| 1 |
<?php |
| 2 |
|
| 3 |
declare( strict_types=1 ); |
| 4 |
|
| 5 |
namespace MaxMind\Db\Reader; |
| 6 |
|
| 7 |
// @codingStandardsIgnoreLine |
| 8 |
|
| 9 |
class Decoder { |
| 10 |
/** |
| 11 |
* @var resource |
| 12 |
*/ |
| 13 |
private $fileStream; |
| 14 |
|
| 15 |
/** |
| 16 |
* @var int |
| 17 |
*/ |
| 18 |
private $pointerBase; |
| 19 |
|
| 20 |
/** |
| 21 |
* This is only used for unit testing. |
| 22 |
* |
| 23 |
* @var bool |
| 24 |
*/ |
| 25 |
private $pointerTestHack; |
| 26 |
|
| 27 |
/** |
| 28 |
* @var bool |
| 29 |
*/ |
| 30 |
private $switchByteOrder; |
| 31 |
|
| 32 |
private const _EXTENDED = 0; |
| 33 |
private const _POINTER = 1; |
| 34 |
private const _UTF8_STRING = 2; |
| 35 |
private const _DOUBLE = 3; |
| 36 |
private const _BYTES = 4; |
| 37 |
private const _UINT16 = 5; |
| 38 |
private const _UINT32 = 6; |
| 39 |
private const _MAP = 7; |
| 40 |
private const _INT32 = 8; |
| 41 |
private const _UINT64 = 9; |
| 42 |
private const _UINT128 = 10; |
| 43 |
private const _ARRAY = 11; |
| 44 |
// 12 is the container type |
| 45 |
// 13 is the end marker type |
| 46 |
private const _BOOLEAN = 14; |
| 47 |
private const _FLOAT = 15; |
| 48 |
|
| 49 |
/** |
| 50 |
* @param resource $fileStream |
| 51 |
*/ |
| 52 |
public function __construct( |
| 53 |
$fileStream, |
| 54 |
int $pointerBase = 0, |
| 55 |
bool $pointerTestHack = false |
| 56 |
) { |
| 57 |
$this->fileStream = $fileStream; |
| 58 |
$this->pointerBase = $pointerBase; |
| 59 |
|
| 60 |
$this->pointerTestHack = $pointerTestHack; |
| 61 |
|
| 62 |
$this->switchByteOrder = $this->isPlatformLittleEndian(); |
| 63 |
} |
| 64 |
|
| 65 |
public function decode( int $offset ): array { |
| 66 |
$ctrlByte = \ord( Util::read( $this->fileStream, $offset, 1 ) ); |
| 67 |
++ $offset; |
| 68 |
|
| 69 |
$type = $ctrlByte >> 5; |
| 70 |
|
| 71 |
// Pointers are a special case, we don't read the next $size bytes, we |
| 72 |
// use the size to determine the length of the pointer and then follow |
| 73 |
// it. |
| 74 |
if ( $type === self::_POINTER ) { |
| 75 |
[ $pointer, $offset ] = $this->decodePointer( $ctrlByte, $offset ); |
| 76 |
|
| 77 |
// for unit testing |
| 78 |
if ( $this->pointerTestHack ) { |
| 79 |
return [ $pointer ]; |
| 80 |
} |
| 81 |
|
| 82 |
[ $result ] = $this->decode( $pointer ); |
| 83 |
|
| 84 |
return [ $result, $offset ]; |
| 85 |
} |
| 86 |
|
| 87 |
if ( $type === self::_EXTENDED ) { |
| 88 |
$nextByte = \ord( Util::read( $this->fileStream, $offset, 1 ) ); |
| 89 |
|
| 90 |
$type = $nextByte + 7; |
| 91 |
|
| 92 |
if ( $type < 8 ) { |
| 93 |
throw new InvalidDatabaseException( |
| 94 |
'Something went horribly wrong in the decoder. An extended type ' |
| 95 |
. 'resolved to a type number < 8 (' |
| 96 |
. $type |
| 97 |
. ')' |
| 98 |
); |
| 99 |
