->mItems[$hash] );
 119                 unset( $this->mItems[$hash] );
 120         }
 121
 122         /**
 123          * Compress the bulk data in the object
 124          */
 125         public function compress() {
 126                 if ( !$this->mCompressed  ) {
 127                         $this->mItems = gzdeflate( serialize( $this->mItems ) );
 128                         $this->mCompressed = true;
 129                 }
 130         }
 131
 132         /**
 133          * Uncompress bulk data
 134          */
 135         public function uncompress() {
 136                 if ( $this->mCompressed ) {
 137                         $this->mItems = unserialize( gzinflate( $this->mItems ) );
 138                         $this->mCompressed = false;
 139                 }
 140         }
 141
 142         /**
 143          * @return array
 144          */
 145         function __sleep() {
 146                 $this->compress();
 147                 return array( 'mVersion', 'mCompressed', 'mItems', 'mDefaultHash' );
 148         }
 149
 150         function __wakeup() {
 151                 $this->uncompress();
 152         }
 153
 154         /**
 155          * Helper function for compression jobs
 156          * Returns true until the object is "full" and ready to be committed
 157          *
 158          * @return bool
 159          */
 160         public function isHappy() {
 161                 return $this->mSize < $this->mMaxSize
 162                         && count( $this->mItems ) < $this->mMaxCount;
 163         }
 164 }
 165
 166
 167 /**
 168  * Pointer object for an item within a CGZ blob stored in the text table.
 169  */
 170 class HistoryBlobStub {
 171         /**
 172          * One-step cache variable to hold base blobs; operations that
 173          * pull multiple revisions may often pull multiple times from
 174          * the same blob. By keeping the last-used one open, we avoid
 175          * redundant unserialization and decompression overhead.
 176          */
 177         protected static $blobCache = array();
 178
 179         var $mOldId, $mHash, $mRef;
 180
 181         /**
 182          * @param $hash Strng: the content hash of the text
 183          * @param $oldid Integer: the old_id for the CGZ object
 184          */
 185         function __construct( $hash = '', $oldid = 0 ) {
 186                 $this->mHash = $hash;
 187         }
 188
 189         /**
 190          * Sets the location (old_id) of the main object to which this object
 191          * points
 192          */
 193         function setLocation( $id ) {
 194                 $this->mOldId = $id;
 195         }
 196
 197         /**
 198          * Sets the location (old_id) of the referring object
 199          */
 200         function setReferrer( $id ) {
 201                 $this->mRef = $id;
 202         }
 203
 204         /**
 205          * Gets the location of the referring object
 206          */
 207         function getReferrer() {
 208                 return $this->mRef;
 209         }
 210
 211         /**
 212          * @return string
 213          */
 214         function getText() {
 215                 $fname = 'HistoryBlobStub::getText';
 216
 217                 if( isset( self::$blobCache[$this->mOldId] ) ) {
 218                         $obj = self::$blobCache[$this->mOldId];
 219                 } else {
 220                         $dbr = wfGetDB( DB_SLAVE );
 221                         $row = $dbr->selectRow( 'text', array( 'old_flags', 'old_text' ), array( 'old_id' => $this->mOldId ) );
 222                         if( !$row ) {
 223                                 return false;
 224                         }
 225                         $flags = explode( ',', $row->old_flags );
 226                         if( in_array( 'external', $flags ) ) {
 227                                 $url=$row->old_text;
 228                                 @list( /* $proto */ ,$path)=explode('://',$url,2);
 229                                 if ( $path == "" ) {
 230                                         wfProfileOut( $fname );
 231                                         return false;
 232                                 }
 233                                 $row->old_text = ExternalStore::fetchFromUrl($url);
 234
 235                         }
 236                         if( !in_array( 'object', $flags ) ) {
 237                                 return false;
 238                         }
 239
 240                         if( in_array( 'gzip', $flags ) ) {
 241                                 // This shouldn't happen, but a bug in the compress script
 242                                 // may at times gzip-compress a HistoryBlob object row.
 243                                 $obj = unserialize( gzinflate( $row->old_text ) );
 244                         } else {
 245                                 $obj = unserialize( $row->old_text );
 246                         }
 247
 248                         if( !is_object( $obj ) ) {
 249                                 // Correct for old double-serialization bug.
 250                                 $obj = unserialize( $obj );
 251                         }
 252
 253                         // Save this item for reference; if pulling many
 254                         // items in a row we'll likely use it again.
