includes/Cdb_PHP.php

   1 <?php
   2 /**
   3  * This is a port of D.J. Bernstein's CDB to PHP. It's based on the copy that
   4  * appears in PHP 5.3. Changes are:
   5  *    * Error returns replaced with exceptions
   6  *    * Exception thrown if sizes or offsets are between 2GB and 4GB
   7  *    * Some variables renamed
   8  *
   9  * @file
  10  */
  11
  12 /**
  13  * Common functions for readers and writers
  14  */
  15 class CdbFunctions {
  16         /**
  17          * Take a modulo of a signed integer as if it were an unsigned integer.
  18          * $b must be less than 0x40000000 and greater than 0
  19          *
  20          * @param $a
  21          * @param $b
  22          *
  23          * @return int
  24          */
  25         public static function unsignedMod( $a, $b ) {
  26                 if ( $a & 0x80000000 ) {
  27                         $m = ( $a & 0x7fffffff ) % $b + 2 * ( 0x40000000 % $b );
  28                         return $m % $b;
  29                 } else {
  30                         return $a % $b;
  31                 }
  32         }
  33
  34         /**
  35          * Shift a signed integer right as if it were unsigned
  36          * @param $a
  37          * @param $b
  38          * @return int
  39          */
  40         public static function unsignedShiftRight( $a, $b ) {
  41                 if ( $b == 0 ) {
  42                         return $a;
  43                 }
  44                 if ( $a & 0x80000000 ) {
  45                         return ( ( $a & 0x7fffffff ) >> $b ) | ( 0x40000000 >> ( $b - 1 ) );
  46                 } else {
  47                         return $a >> $b;
  48                 }
  49         }
  50
  51         /**
  52          * The CDB hash function.
  53          *
  54          * @param $s
  55          *
  56          * @return
  57          */
  58         public static function hash( $s ) {
  59                 $h = 5381;
  60                 for ( $i = 0; $i < strlen( $s ); $i++ ) {
  61                         $h5 = ($h << 5) & 0xffffffff;
  62                         // Do a 32-bit sum
  63                         // Inlined here for speed
  64                         $sum = ($h & 0x3fffffff) + ($h5 & 0x3fffffff);
  65                         $h =
  66                                 (
  67                                         ( $sum & 0x40000000 ? 1 : 0 )
  68                                         + ( $h & 0x80000000 ? 2 : 0 )
  69                                         + ( $h & 0x40000000 ? 1 : 0 )
  70                                         + ( $h5 & 0x80000000 ? 2 : 0 )
  71                                         + ( $h5 & 0x40000000 ? 1 : 0 )
  72                                 ) << 30
  73                                 | ( $sum & 0x3fffffff );
  74                         $h ^= ord( $s[$i] );
  75                         $h &= 0xffffffff;
  76                 }
  77                 return $h;
  78         }
  79 }
  80
  81 /**
  82  * CDB reader class
  83  */
  84 class CdbReader_PHP extends CdbReader {
  85         /** The file handle */
  86         var $handle;
  87
  88         /* number of hash slots searched under this key */
  89         var $loop;
  90
  91         /* initialized if loop is nonzero */
  92         var $khash;
  93
  94         /* initialized if loop is nonzero */
  95         var $kpos;
  96
  97         /* initialized if loop is nonzero */
  98         var $hpos;
  99
 100         /* initialized if loop is nonzero */
 101         var $hslots;
 102
 103         /* initialized if findNext() returns true */
 104         var $dpos;
 105
 106         /* initialized if cdb_findnext() returns 1 */
 107         var $dlen;
 108
 109         /**
 110          * @param $fileName string
 111          */
 112         function __construct( $fileName ) {
 113                 $this->handle = fopen( $fileName, 'rb' );
 114                 if ( !$this->handle ) {
 115                         throw new MWException( 'Unable to open CDB file "' . $fileName . '"' );
 116                 }
 117                 $this->findStart();
 118         }
 119
 120         function close() {
 121                 if( isset( $this->handle ) ) {
 122                         fclose( $this->handle );
 123                 }
 124                 unset( $this->handle );
 125         }
 126
 127         /**
 128          * @param $key
 129          * @return bool|string
 130          */
 131         public function get( $key ) {