} |
| 100 |
|
| 101 |
++ $offset; |
| 102 |
} |
| 103 |
|
| 104 |
[ $size, $offset ] = $this->sizeFromCtrlByte( $ctrlByte, $offset ); |
| 105 |
|
| 106 |
return $this->decodeByType( $type, $offset, $size ); |
| 107 |
} |
| 108 |
|
| 109 |
private function decodeByType( int $type, int $offset, int $size ): array { |
| 110 |
switch ( $type ) { |
| 111 |
case self::_MAP: |
| 112 |
return $this->decodeMap( $size, $offset ); |
| 113 |
|
| 114 |
case self::_ARRAY: |
| 115 |
return $this->decodeArray( $size, $offset ); |
| 116 |
|
| 117 |
case self::_BOOLEAN: |
| 118 |
return [ $this->decodeBoolean( $size ), $offset ]; |
| 119 |
} |
| 120 |
|
| 121 |
$newOffset = $offset + $size; |
| 122 |
$bytes = Util::read( $this->fileStream, $offset, $size ); |
| 123 |
|
| 124 |
switch ( $type ) { |
| 125 |
case self::_BYTES: |
| 126 |
case self::_UTF8_STRING: |
| 127 |
return [ $bytes, $newOffset ]; |
| 128 |
|
| 129 |
case self::_DOUBLE: |
| 130 |
$this->verifySize( 8, $size ); |
| 131 |
|
| 132 |
return [ $this->decodeDouble( $bytes ), $newOffset ]; |
| 133 |
|
| 134 |
case self::_FLOAT: |
| 135 |
$this->verifySize( 4, $size ); |
| 136 |
|
| 137 |
return [ $this->decodeFloat( $bytes ), $newOffset ]; |
| 138 |
|
| 139 |
case self::_INT32: |
| 140 |
return [ $this->decodeInt32( $bytes, $size ), $newOffset ]; |
| 141 |
|
| 142 |
case self::_UINT16: |
| 143 |
case self::_UINT32: |
| 144 |
case self::_UINT64: |
| 145 |
case self::_UINT128: |
| 146 |
return [ $this->decodeUint( $bytes, $size ), $newOffset ]; |
| 147 |
|
| 148 |
default: |
| 149 |
throw new InvalidDatabaseException( |
| 150 |
'Unknown or unexpected type: ' . $type |
| 151 |
); |
| 152 |
} |
| 153 |
} |
| 154 |
|
| 155 |
private function verifySize( int $expected, int $actual ): void { |
| 156 |
if ( $expected !== $actual ) { |
| 157 |
throw new InvalidDatabaseException( |
| 158 |
"The MaxMind DB file's data section contains bad data (unknown data type or corrupt data)" |
| 159 |
); |
| 160 |
} |
| 161 |
} |
| 162 |
|
| 163 |
private function decodeArray( int $size, int $offset ): array { |
| 164 |
$array = []; |
| 165 |
|
| 166 |
for ( $i = 0; $i < $size; ++ $i ) { |
| 167 |
[ $value, $offset ] = $this->decode( $offset ); |
| 168 |
$array[] = $value; |
| 169 |
} |
| 170 |
|
| 171 |
return [ $array, $offset ]; |
| 172 |
} |
| 173 |
|
| 174 |
private function decodeBoolean( int $size ): bool { |
| 175 |
return $size !== 0; |
| 176 |
} |
| 177 |
|
| 178 |
private function decodeDouble( string $bytes ): float { |
| 179 |
// This assumes IEEE 754 doubles, but most (all?) modern platforms |
| 180 |
// use them. |
| 181 |
[ , $double ] = unpack( 'E', $bytes ); |
| 182 |
|
| 183 |
return $double; |
| 184 |
} |
| 185 |
|
| 186 |
private function decodeFloat( string $bytes ): float { |
| 187 |