 255                         $obj->uncompress();
 256                         self::$blobCache = array( $this->mOldId => $obj );
 257                 }
 258                 return $obj->getItem( $this->mHash );
 259         }
 260
 261         /**
 262          * Get the content hash
 263          *
 264          * @return string
 265          */
 266         function getHash() {
 267                 return $this->mHash;
 268         }
 269 }
 270
 271
 272 /**
 273  * To speed up conversion from 1.4 to 1.5 schema, text rows can refer to the
 274  * leftover cur table as the backend. This avoids expensively copying hundreds
 275  * of megabytes of data during the conversion downtime.
 276  *
 277  * Serialized HistoryBlobCurStub objects will be inserted into the text table
 278  * on conversion if $wgFastSchemaUpgrades is set to true.
 279  */
 280 class HistoryBlobCurStub {
 281         var $mCurId;
 282
 283         /**
 284          * @param $curid Integer: the cur_id pointed to
 285          */
 286         function __construct( $curid = 0 ) {
 287                 $this->mCurId = $curid;
 288         }
 289
 290         /**
 291          * Sets the location (cur_id) of the main object to which this object
 292          * points
 293          */
 294         function setLocation( $id ) {
 295                 $this->mCurId = $id;
 296         }
 297
 298         /**
 299          * @return string|false
 300          */
 301         function getText() {
 302                 $dbr = wfGetDB( DB_SLAVE );
 303                 $row = $dbr->selectRow( 'cur', array( 'cur_text' ), array( 'cur_id' => $this->mCurId ) );
 304                 if( !$row ) {
 305                         return false;
 306                 }
 307                 return $row->cur_text;
 308         }
 309 }
 310
 311 /**
 312  * Diff-based history compression
 313  * Requires xdiff 1.5+ and zlib
 314  */
 315 class DiffHistoryBlob implements HistoryBlob {
 316         /** Uncompressed item cache */
 317         var $mItems = array();
 318
 319         /** Total uncompressed size */
 320         var $mSize = 0;
 321
 322         /**
 323          * Array of diffs. If a diff D from A to B is notated D = B - A, and Z is
 324          * an empty string:
 325          *
 326          *              { item[map[i]] - item[map[i-1]]   where i > 0
 327          *    diff[i] = {
 328          *              { item[map[i]] - Z                where i = 0
 329          */
 330         var $mDiffs;
 331
 332         /** The diff map, see above */
 333         var $mDiffMap;
 334
 335         /**
 336          * The key for getText()
 337          */
 338         var $mDefaultKey;
 339
 340         /**
 341          * Compressed storage
 342          */
 343         var $mCompressed;
 344
 345         /**
 346          * True if the object is locked against further writes
 347          */
 348         var $mFrozen = false;
 349
 350         /**
 351          * The maximum uncompressed size before the object becomes sad
 352          * Should be less than max_allowed_packet
 353          */
 354         var $mMaxSize = 10000000;
 355
 356         /**
 357          * The maximum number of text items before the object becomes sad
 358          */
 359         var $mMaxCount = 100;
 360
 361         /** Constants from xdiff.h */
 362         const XDL_BDOP_INS = 1;
 363         const XDL_BDOP_CPY = 2;
 364         const XDL_BDOP_INSB = 3;
 365
 366         function __construct() {
 367                 if ( !function_exists( 'gzdeflate' ) ) {
 368                         throw new MWException( "Need zlib support to read or write DiffHistoryBlob\n" );
 369                 }
 370         }
 371
 372         /**
 373          * @throws MWException
 374          * @param $text string
 375          * @return int
 376          */
 377         function addItem( $text ) {
 378                 if ( $this->mFrozen ) {
 379                         throw new MWException( __METHOD__.": Cannot add more items after sleep/wakeup" );
 380                 }
 381
 382                 $this->mItems[] = $text;
 383                 $this->mSize += strlen( $text );
 384                 $this->mDiffs = null; // later
 385                 return count( $this->mItems ) - 1;
 386         }
 387
 388         /**
 389          * @param $key string
 390          * @return string
 391          */
 392         function getItem( $key ) {
 393                 return $this->mItems[$key];
 394         }
 395
 396         /**
 397          * @param $text string
 398          */
 399         function setText( $text ) {
 400                 $this->mDefaultKey = $this->addItem( $text );
 401         }
 402
 403         /**
 404          * @return string
 405          */
 406         function getText() {
 407                 return $this->getItem( $this->mDefaultKey );
 408         }
 409
 410         /**
 411          * @throws MWException
 412          */
 413         function compress() {
 414                 if ( !function_exists( 'xdiff_string_rabdiff' ) ){
 415                         throw new MWException( "Need xdiff 1.5+ support to write DiffHistoryBlob\n" );