 132                 // strval is required
 133                 if ( $this->find( strval( $key ) ) ) {
 134                         return $this->read( $this->dlen, $this->dpos );
 135                 } else {
 136                         return false;
 137                 }
 138         }
 139
 140         /**
 141          * @param $key
 142          * @param $pos
 143          * @return bool
 144          */
 145         protected function match( $key, $pos ) {
 146                 $buf = $this->read( strlen( $key ), $pos );
 147                 return $buf === $key;
 148         }
 149
 150         protected function findStart() {
 151                 $this->loop = 0;
 152         }
 153
 154         /**
 155          * @throws MWException
 156          * @param $length
 157          * @param $pos
 158          * @return string
 159          */
 160         protected function read( $length, $pos ) {
 161                 if ( fseek( $this->handle, $pos ) == -1 ) {
 162                         // This can easily happen if the internal pointers are incorrect
 163                         throw new MWException( __METHOD__.': seek failed, file may be corrupted.' );
 164                 }
 165
 166                 if ( $length == 0 ) {
 167                         return '';
 168                 }
 169
 170                 $buf = fread( $this->handle, $length );
 171                 if ( $buf === false || strlen( $buf ) !== $length ) {
 172                         throw new MWException( __METHOD__.': read from CDB file failed, file may be corrupted' );
 173                 }
 174                 return $buf;
 175         }
 176
 177         /**
 178          * Unpack an unsigned integer and throw an exception if it needs more than 31 bits
 179          * @param $s
 180          * @return
 181          */
 182         protected function unpack31( $s ) {
 183                 $data = unpack( 'V', $s );
 184                 if ( $data[1] > 0x7fffffff ) {
 185                         throw new MWException( __METHOD__.': error in CDB file, integer too big' );
 186                 }
 187                 return $data[1];
 188         }
 189
 190         /**
 191          * Unpack a 32-bit signed integer
 192          * @param $s
 193          * @return int
 194          */
 195         protected function unpackSigned( $s ) {
 196                 $data = unpack( 'va/vb', $s );
 197                 return $data['a'] | ( $data['b'] << 16 );
 198         }
 199
 200         /**
 201          * @param $key
 202          * @return bool
 203          */
 204         protected function findNext( $key ) {
 205                 if ( !$this->loop ) {
 206                         $u = CdbFunctions::hash( $key );
 207                         $buf = $this->read( 8, ( $u << 3 ) & 2047 );
 208                         $this->hslots = $this->unpack31( substr( $buf, 4 ) );
 209                         if ( !$this->hslots ) {
 210                                 return false;
 211                         }
 212                         $this->hpos = $this->unpack31( substr( $buf, 0, 4 ) );
 213                         $this->khash = $u;
 214                         $u = CdbFunctions::unsignedShiftRight( $u, 8 );
 215                         $u = CdbFunctions::unsignedMod( $u, $this->hslots );
 216                         $u <<= 3;
 217                         $this->kpos = $this->hpos + $u;
 218                 }
 219
 220                 while ( $this->loop < $this->hslots ) {
 221                         $buf = $this->read( 8, $this->kpos );
 222                         $pos = $this->unpack31( substr( $buf, 4 ) );
 223                         if ( !$pos ) {
 224                                 return false;
 225                         }
 226                         $this->loop += 1;
 227                         $this->kpos += 8;
 228                         if ( $this->kpos == $this->hpos + ( $this->hslots << 3 ) ) {
 229                                 $this->kpos = $this->hpos;
 230                         }
 231                         $u = $this->unpackSigned( substr( $buf, 0, 4 ) );
 232                         if ( $u === $this->khash ) {
 233                                 $buf = $this->read( 8, $pos );
 234                                 $keyLen = $this->unpack31( substr( $buf, 0, 4 ) );
 235                                 if ( $keyLen == strlen( $key ) && $this->match( $key, $pos + 8 ) ) {
 236                                         // Found