// This assumes IEEE 754 floats, but most (all?) modern platforms |
| 188 |
// use them. |
| 189 |
[ , $float ] = unpack( 'G', $bytes ); |
| 190 |
|
| 191 |
return $float; |
| 192 |
} |
| 193 |
|
| 194 |
private function decodeInt32( string $bytes, int $size ): int { |
| 195 |
switch ( $size ) { |
| 196 |
case 0: |
| 197 |
return 0; |
| 198 |
|
| 199 |
case 1: |
| 200 |
case 2: |
| 201 |
case 3: |
| 202 |
$bytes = str_pad( $bytes, 4, "\x00", \STR_PAD_LEFT ); |
| 203 |
|
| 204 |
break; |
| 205 |
|
| 206 |
case 4: |
| 207 |
break; |
| 208 |
|
| 209 |
default: |
| 210 |
throw new InvalidDatabaseException( |
| 211 |
"The MaxMind DB file's data section contains bad data (unknown data type or corrupt data)" |
| 212 |
); |
| 213 |
} |
| 214 |
|
| 215 |
[ , $int ] = unpack( 'l', $this->maybeSwitchByteOrder( $bytes ) ); |
| 216 |
|
| 217 |
return $int; |
| 218 |
} |
| 219 |
|
| 220 |
private function decodeMap( int $size, int $offset ): array { |
| 221 |
$map = []; |
| 222 |
|
| 223 |
for ( $i = 0; $i < $size; ++ $i ) { |
| 224 |
[ $key, $offset ] = $this->decode( $offset ); |
| 225 |
[ $value, $offset ] = $this->decode( $offset ); |
| 226 |
$map[ $key ] = $value; |
| 227 |
} |
| 228 |
|
| 229 |
return [ $map, $offset ]; |
| 230 |
} |
| 231 |
|
| 232 |
private function decodePointer( int $ctrlByte, int $offset ): array { |
| 233 |
$pointerSize = ( ( $ctrlByte >> 3 ) & 0x3 ) + 1; |
| 234 |
|
| 235 |
$buffer = Util::read( $this->fileStream, $offset, $pointerSize ); |
| 236 |
$offset = $offset + $pointerSize; |
| 237 |
|
| 238 |
switch ( $pointerSize ) { |
| 239 |
case 1: |
| 240 |
$packed = \chr( $ctrlByte & 0x7 ) . $buffer; |
| 241 |
[ , $pointer ] = unpack( 'n', $packed ); |
| 242 |
$pointer += $this->pointerBase; |
| 243 |
|
| 244 |
break; |
| 245 |
|
| 246 |
case 2: |
| 247 |
$packed = "\x00" . \chr( $ctrlByte & 0x7 ) . $buffer; |
| 248 |
[ , $pointer ] = unpack( 'N', $packed ); |
| 249 |
$pointer += $this->pointerBase + 2048; |
| 250 |
|
| 251 |
break; |
| 252 |
|
| 253 |
case 3: |
| 254 |
$packed = \chr( $ctrlByte & 0x7 ) . $buffer; |
| 255 |
|
| 256 |
// It is safe to use 'N' here, even on 32 bit machines as the |
| 257 |
// first bit is 0. |
| 258 |
[ , $pointer ] = unpack( 'N', $packed ); |
| 259 |
$pointer += $this->pointerBase + 526336; |
| 260 |
|
| 261 |
break; |
| 262 |
|
| 263 |
case 4: |
| 264 |
// We cannot use unpack here as we might overflow on 32 bit |
| 265 |
// machines |
| 266 |
$pointerOffset = $this->decodeUint( $buffer, $pointerSize ); |
| 267 |
|
| 268 |
$pointerBase = $this->pointerBase; |
| 269 |
|
| 270 |
if ( \PHP_INT_MAX - $pointerBase >= $pointerOffset ) { |
| 271 |
$pointer = $pointerOffset + $pointerBase; |
| 272 |
} else { |
| 273 |
throw new \RuntimeException( |
| 274 |
'The database offset is too large to be represented on your platform.' |