 416                 }
 417                 if ( isset( $this->mDiffs ) ) {
 418                         // Already compressed
 419                         return;
 420                 }
 421                 if ( !count( $this->mItems ) ) {
 422                         // Empty
 423                         return;
 424                 }
 425
 426                 // Create two diff sequences: one for main text and one for small text
 427                 $sequences = array(
 428                         'small' => array(
 429                                 'tail' => '',
 430                                 'diffs' => array(),
 431                                 'map' => array(),
 432                         ),
 433                         'main' => array(
 434                                 'tail' => '',
 435                                 'diffs' => array(),
 436                                 'map' => array(),
 437                         ),
 438                 );
 439                 $smallFactor = 0.5;
 440
 441                 for ( $i = 0; $i < count( $this->mItems ); $i++ ) {
 442                         $text = $this->mItems[$i];
 443                         if ( $i == 0 ) {
 444                                 $seqName = 'main';
 445                         } else {
 446                                 $mainTail = $sequences['main']['tail'];
 447                                 if ( strlen( $text ) < strlen( $mainTail ) * $smallFactor ) {
 448                                         $seqName = 'small';
 449                                 } else {
 450                                         $seqName = 'main';
 451                                 }
 452                         }
 453                         $seq =& $sequences[$seqName];
 454                         $tail = $seq['tail'];
 455                         $diff = $this->diff( $tail, $text );
 456                         $seq['diffs'][] = $diff;
 457                         $seq['map'][] = $i;
 458                         $seq['tail'] = $text;
 459                 }
 460                 unset( $seq ); // unlink dangerous alias
 461
 462                 // Knit the sequences together
 463                 $tail = '';
 464                 $this->mDiffs = array();
 465                 $this->mDiffMap = array();
 466                 foreach ( $sequences as $seq ) {
 467                         if ( !count( $seq['diffs'] ) ) {
 468                                 continue;
 469                         }
 470                         if ( $tail === '' ) {
 471                                 $this->mDiffs[] = $seq['diffs'][0];
 472                         } else {
 473                                 $head = $this->patch( '', $seq['diffs'][0] );
 474                                 $this->mDiffs[] = $this->diff( $tail, $head );
 475                         }
 476                         $this->mDiffMap[] = $seq['map'][0];
 477                         for ( $i = 1; $i < count( $seq['diffs'] ); $i++ ) {
 478                                 $this->mDiffs[] = $seq['diffs'][$i];
 479                                 $this->mDiffMap[] = $seq['map'][$i];
 480                         }
 481                         $tail = $seq['tail'];
 482                 }
 483         }
 484
 485         function diff( $t1, $t2 ) {
 486                 # Need to do a null concatenation with warnings off, due to bugs in the current version of xdiff
 487                 # "String is not zero-terminated"
 488                 wfSuppressWarnings();
 489                 $diff = xdiff_string_rabdiff( $t1, $t2 ) . '';
 490                 wfRestoreWarnings();
 491                 return $diff;
 492         }
 493
 494         function patch( $base, $diff ) {
 495                 if ( function_exists( 'xdiff_string_bpatch' ) ) {
 496                         wfSuppressWarnings();
 497                         $text = xdiff_string_bpatch( $base, $diff ) . '';
 498                         wfRestoreWarnings();
 499                         return $text;
 500                 }
 501
 502                 # Pure PHP implementation
 503
 504                 $header = unpack( 'Vofp/Vcsize', substr( $diff, 0, 8 ) );
 505
 506                 # Check the checksum if mhash is available
 507                 if ( extension_loaded( 'mhash' ) ) {
 508                         $ofp = mhash( MHASH_ADLER32, $base );
 509                         if ( $ofp !== substr( $diff, 0, 4 ) ) {
 510                                 wfDebug( __METHOD__. ": incorrect base checksum\n" );
 511                                 return false;
 512                         }
 513                 }
 514                 if ( $header['csize'] != strlen( $base ) ) {
 515                         wfDebug( __METHOD__. ": incorrect base length\n" );
 516                         return false;
 517                 }
 518
 519                 $p = 8;
 520                 $out = '';
 521                 while ( $p < strlen( $diff ) ) {
 522                         $x = unpack( 'Cop', substr( $diff, $p, 1 ) );
 523                         $op = $x['op'];
 524                         ++$p;
