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         function __construct( $fileName ) {
 110                 $this->handle = fopen( $fileName, 'rb' );
 111                 if ( !$this->handle ) {
 112                         throw new MWException( 'Unable to open CDB file "' . $fileName . '"' );
 113                 }
 114                 $this->findStart();
 115         }
 116
 117         function close() {
 118                 if( isset( $this->handle ) ) {
 119                         fclose( $this->handle );
 120                 }
 121                 unset( $this->handle );
 122         }
 123
 124         /**
 125          * @param $key
 126          * @return bool|string
 127          */
 128         public function get( $key ) {
 129                 // strval is required
 130                 if ( $this->find( strval( $key ) ) ) {
 131                         return $this->read( $this->dlen, $this->dpos );
 132                 } else {
 133                         return false;
 134                 }
 135         }
 136
 137         /**
 138          * @param $key
 139          * @param $pos
 140          * @return bool
 141          */
 142         protected function match( $key, $pos ) {
 143                 $buf = $this->read( strlen( $key ), $pos );
 144                 return $buf === $key;
 145         }
 146
 147         protected function findStart() {
 148                 $this->loop = 0;
 149         }
 150
 151         /**
 152          * @throws MWException
 153          * @param $length
 154          * @param $pos
 155          * @return string
 156          */
 157         protected function read( $length, $pos ) {
 158                 if ( fseek( $this->handle, $pos ) == -1 ) {
 159                         // This can easily happen if the internal pointers are incorrect
 160                         throw new MWException( __METHOD__.': seek failed, file may be corrupted.' );
 161                 }
 162
 163                 if ( $length == 0 ) {
 164                         return '';
 165                 }
 166
 167                 $buf = fread( $this->handle, $length );
 168                 if ( $buf === false || strlen( $buf ) !== $length ) {
 169                         throw new MWException( __METHOD__.': read from CDB file failed, file may be corrupted' );
 170                 }
 171                 return $buf;
 172         }
 173
 174         /**
 175          * Unpack an unsigned integer and throw an exception if it needs more than 31 bits
 176          * @param $s
 177          * @return
 178          */
 179         protected function unpack31( $s ) {
 180                 $data = unpack( 'V', $s );
 181                 if ( $data[1] > 0x7fffffff ) {
 182                         throw new MWException( __METHOD__.': error in CDB file, integer too big' );
 183                 }
 184                 return $data[1];
 185         }
 186
 187         /**
 188          * Unpack a 32-bit signed integer
 189          * @param $s
 190          * @return int
 191          */
 192         protected function unpackSigned( $s ) {
 193                 $data = unpack( 'va/vb', $s );
 194                 return $data['a'] | ( $data['b'] << 16 );
 195         }
 196
 197         /**
 198          * @param $key
 199          * @return bool
 200          */
 201         protected function findNext( $key ) {
 202                 if ( !$this->loop ) {
 203                         $u = CdbFunctions::hash( $key );
 204                         $buf = $this->read( 8, ( $u << 3 ) & 2047 );
 205                         $this->hslots = $this->unpack31( substr( $buf, 4 ) );
 206                         if ( !$this->hslots ) {
 207                                 return false;
 208                         }
 209                         $this->hpos = $this->unpack31( substr( $buf, 0, 4 ) );
 210                         $this->khash = $u;
 211                         $u = CdbFunctions::unsignedShiftRight( $u, 8 );
 212                         $u = CdbFunctions::unsignedMod( $u, $this->hslots );
 213                         $u <<= 3;
 214                         $this->kpos = $this->hpos + $u;
 215                 }
 216
 217                 while ( $this->loop < $this->hslots ) {
 218                         $buf = $this->read( 8, $this->kpos );
 219                         $pos = $this->unpack31( substr( $buf, 4 ) );
 220                         if ( !$pos ) {
 221                                 return false;
 222                         }
 223                         $this->loop += 1;
 224                         $this->kpos += 8;
 225                         if ( $this->kpos == $this->hpos + ( $this->hslots << 3 ) ) {
