includes/HistoryBlob.php

   1 <?php
   2 /**
   3  * Efficient concatenated text storage.
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  * http://www.gnu.org/copyleft/gpl.html
  19  *
  20  * @file
  21  */
  22
  23 /**
  24  * Base class for general text storage via the "object" flag in old_flags, or
  25  * two-part external storage URLs. Used for represent efficient concatenated
  26  * storage, and migration-related pointer objects.
  27  */
  28 interface HistoryBlob
  29 {
  30         /**
  31          * Adds an item of text, returns a stub object which points to the item.
  32          * You must call setLocation() on the stub object before storing it to the
  33          * database
  34          *
  35          * @param $text string
  36          *
  37          * @return String: the key for getItem()
  38          */
  39         function addItem( $text );
  40
  41         /**
  42          * Get item by key, or false if the key is not present
  43          *
  44          * @param $key string
  45          *
  46          * @return String or false
  47          */
  48         function getItem( $key );
  49
  50         /**
  51          * Set the "default text"
  52          * This concept is an odd property of the current DB schema, whereby each text item has a revision
  53          * associated with it. The default text is the text of the associated revision. There may, however,
  54          * be other revisions in the same object.
  55          *
  56          * Default text is not required for two-part external storage URLs.
  57          *
  58          * @param $text string
  59          */
  60         function setText( $text );
  61
  62         /**
  63          * Get default text. This is called from Revision::getRevisionText()
  64          *
  65          * @return String
  66          */
  67         function getText();
  68 }
  69
  70 /**
  71  * Concatenated gzip (CGZ) storage
  72  * Improves compression ratio by concatenating like objects before gzipping
  73  */
  74 class ConcatenatedGzipHistoryBlob implements HistoryBlob
  75 {
  76         public $mVersion = 0, $mCompressed = false, $mItems = array(), $mDefaultHash = '';
  77         public $mSize = 0;
  78         public $mMaxSize = 10000000;
  79         public $mMaxCount = 100;
  80
  81         /** Constructor */
  82         public function __construct() {
  83                 if ( !function_exists( 'gzdeflate' ) ) {
  84                         throw new MWException( "Need zlib support to read or write this kind of history object (ConcatenatedGzipHistoryBlob)\n" );
  85                 }
  86         }
  87
  88         /**
  89          * @param $text string
  90          * @return string
  91          */
  92         public function addItem( $text ) {
  93                 $this->uncompress();
  94                 $hash = md5( $text );
  95                 if ( !isset( $this->mItems[$hash] ) ) {
  96                         $this->mItems[$hash] = $text;
  97                         $this->mSize += strlen( $text );
  98                 }
  99                 return $hash;
 100         }
 101
 102         /**
 103          * @param $hash string
 104          * @return array|bool
 105          */
 106         public function getItem( $hash ) {
 107                 $this->uncompress();
 108                 if ( array_key_exists( $hash, $this->mItems ) ) {
 109                         return $this->mItems[$hash];
 110                 } else {
 111                         return false;
 112                 }
 113         }
 114
 115         /**
 116          * @param $text string
 117          * @return void
 118          */
 119         public function setText( $text ) {
 120                 $this->uncompress();
 121                 $this->mDefaultHash = $this->addItem( $text );
 122         }
 123
 124         /**
 125          * @return array|bool
 126          */
 127         public function getText() {
 128                 $this->uncompress();
 129                 return $this->getItem( $this->mDefaultHash );
 130         }
 131
 132         /**
 133          * Remove an item
 134          *
 135          * @param $hash string
 136          */
 137         public function removeItem( $hash ) {
 138                 $this->mSize -= strlen( $this->mItems[$hash] );
 139                 unset( $this->mItems[$hash] );
 140         }
 141
 142         /**
 143          * Compress the bulk data in the object
 144          */
 145         public function compress() {
 146                 if ( !$this->mCompressed  ) {
 147                         $this->mItems = gzdeflate( serialize( $this->mItems ) );
 148                         $this->mCompressed = true;
 149                 }
 150         }
 151
 152         /**
 153          * Uncompress bulk data
 154          */
 155         public function uncompress() {
 156                 if ( $this->mCompressed ) {
 157                         $this->mItems = unserialize( gzinflate( $this->mItems ) );
 158                         $this->mCompressed = false;
 159                 }
 160         }
 161
 162         /**
 163          * @return array
 164          */
 165         function __sleep() {
 166                 $this->compress();
 167                 return array( 'mVersion', 'mCompressed', 'mItems', 'mDefaultHash' );
 168         }
 169
 170         function __wakeup() {
 171                 $this->uncompress();
 172         }
 173
 174         /**
 175          * Helper function for compression jobs
 176          * Returns true until the object is "full" and ready to be committed
 177          *
 178          * @return bool
 179          */
 180         public function isHappy() {
 181                 return $this->mSize < $this->mMaxSize
 182                         && count( $this->mItems ) < $this->mMaxCount;
 183         }
 184 }
 185
 186
 187 /**
 188  * Pointer object for an item within a CGZ blob stored in the text table.
