includes/Cdb_PHP.php

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