 237                                         $this->dlen = $this->unpack31( substr( $buf, 4 ) );
 238                                         $this->dpos = $pos + 8 + $keyLen;
 239                                         return true;
 240                                 }
 241                         }
 242                 }
 243                 return false;
 244         }
 245
 246         /**
 247          * @param $key
 248          * @return bool
 249          */
 250         protected function find( $key ) {
 251                 $this->findStart();
 252                 return $this->findNext( $key );
 253         }
 254 }
 255
 256 /**
 257  * CDB writer class
 258  */
 259 class CdbWriter_PHP extends CdbWriter {
 260         var $handle, $realFileName, $tmpFileName;
 261
 262         var $hplist;
 263         var $numentries, $pos;
 264
 265         /**
 266          * @param $fileName string
 267          */
 268         function __construct( $fileName ) {
 269                 $this->realFileName = $fileName;
 270                 $this->tmpFileName = $fileName . '.tmp.' . mt_rand( 0, 0x7fffffff );
 271                 $this->handle = fopen( $this->tmpFileName, 'wb' );
 272                 if ( !$this->handle ) {
 273                         throw new MWException( 'Unable to open CDB file for write "' . $fileName . '"' );
 274                 }
 275                 $this->hplist = array();
 276                 $this->numentries = 0;
 277                 $this->pos = 2048; // leaving space for the pointer array, 256 * 8
 278                 if ( fseek( $this->handle, $this->pos ) == -1 ) {
 279                         throw new MWException( __METHOD__.': fseek failed' );
 280                 }
 281         }
 282
 283         function __destruct() {
 284                 if ( isset( $this->handle ) ) {
 285                         $this->close();
 286                 }
 287         }
 288
 289         /**
 290          * @param $key
 291          * @param $value
 292          * @return
 293          */
 294         public function set( $key, $value ) {
 295                 if ( strval( $key ) === '' ) {
 296                         // DBA cross-check hack
 297                         return;
 298                 }
 299                 $this->addbegin( strlen( $key ), strlen( $value ) );
 300                 $this->write( $key );
 301                 $this->write( $value );
 302                 $this->addend( strlen( $key ), strlen( $value ), CdbFunctions::hash( $key ) );
 303         }
 304
 305         /**
 306          * @throws MWException
 307          */
 308         public function close() {
 309                 $this->finish();
 310                 if( isset($this->handle) ) {
 311                         fclose( $this->handle );
 312                 }
 313                 if ( wfIsWindows() && file_exists($this->realFileName) ) {
 314                         unlink( $this->realFileName );
 315                 }
 316                 if ( !rename( $this->tmpFileName, $this->realFileName ) ) {
 317                         throw new MWException( 'Unable to move the new CDB file into place.' );
 318                 }
 319                 unset( $this->handle );
 320         }
 321
 322         /**
 323          * @throws MWException
 324          * @param $buf
 325          */
 326         protected function write( $buf ) {
 327                 $len = fwrite( $this->handle, $buf );
 328                 if ( $len !== strlen( $buf ) ) {
 329                         throw new MWException( 'Error writing to CDB file.' );
 330                 }
 331         }
 332
 333         /**
 334          * @throws MWException
 335          * @param $len
 336          */
 337         protected function posplus( $len ) {
 338                 $newpos = $this->pos + $len;
 339                 if ( $newpos > 0x7fffffff ) {
 340                         throw new MWException( 'A value in the CDB file is too large' );
 341                 }
 342                 $this->pos = $newpos;
 343         }
 344
 345         /**
 346          * @param $keylen
 347          * @param $datalen
 348          * @param $h
 349          */
 350         protected function addend( $keylen, $datalen, $h ) {
 351                 $this->hplist[] = array(
 352                         'h' => $h,
 353                         'p' => $this->pos
 354                 );
 355
 356                 $this->numentries++;
 357                 $this->posplus( 8 );