 189  */
 190 class HistoryBlobStub {
 191         /**
 192          * One-step cache variable to hold base blobs; operations that
 193          * pull multiple revisions may often pull multiple times from
 194          * the same blob. By keeping the last-used one open, we avoid
 195          * redundant unserialization and decompression overhead.
 196          */
 197         protected static $blobCache = array();
 198
 199         var $mOldId, $mHash, $mRef;
 200
 201         /**
 202          * @param $hash string the content hash of the text
 203          * @param $oldid Integer the old_id for the CGZ object
 204          */
 205         function __construct( $hash = '', $oldid = 0 ) {
 206                 $this->mHash = $hash;
 207         }
 208
 209         /**
 210          * Sets the location (old_id) of the main object to which this object
 211          * points
 212          */
 213         function setLocation( $id ) {
 214                 $this->mOldId = $id;
 215         }
 216
 217         /**
 218          * Sets the location (old_id) of the referring object
 219          */
 220         function setReferrer( $id ) {
 221                 $this->mRef = $id;
 222         }
 223
 224         /**
 225          * Gets the location of the referring object
 226          */
 227         function getReferrer() {
 228                 return $this->mRef;
 229         }
 230
 231         /**
 232          * @return string
 233          */
 234         function getText() {
 235                 if( isset( self::$blobCache[$this->mOldId] ) ) {
 236                         $obj = self::$blobCache[$this->mOldId];
 237                 } else {
 238                         $dbr = wfGetDB( DB_SLAVE );
 239                         $row = $dbr->selectRow( 'text', array( 'old_flags', 'old_text' ), array( 'old_id' => $this->mOldId ) );
 240                         if( !$row ) {
 241                                 return false;
 242                         }
 243                         $flags = explode( ',', $row->old_flags );
 244                         if( in_array( 'external', $flags ) ) {
 245                                 $url = $row->old_text;
 246                                 $parts = explode( '://', $url, 2 );
 247                                 if ( !isset( $parts[1] ) || $parts[1] == '' ) {
 248                                         return false;
 249                                 }
 250                                 $row->old_text = ExternalStore::fetchFromUrl($url);
 251
 252                         }
 253                         if( !in_array( 'object', $flags ) ) {
 254                                 return false;
 255                         }
 256
 257                         if( in_array( 'gzip', $flags ) ) {
 258                                 // This shouldn't happen, but a bug in the compress script
 259                                 // may at times gzip-compress a HistoryBlob object row.
 260                                 $obj = unserialize( gzinflate( $row->old_text ) );
 261                         } else {
 262                                 $obj = unserialize( $row->old_text );
 263                         }
 264
 265                         if( !is_object( $obj ) ) {
 266                                 // Correct for old double-serialization bug.
 267                                 $obj = unserialize( $obj );
 268                         }
 269
 270                         // Save this item for reference; if pulling many
 271                         // items in a row we'll likely use it again.
 272                         $obj->uncompress();
 273                         self::$blobCache = array( $this->mOldId => $obj );
 274                 }
 275                 return $obj->getItem( $this->mHash );
 276         }
 277
 278         /**
 279          * Get the content hash
 280          *
 281          * @return string
 282          */
 283         function getHash() {
 284                 return $this->mHash;
 285         }
 286 }
 287
 288
 289 /**
 290  * To speed up conversion from 1.4 to 1.5 schema, text rows can refer to the
 291  * leftover cur table as the backend. This avoids expensively copying hundreds
 292  * of megabytes of data during the conversion downtime.
 293  *
 294  * Serialized HistoryBlobCurStub objects will be inserted into the text table
 295  * on conversion if $wgFastSchemaUpgrades is set to true.