 226                                 $this->kpos = $this->hpos;
 227                         }
 228                         $u = $this->unpackSigned( substr( $buf, 0, 4 ) );
 229                         if ( $u === $this->khash ) {
 230                                 $buf = $this->read( 8, $pos );
 231                                 $keyLen = $this->unpack31( substr( $buf, 0, 4 ) );
 232                                 if ( $keyLen == strlen( $key ) && $this->match( $key, $pos + 8 ) ) {
 233                                         // Found
 234                                         $this->dlen = $this->unpack31( substr( $buf, 4 ) );
 235                                         $this->dpos = $pos + 8 + $keyLen;
 236                                         return true;
 237                                 }
 238                         }
 239                 }
 240                 return false;
 241         }
 242
 243         /**
 244          * @param $key
 245          * @return bool
 246          */
 247         protected function find( $key ) {
 248                 $this->findStart();
 249                 return $this->findNext( $key );
 250         }
 251 }
 252
 253 /**
 254  * CDB writer class
 255  */
 256 class CdbWriter_PHP extends CdbWriter {
 257         var $handle, $realFileName, $tmpFileName;
 258
 259         var $hplist;
 260         var $numEntries, $pos;
 261
 262         function __construct( $fileName ) {
 263                 $this->realFileName = $fileName;
 264                 $this->tmpFileName = $fileName . '.tmp.' . mt_rand( 0, 0x7fffffff );
 265                 $this->handle = fopen( $this->tmpFileName, 'wb' );
 266                 if ( !$this->handle ) {
 267                         throw new MWException( 'Unable to open CDB file for write "' . $fileName . '"' );
 268                 }
 269                 $this->hplist = array();
 270                 $this->numentries = 0;
 271                 $this->pos = 2048; // leaving space for the pointer array, 256 * 8
 272                 if ( fseek( $this->handle, $this->pos ) == -1 ) {
 273                         throw new MWException( __METHOD__.': fseek failed' );
 274                 }
 275         }
 276
 277         function __destruct() {
 278                 if ( isset( $this->handle ) ) {
 279                         $this->close();
 280                 }
 281         }
 282
 283         /**
 284          * @param $key
 285          * @param $value
 286          * @return
 287          */
 288         public function set( $key, $value ) {
 289                 if ( strval( $key ) === '' ) {
 290                         // DBA cross-check hack
 291                         return;
 292                 }
 293                 $this->addbegin( strlen( $key ), strlen( $value ) );
 294                 $this->write( $key );
 295                 $this->write( $value );
 296                 $this->addend( strlen( $key ), strlen( $value ), CdbFunctions::hash( $key ) );
 297         }
 298
 299         /**
 300          * @throws MWException
 301          */
 302         public function close() {
 303                 $this->finish();
 304                 if( isset($this->handle) ) {
 305                         fclose( $this->handle );
 306                 }
 307                 if ( wfIsWindows() && file_exists($this->realFileName) ) {
 308                         unlink( $this->realFileName );
 309                 }
 310                 if ( !rename( $this->tmpFileName, $this->realFileName ) ) {
 311                         throw new MWException( 'Unable to move the new CDB file into place.' );
 312                 }
 313                 unset( $this->handle );
 314         }
 315
 316         /**
 317          * @throws MWException
 318          * @param $buf
 319          */
 320         protected function write( $buf ) {
 321                 $len = fwrite( $this->handle, $buf );
 322                 if ( $len !== strlen( $buf ) ) {
 323                         throw new MWException( 'Error writing to CDB file.' );
 324                 }
 325         }
 326
 327         /**
 328          * @throws MWException
 329          * @param $len
 330          */
 331         protected function posplus( $len ) {
 332                 $newpos = $this->pos + $len;
 333                 if ( $newpos > 0x7fffffff ) {
 334                         throw new MWException( 'A value in the CDB file is too large' );
 335                 }
 336                 $this->pos = $newpos;
 337         }
 338
 339         /**
 340          * @param $keylen
 341          * @param $datalen
 342          * @param $h
 343          */
 344         protected function addend( $keylen, $datalen, $h ) {