| 275 |
); |
| 276 |
} |
| 277 |
|
| 278 |
break; |
| 279 |
|
| 280 |
default: |
| 281 |
throw new InvalidDatabaseException( |
| 282 |
'Unexpected pointer size ' . $pointerSize |
| 283 |
); |
| 284 |
} |
| 285 |
|
| 286 |
return [ $pointer, $offset ]; |
| 287 |
} |
| 288 |
|
| 289 |
// @phpstan-ignore-next-line |
| 290 |
private function decodeUint( string $bytes, int $byteLength ) { |
| 291 |
if ( $byteLength === 0 ) { |
| 292 |
return 0; |
| 293 |
} |
| 294 |
|
| 295 |
$integer = 0; |
| 296 |
|
| 297 |
// PHP integers are signed. PHP_INT_SIZE - 1 is the number of |
| 298 |
// complete bytes that can be converted to an integer. However, |
| 299 |
// we can convert another byte if the leading bit is zero. |
| 300 |
$useRealInts = $byteLength <= \PHP_INT_SIZE - 1 |
| 301 |
|| ( $byteLength === \PHP_INT_SIZE && ( \ord( $bytes[0] ) & 0x80 ) === 0 ); |
| 302 |
|
| 303 |
for ( $i = 0; $i < $byteLength; ++ $i ) { |
| 304 |
$part = \ord( $bytes[ $i ] ); |
| 305 |
|
| 306 |
// We only use gmp or bcmath if the final value is too big |
| 307 |
if ( $useRealInts ) { |
| 308 |
$integer = ( $integer << 8 ) + $part; |
| 309 |
} elseif ( \extension_loaded( 'gmp' ) ) { |
| 310 |
$integer = gmp_strval( gmp_add( gmp_mul( (string) $integer, '256' ), $part ) ); |
| 311 |
} elseif ( \extension_loaded( 'bcmath' ) ) { |
| 312 |
$integer = bcadd( bcmul( (string) $integer, '256' ), (string) $part ); |
| 313 |
} else { |
| 314 |
throw new \RuntimeException( |
| 315 |
'The gmp or bcmath extension must be installed to read this database.' |
| 316 |
); |
| 317 |
} |
| 318 |
} |
| 319 |
|
| 320 |
return $integer; |
| 321 |
} |
| 322 |
|
| 323 |
private function sizeFromCtrlByte( int $ctrlByte, int $offset ): array { |
| 324 |
$size = $ctrlByte & 0x1F; |
| 325 |
|
| 326 |
if ( $size < 29 ) { |
| 327 |
return [ $size, $offset ]; |
| 328 |
} |
| 329 |
|
| 330 |
$bytesToRead = $size - 28; |
| 331 |
$bytes = Util::read( $this->fileStream, $offset, $bytesToRead ); |
| 332 |
|
| 333 |
if ( $size === 29 ) { |
| 334 |
$size = 29 + \ord( $bytes ); |
| 335 |
} elseif ( $size === 30 ) { |
| 336 |
[ , $adjust ] = unpack( 'n', $bytes ); |
| 337 |
$size = 285 + $adjust; |
| 338 |
} else { |
| 339 |
[ , $adjust ] = unpack( 'N', "\x00" . $bytes ); |
| 340 |
$size = $adjust + 65821; |
| 341 |
} |
| 342 |
|
| 343 |
return [ $size, $offset + $bytesToRead ]; |
| 344 |
} |
| 345 |
|
| 346 |
private function maybeSwitchByteOrder( string $bytes ): string { |
| 347 |
return $this->switchByteOrder ? strrev( $bytes ) : $bytes; |
| 348 |
} |
| 349 |
|
| 350 |
private function isPlatformLittleEndian(): bool { |
| 351 |
$testint = 0x00FF; |
| 352 |
$packed = pack( 'S', $testint ); |
| 353 |
|
| 354 |
return $testint === current( unpack( 'v', $packed ) ); |
| 355 |
} |
| 356 |
} |
| 357 |
|