 296  */
 297 class HistoryBlobCurStub {
 298         var $mCurId;
 299
 300         /**
 301          * @param $curid Integer: the cur_id pointed to
 302          */
 303         function __construct( $curid = 0 ) {
 304                 $this->mCurId = $curid;
 305         }
 306
 307         /**
 308          * Sets the location (cur_id) of the main object to which this object
 309          * points
 310          *
 311          * @param $id int
 312          */
 313         function setLocation( $id ) {
 314                 $this->mCurId = $id;
 315         }
 316
 317         /**
 318          * @return string|bool
 319          */
 320         function getText() {
 321                 $dbr = wfGetDB( DB_SLAVE );
 322                 $row = $dbr->selectRow( 'cur', array( 'cur_text' ), array( 'cur_id' => $this->mCurId ) );
 323                 if( !$row ) {
 324                         return false;
 325                 }
 326                 return $row->cur_text;
 327         }
 328 }
 329
 330 /**
 331  * Diff-based history compression
 332  * Requires xdiff 1.5+ and zlib
 333  */
 334 class DiffHistoryBlob implements HistoryBlob {
 335         /** Uncompressed item cache */
 336         var $mItems = array();
 337
 338         /** Total uncompressed size */
 339         var $mSize = 0;
 340
 341         /**
 342          * Array of diffs. If a diff D from A to B is notated D = B - A, and Z is
 343          * an empty string:
 344          *
 345          *              { item[map[i]] - item[map[i-1]]   where i > 0
 346          *    diff[i] = {
 347          *              { item[map[i]] - Z                where i = 0
 348          */
 349         var $mDiffs;
 350
 351         /** The diff map, see above */
 352         var $mDiffMap;
 353
 354         /**
 355          * The key for getText()
 356          */
 357         var $mDefaultKey;
 358
 359         /**
 360          * Compressed storage
 361          */
 362         var $mCompressed;
 363
 364         /**
 365          * True if the object is locked against further writes
 366          */
 367         var $mFrozen = false;
 368
 369         /**
 370          * The maximum uncompressed size before the object becomes sad
 371          * Should be less than max_allowed_packet
 372          */
 373         var $mMaxSize = 10000000;
 374
 375         /**
 376          * The maximum number of text items before the object becomes sad
 377          */
 378         var $mMaxCount = 100;
 379
 380         /** Constants from xdiff.h */
 381         const XDL_BDOP_INS = 1;
 382         const XDL_BDOP_CPY = 2;
 383         const XDL_BDOP_INSB = 3;
 384
 385         function __construct() {
 386                 if ( !function_exists( 'gzdeflate' ) ) {
 387                         throw new MWException( "Need zlib support to read or write DiffHistoryBlob\n" );
 388                 }
 389         }
 390
 391         /**
 392          * @throws MWException
 393          * @param $text string
 394          * @return int
 395          */
 396         function addItem( $text ) {
 397                 if ( $this->mFrozen ) {
 398                         throw new MWException( __METHOD__.": Cannot add more items after sleep/wakeup" );
 399                 }
 400
 401                 $this->mItems[] = $text;
 402                 $this->mSize += strlen( $text );
 403                 $this->mDiffs = null; // later
 404                 return count( $this->mItems ) - 1;
 405         }
 406
 407         /**
 408          * @param $key string
 409          * @return string
 410          */
 411         function getItem( $key ) {
 412                 return $this->mItems[$key];
 413         }
 414
 415         /**
 416          * @param $text string
 417          */
 418         function setText( $text ) {
 419                 $this->mDefaultKey = $this->addItem( $text );
 420         }
 421
 422         /**
 423          * @return string
 424          */
 425         function getText() {
 426                 return $this->getItem( $this->mDefaultKey );
 427         }
 428
 429         /**
 430          * @throws MWException
 431          */
 432         function compress() {
 433                 if ( !function_exists( 'xdiff_string_rabdiff' ) ) {
 434                         throw new MWException( "Need xdiff 1.5+ support to write DiffHistoryBlob\n" );
 435                 }
 436                 if ( isset( $this->mDiffs ) ) {
 437                         // Already compressed
 438                         return;
 439                 }
 440                 if ( !count( $this->mItems ) ) {
 441                         // Empty
 442                         return;
 443                 }
 444
 445                 // Create two diff sequences: one for main text and one for small text
 446                 $sequences = array(
 447                         'small' => array(
 448                                 'tail' => '',
 449                                 'diffs' => array(),
 450                                 'map' => array(),
 451                         ),
 452                         'main' => array(
 453                                 'tail' => '',
 454                                 'diffs' => array(),
 455                                 'map' => array(),
 456                         ),
 457                 );
 458                 $smallFactor = 0.5;
 459
 460                 for ( $i = 0; $i < count( $this->mItems ); $i++ ) {
 461                         $text = $this->mItems[$i];
 462                         if ( $i == 0 ) {
 463                                 $seqName = 'main';