 345                 $this->hplist[] = array(
 346                         'h' => $h,
 347                         'p' => $this->pos
 348                 );
 349
 350                 $this->numentries++;
 351                 $this->posplus( 8 );
 352                 $this->posplus( $keylen );
 353                 $this->posplus( $datalen );
 354         }
 355
 356         /**
 357          * @throws MWException
 358          * @param $keylen
 359          * @param $datalen
 360          */
 361         protected function addbegin( $keylen, $datalen ) {
 362                 if ( $keylen > 0x7fffffff ) {
 363                         throw new MWException( __METHOD__.': key length too long' );
 364                 }
 365                 if ( $datalen > 0x7fffffff ) {
 366                         throw new MWException( __METHOD__.': data length too long' );
 367                 }
 368                 $buf = pack( 'VV', $keylen, $datalen );
 369                 $this->write( $buf );
 370         }
 371
 372         /**
 373          * @throws MWException
 374          */
 375         protected function finish() {
 376                 // Hack for DBA cross-check
 377                 $this->hplist = array_reverse( $this->hplist );
 378
 379                 // Calculate the number of items that will be in each hashtable
 380                 $counts = array_fill( 0, 256, 0 );
 381                 foreach ( $this->hplist as $item ) {
 382                         ++ $counts[ 255 & $item['h'] ];
 383                 }
 384
 385                 // Fill in $starts with the *end* indexes
 386                 $starts = array();
 387                 $pos = 0;
 388                 for ( $i = 0; $i < 256; ++$i ) {
 389                         $pos += $counts[$i];
 390                         $starts[$i] = $pos;
 391                 }
 392
 393                 // Excessively clever and indulgent code to simultaneously fill $packedTables
 394                 // with the packed hashtables, and adjust the elements of $starts
 395                 // to actually point to the starts instead of the ends.
 396                 $packedTables = array_fill( 0, $this->numentries, false );
 397                 foreach ( $this->hplist as $item ) {
 398                         $packedTables[--$starts[255 & $item['h']]] = $item;
 399                 }
 400
 401                 $final = '';
 402                 for ( $i = 0; $i < 256; ++$i ) {
 403                         $count = $counts[$i];
 404
 405                         // The size of the hashtable will be double the item count.
 406                         // The rest of the slots will be empty.
 407                         $len = $count + $count;
 408                         $final .= pack( 'VV', $this->pos, $len );
 409
 410                         $hashtable = array();
 411                         for ( $u = 0; $u < $len; ++$u ) {
 412                                 $hashtable[$u] = array( 'h' => 0, 'p' => 0 );
 413                         }
 414
 415                         // Fill the hashtable, using the next empty slot if the hashed slot
 416                         // is taken.
 417                         for ( $u = 0; $u < $count; ++$u ) {
 418                                 $hp = $packedTables[$starts[$i] + $u];
 419                                 $where = CdbFunctions::unsignedMod(
 420                                         CdbFunctions::unsignedShiftRight( $hp['h'], 8 ), $len );
 421                                 while ( $hashtable[$where]['p'] )
 422                                         if ( ++$where == $len )
 423                                                 $where = 0;
 424                                 $hashtable[$where] = $hp;
 425                         }
 426
 427                         // Write the hashtable
 428                         for ( $u = 0; $u < $len; ++$u ) {
 429                                 $buf = pack( 'vvV',
 430                                         $hashtable[$u]['h'] & 0xffff,
 431                                         CdbFunctions::unsignedShiftRight( $hashtable[$u]['h'], 16 ),
 432                                         $hashtable[$u]['p'] );
 433                                 $this->write( $buf );
 434                                 $this->posplus( 8 );
 435                         }
 436                 }
 437
 438                 // Write the pointer array at the start of the file
 439                 rewind( $this->handle );
 440                 if ( ftell( $this->handle ) != 0 ) {
 441                         throw new MWException( __METHOD__.': Error rewinding to start of file' );
 442                 }
 443                 $this->write( $final );
 444         }
 445 }