 525                         switch ( $op ) {
 526                         case self::XDL_BDOP_INS:
 527                                 $x = unpack( 'Csize', substr( $diff, $p, 1 ) );
 528                                 $p++;
 529                                 $out .= substr( $diff, $p, $x['size'] );
 530                                 $p += $x['size'];
 531                                 break;
 532                         case self::XDL_BDOP_INSB:
 533                                 $x = unpack( 'Vcsize', substr( $diff, $p, 4 ) );
 534                                 $p += 4;
 535                                 $out .= substr( $diff, $p, $x['csize'] );
 536                                 $p += $x['csize'];
 537                                 break;
 538                         case self::XDL_BDOP_CPY:
 539                                 $x = unpack( 'Voff/Vcsize', substr( $diff, $p, 8 ) );
 540                                 $p += 8;
 541                                 $out .= substr( $base, $x['off'], $x['csize'] );
 542                                 break;
 543                         default:
 544                                 wfDebug( __METHOD__.": invalid op\n" );
 545                                 return false;
 546                         }
 547                 }
 548                 return $out;
 549         }
 550
 551         function uncompress() {
 552                 if ( !$this->mDiffs ) {
 553                         return;
 554                 }
 555                 $tail = '';
 556                 for ( $diffKey = 0; $diffKey < count( $this->mDiffs ); $diffKey++ ) {
 557                         $textKey = $this->mDiffMap[$diffKey];
 558                         $text = $this->patch( $tail, $this->mDiffs[$diffKey] );
 559                         $this->mItems[$textKey] = $text;
 560                         $tail = $text;
 561                 }
 562         }
 563
 564         function __sleep() {
 565                 $this->compress();
 566                 if ( !count( $this->mItems ) ) {
 567                         // Empty object
 568                         $info = false;
 569                 } else {
 570                         // Take forward differences to improve the compression ratio for sequences
 571                         $map = '';
 572                         $prev = 0;
 573                         foreach ( $this->mDiffMap as $i ) {
 574                                 if ( $map !== '' ) {
 575                                         $map .= ',';
 576                                 }
 577                                 $map .= $i - $prev;
 578                                 $prev = $i;
 579                         }
 580                         $info = array(
 581                                 'diffs' => $this->mDiffs,
 582                                 'map' => $map
 583                         );
 584                 }
 585                 if ( isset( $this->mDefaultKey ) ) {
 586                         $info['default'] = $this->mDefaultKey;
 587                 }
 588                 $this->mCompressed = gzdeflate( serialize( $info ) );
 589                 return array( 'mCompressed' );
 590         }
 591
 592         function __wakeup() {
 593                 // addItem() doesn't work if mItems is partially filled from mDiffs
 594                 $this->mFrozen = true;
 595                 $info = unserialize( gzinflate( $this->mCompressed ) );
 596                 unset( $this->mCompressed );
 597
 598                 if ( !$info ) {
 599                         // Empty object
 600                         return;
 601                 }
 602
 603                 if ( isset( $info['default'] ) ) {
 604                         $this->mDefaultKey = $info['default'];
 605                 }
 606                 $this->mDiffs = $info['diffs'];
 607                 if ( isset( $info['base'] ) ) {
 608                         // Old format
 609                         $this->mDiffMap = range( 0, count( $this->mDiffs ) - 1 );
 610                         array_unshift( $this->mDiffs,
 611                                 pack( 'VVCV', 0, 0, self::XDL_BDOP_INSB, strlen( $info['base'] ) ) .
 612                                 $info['base'] );
 613                 } else {
 614                         // New format
 615                         $map = explode( ',', $info['map'] );
 616                         $cur = 0;
 617                         $this->mDiffMap = array();
 618                         foreach ( $map as $i ) {
 619                                 $cur += $i;
 620                                 $this->mDiffMap[] = $cur;
 621                         }
 622                 }
 623                 $this->uncompress();
 624         }
 625
 626         /**
 627          * Helper function for compression jobs
 628          * Returns true until the object is "full" and ready to be committed
 629          */
 630         function isHappy() {
 631                 return $this->mSize < $this->mMaxSize
 632                         && count( $this->mItems ) < $this->mMaxCount;
 633         }
 634
 635 }