 464                         } else {
 465                                 $mainTail = $sequences['main']['tail'];
 466                                 if ( strlen( $text ) < strlen( $mainTail ) * $smallFactor ) {
 467                                         $seqName = 'small';
 468                                 } else {
 469                                         $seqName = 'main';
 470                                 }
 471                         }
 472                         $seq =& $sequences[$seqName];
 473                         $tail = $seq['tail'];
 474                         $diff = $this->diff( $tail, $text );
 475                         $seq['diffs'][] = $diff;
 476                         $seq['map'][] = $i;
 477                         $seq['tail'] = $text;
 478                 }
 479                 unset( $seq ); // unlink dangerous alias
 480
 481                 // Knit the sequences together
 482                 $tail = '';
 483                 $this->mDiffs = array();
 484                 $this->mDiffMap = array();
 485                 foreach ( $sequences as $seq ) {
 486                         if ( !count( $seq['diffs'] ) ) {
 487                                 continue;
 488                         }
 489                         if ( $tail === '' ) {
 490                                 $this->mDiffs[] = $seq['diffs'][0];
 491                         } else {
 492                                 $head = $this->patch( '', $seq['diffs'][0] );
 493                                 $this->mDiffs[] = $this->diff( $tail, $head );
 494                         }
 495                         $this->mDiffMap[] = $seq['map'][0];
 496                         for ( $i = 1; $i < count( $seq['diffs'] ); $i++ ) {
 497                                 $this->mDiffs[] = $seq['diffs'][$i];
 498                                 $this->mDiffMap[] = $seq['map'][$i];
 499                         }
 500                         $tail = $seq['tail'];
 501                 }
 502         }
 503
 504         /**
 505          * @param $t1
 506          * @param $t2
 507          * @return string
 508          */
 509         function diff( $t1, $t2 ) {
 510                 # Need to do a null concatenation with warnings off, due to bugs in the current version of xdiff
 511                 # "String is not zero-terminated"
 512                 wfSuppressWarnings();
 513                 $diff = xdiff_string_rabdiff( $t1, $t2 ) . '';
 514                 wfRestoreWarnings();
 515                 return $diff;
 516         }
 517
 518         /**
 519          * @param $base
 520          * @param $diff
 521          * @return bool|string
 522          */
 523         function patch( $base, $diff ) {
 524                 if ( function_exists( 'xdiff_string_bpatch' ) ) {
 525                         wfSuppressWarnings();
 526                         $text = xdiff_string_bpatch( $base, $diff ) . '';
 527                         wfRestoreWarnings();
 528                         return $text;
 529                 }
 530
 531                 # Pure PHP implementation
 532
 533                 $header = unpack( 'Vofp/Vcsize', substr( $diff, 0, 8 ) );
 534
 535                 # Check the checksum if hash/mhash is available
 536                 $ofp = $this->xdiffAdler32( $base );
 537                 if ( $ofp !== false && $ofp !== substr( $diff, 0, 4 ) ) {
 538                         wfDebug( __METHOD__. ": incorrect base checksum\n" );
 539                         return false;
 540                 }
 541                 if ( $header['csize'] != strlen( $base ) ) {
 542                         wfDebug( __METHOD__. ": incorrect base length\n" );
 543                         return false;
 544                 }
 545
 546                 $p = 8;
 547                 $out = '';
 548                 while ( $p < strlen( $diff ) ) {
 549                         $x = unpack( 'Cop', substr( $diff, $p, 1 ) );
 550                         $op = $x['op'];
 551                         ++$p;
 552                         switch ( $op ) {
 553                         case self::XDL_BDOP_INS:
 554                                 $x = unpack( 'Csize', substr( $diff, $p, 1 ) );
 555                                 $p++;
 556                                 $out .= substr( $diff, $p, $x['size'] );
 557                                 $p += $x['size'];
 558                                 break;
 559                         case self::XDL_BDOP_INSB:
 560                                 $x = unpack( 'Vcsize', substr( $diff, $p, 4 ) );
 561                                 $p += 4;
 562                                 $out .= substr( $diff, $p, $x['csize'] );
 563                                 $p += $x['csize'];
 564                                 break;
 565                         case self::XDL_BDOP_CPY:
 566                                 $x = unpack( 'Voff/Vcsize', substr( $diff, $p, 8 ) );
 567                                 $p += 8;
 568                                 $out .= substr( $base, $x['off'], $x['csize'] );
 569                                 break;
 570                         default:
 571                                 wfDebug( __METHOD__.": invalid op\n" );
 572                                 return false;
 573                         }
 574                 }
 575                 return $out;
 576         }
 577
 578         /**
 579          * Compute a binary "Adler-32" checksum as defined by LibXDiff, i.e. with
 580          * the bytes backwards and initialised with 0 instead of 1. See bug 34428.