 358                 $this->posplus( $keylen );
 359                 $this->posplus( $datalen );
 360         }
 361
 362         /**
 363          * @throws MWException
 364          * @param $keylen
 365          * @param $datalen
 366          */
 367         protected function addbegin( $keylen, $datalen ) {
 368                 if ( $keylen > 0x7fffffff ) {
 369                         throw new MWException( __METHOD__.': key length too long' );
 370                 }
 371                 if ( $datalen > 0x7fffffff ) {
 372                         throw new MWException( __METHOD__.': data length too long' );
 373                 }
 374                 $buf = pack( 'VV', $keylen, $datalen );
 375                 $this->write( $buf );
 376         }
 377
 378         /**
 379          * @throws MWException
 380          */
 381         protected function finish() {
 382                 // Hack for DBA cross-check
 383                 $this->hplist = array_reverse( $this->hplist );
 384
 385                 // Calculate the number of items that will be in each hashtable
 386                 $counts = array_fill( 0, 256, 0 );
 387                 foreach ( $this->hplist as $item ) {
 388                         ++ $counts[ 255 & $item['h'] ];
 389                 }
 390
 391                 // Fill in $starts with the *end* indexes
 392                 $starts = array();
 393                 $pos = 0;
 394                 for ( $i = 0; $i < 256; ++$i ) {
 395                         $pos += $counts[$i];
 396                         $starts[$i] = $pos;
 397                 }
 398
 399                 // Excessively clever and indulgent code to simultaneously fill $packedTables
 400                 // with the packed hashtables, and adjust the elements of $starts
 401                 // to actually point to the starts instead of the ends.
 402                 $packedTables = array_fill( 0, $this->numentries, false );
 403                 foreach ( $this->hplist as $item ) {
 404                         $packedTables[--$starts[255 & $item['h']]] = $item;
 405                 }
 406
 407                 $final = '';
 408                 for ( $i = 0; $i < 256; ++$i ) {
 409                         $count = $counts[$i];
 410
 411                         // The size of the hashtable will be double the item count.
 412                         // The rest of the slots will be empty.
 413                         $len = $count + $count;
 414                         $final .= pack( 'VV', $this->pos, $len );
 415
 416                         $hashtable = array();
 417                         for ( $u = 0; $u < $len; ++$u ) {
 418                                 $hashtable[$u] = array( 'h' => 0, 'p' => 0 );
 419                         }
 420
 421                         // Fill the hashtable, using the next empty slot if the hashed slot
 422                         // is taken.
 423                         for ( $u = 0; $u < $count; ++$u ) {
 424                                 $hp = $packedTables[$starts[$i] + $u];
 425                                 $where = CdbFunctions::unsignedMod(
 426                                         CdbFunctions::unsignedShiftRight( $hp['h'], 8 ), $len );
 427                                 while ( $hashtable[$where]['p'] )
 428                                         if ( ++$where == $len )
 429                                                 $where = 0;
 430                                 $hashtable[$where] = $hp;
 431                         }
 432
 433                         // Write the hashtable
 434                         for ( $u = 0; $u < $len; ++$u ) {
 435                                 $buf = pack( 'vvV',
 436                                         $hashtable[$u]['h'] & 0xffff,
 437                                         CdbFunctions::unsignedShiftRight( $hashtable[$u]['h'], 16 ),
 438                                         $hashtable[$u]['p'] );
 439                                 $this->write( $buf );
 440                                 $this->posplus( 8 );
 441                         }
 442                 }
 443
 444                 // Write the pointer array at the start of the file
 445                 rewind( $this->handle );
 446                 if ( ftell( $this->handle ) != 0 ) {
 447                         throw new MWException( __METHOD__.': Error rewinding to start of file' );
 448                 }
 449                 $this->write( $final );
 450         }
 451 }