 581          *
 582          * Returns false if no hashing library is available
 583          */
 584         function xdiffAdler32( $s ) {
 585                 static $init;
 586                 if ( $init === null ) {
 587                         $init = str_repeat( "\xf0", 205 ) . "\xee" . str_repeat( "\xf0", 67 ) . "\x02";
 588                 }
 589                 // The real Adler-32 checksum of $init is zero, so it initialises the
 590                 // state to zero, as it is at the start of LibXDiff's checksum
 591                 // algorithm. Appending the subject string then simulates LibXDiff.
 592                 if ( function_exists( 'hash' ) ) {
 593                         $hash = hash( 'adler32', $init . $s, true );
 594                 } elseif ( function_exists( 'mhash' ) ) {
 595                         $hash = mhash( MHASH_ADLER32, $init . $s );
 596                 } else {
 597                         return false;
 598                 }
 599                 return strrev( $hash );
 600         }
 601
 602         function uncompress() {
 603                 if ( !$this->mDiffs ) {
 604                         return;
 605                 }
 606                 $tail = '';
 607                 for ( $diffKey = 0; $diffKey < count( $this->mDiffs ); $diffKey++ ) {
 608                         $textKey = $this->mDiffMap[$diffKey];
 609                         $text = $this->patch( $tail, $this->mDiffs[$diffKey] );
 610                         $this->mItems[$textKey] = $text;
 611                         $tail = $text;
 612                 }
 613         }
 614
 615         /**
 616          * @return array
 617          */
 618         function __sleep() {
 619                 $this->compress();
 620                 if ( !count( $this->mItems ) ) {
 621                         // Empty object
 622                         $info = false;
 623                 } else {
 624                         // Take forward differences to improve the compression ratio for sequences
 625                         $map = '';
 626                         $prev = 0;
 627                         foreach ( $this->mDiffMap as $i ) {
 628                                 if ( $map !== '' ) {
 629                                         $map .= ',';
 630                                 }
 631                                 $map .= $i - $prev;
 632                                 $prev = $i;
 633                         }
 634                         $info = array(
 635                                 'diffs' => $this->mDiffs,
 636                                 'map' => $map
 637                         );
 638                 }
 639                 if ( isset( $this->mDefaultKey ) ) {
 640                         $info['default'] = $this->mDefaultKey;
 641                 }
 642                 $this->mCompressed = gzdeflate( serialize( $info ) );
 643                 return array( 'mCompressed' );
 644         }
 645
 646         function __wakeup() {
 647                 // addItem() doesn't work if mItems is partially filled from mDiffs
 648                 $this->mFrozen = true;
 649                 $info = unserialize( gzinflate( $this->mCompressed ) );
 650                 unset( $this->mCompressed );
 651
 652                 if ( !$info ) {
 653                         // Empty object
 654                         return;
 655                 }
 656
 657                 if ( isset( $info['default'] ) ) {
 658                         $this->mDefaultKey = $info['default'];
 659                 }
 660                 $this->mDiffs = $info['diffs'];
 661                 if ( isset( $info['base'] ) ) {
 662                         // Old format
 663                         $this->mDiffMap = range( 0, count( $this->mDiffs ) - 1 );
 664                         array_unshift( $this->mDiffs,
 665                                 pack( 'VVCV', 0, 0, self::XDL_BDOP_INSB, strlen( $info['base'] ) ) .
 666                                 $info['base'] );
 667                 } else {
 668                         // New format
 669                         $map = explode( ',', $info['map'] );
 670                         $cur = 0;
 671                         $this->mDiffMap = array();
 672                         foreach ( $map as $i ) {
 673                                 $cur += $i;
 674                                 $this->mDiffMap[] = $cur;
 675                         }
 676                 }
 677                 $this->uncompress();
 678         }
 679
 680         /**
 681          * Helper function for compression jobs
 682          * Returns true until the object is "full" and ready to be committed
 683          *
 684          * @return bool
 685          */
 686         function isHappy() {
 687                 return $this->mSize < $this->mMaxSize
 688                         && count( $this->mItems ) < $this->mMaxCount;
 689         }
 690
 691 }