Decoder.php
3 years ago
InvalidDatabaseException.php
3 years ago
Metadata.php
3 years ago
Util.php
3 years ago
Decoder.php
276 lines
| 1 | <?php |
| 2 | |
| 3 | declare (strict_types=1); |
| 4 | namespace IAWP\MaxMind\Db\Reader; |
| 5 | |
| 6 | // @codingStandardsIgnoreLine |
| 7 | use RuntimeException; |
| 8 | class Decoder |
| 9 | { |
| 10 | /** |
| 11 | * @var resource |
| 12 | */ |
| 13 | private $fileStream; |
| 14 | /** |
| 15 | * @var int |
| 16 | */ |
| 17 | private $pointerBase; |
| 18 | /** |
| 19 | * @var float |
| 20 | */ |
| 21 | private $pointerBaseByteSize; |
| 22 | /** |
| 23 | * This is only used for unit testing. |
| 24 | * |
| 25 | * @var bool |
| 26 | */ |
| 27 | private $pointerTestHack; |
| 28 | /** |
| 29 | * @var bool |
| 30 | */ |
| 31 | private $switchByteOrder; |
| 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 | private const _CONTAINER = 12; |
| 45 | private const _END_MARKER = 13; |
| 46 | private const _BOOLEAN = 14; |
| 47 | private const _FLOAT = 15; |
| 48 | /** |
| 49 | * @param resource $fileStream |
| 50 | */ |
| 51 | public function __construct($fileStream, int $pointerBase = 0, bool $pointerTestHack = \false) |
| 52 | { |
| 53 | $this->fileStream = $fileStream; |
| 54 | $this->pointerBase = $pointerBase; |
| 55 | $this->pointerBaseByteSize = $pointerBase > 0 ? \log($pointerBase, 2) / 8 : 0; |
| 56 | $this->pointerTestHack = $pointerTestHack; |
| 57 | $this->switchByteOrder = $this->isPlatformLittleEndian(); |
| 58 | } |
| 59 | public function decode(int $offset) : array |
| 60 | { |
| 61 | $ctrlByte = \ord(Util::read($this->fileStream, $offset, 1)); |
| 62 | ++$offset; |
| 63 | $type = $ctrlByte >> 5; |
| 64 | // Pointers are a special case, we don't read the next $size bytes, we |
| 65 | // use the size to determine the length of the pointer and then follow |
| 66 | // it. |
| 67 | if ($type === self::_POINTER) { |
| 68 | [$pointer, $offset] = $this->decodePointer($ctrlByte, $offset); |
| 69 | // for unit testing |
| 70 | if ($this->pointerTestHack) { |
| 71 | return [$pointer]; |
| 72 | } |
| 73 | [$result] = $this->decode($pointer); |
| 74 | return [$result, $offset]; |
| 75 | } |
| 76 | if ($type === self::_EXTENDED) { |
| 77 | $nextByte = \ord(Util::read($this->fileStream, $offset, 1)); |
| 78 | $type = $nextByte + 7; |
| 79 | if ($type < 8) { |
| 80 | throw new InvalidDatabaseException('Something went horribly wrong in the decoder. An extended type ' . 'resolved to a type number < 8 (' . $type . ')'); |
| 81 | } |
| 82 | ++$offset; |
| 83 | } |
| 84 | [$size, $offset] = $this->sizeFromCtrlByte($ctrlByte, $offset); |
| 85 | return $this->decodeByType($type, $offset, $size); |
| 86 | } |
| 87 | private function decodeByType(int $type, int $offset, int $size) : array |
| 88 | { |
| 89 | switch ($type) { |
| 90 | case self::_MAP: |
| 91 | return $this->decodeMap($size, $offset); |
| 92 | case self::_ARRAY: |
| 93 | return $this->decodeArray($size, $offset); |
| 94 | case self::_BOOLEAN: |
| 95 | return [$this->decodeBoolean($size), $offset]; |
| 96 | } |
| 97 | $newOffset = $offset + $size; |
| 98 | $bytes = Util::read($this->fileStream, $offset, $size); |
| 99 | switch ($type) { |
| 100 | case self::_BYTES: |
| 101 | case self::_UTF8_STRING: |
| 102 | return [$bytes, $newOffset]; |
| 103 | case self::_DOUBLE: |
| 104 | $this->verifySize(8, $size); |
| 105 | return [$this->decodeDouble($bytes), $newOffset]; |
| 106 | case self::_FLOAT: |
| 107 | $this->verifySize(4, $size); |
| 108 | return [$this->decodeFloat($bytes), $newOffset]; |
| 109 | case self::_INT32: |
| 110 | return [$this->decodeInt32($bytes, $size), $newOffset]; |
| 111 | case self::_UINT16: |
| 112 | case self::_UINT32: |
| 113 | case self::_UINT64: |
| 114 | case self::_UINT128: |
| 115 | return [$this->decodeUint($bytes, $size), $newOffset]; |
| 116 | default: |
| 117 | throw new InvalidDatabaseException('Unknown or unexpected type: ' . $type); |
| 118 | } |
| 119 | } |
| 120 | private function verifySize(int $expected, int $actual) : void |
| 121 | { |
| 122 | if ($expected !== $actual) { |
| 123 | throw new InvalidDatabaseException("The MaxMind DB file's data section contains bad data (unknown data type or corrupt data)"); |
| 124 | } |
| 125 | } |
| 126 | private function decodeArray(int $size, int $offset) : array |
| 127 | { |
| 128 | $array = []; |
| 129 | for ($i = 0; $i < $size; ++$i) { |
| 130 | [$value, $offset] = $this->decode($offset); |
| 131 | $array[] = $value; |
| 132 | } |
| 133 | return [$array, $offset]; |
| 134 | } |
| 135 | private function decodeBoolean(int $size) : bool |
| 136 | { |
| 137 | return $size !== 0; |
| 138 | } |
| 139 | private function decodeDouble(string $bytes) : float |
| 140 | { |
| 141 | // This assumes IEEE 754 doubles, but most (all?) modern platforms |
| 142 | // use them. |
| 143 | [, $double] = \unpack('E', $bytes); |
| 144 | return $double; |
| 145 | } |
| 146 | private function decodeFloat(string $bytes) : float |
| 147 | { |
| 148 | // This assumes IEEE 754 floats, but most (all?) modern platforms |
| 149 | // use them. |
| 150 | [, $float] = \unpack('G', $bytes); |
| 151 | return $float; |
| 152 | } |
| 153 | private function decodeInt32(string $bytes, int $size) : int |
| 154 | { |
| 155 | switch ($size) { |
| 156 | case 0: |
| 157 | return 0; |
| 158 | case 1: |
| 159 | case 2: |
| 160 | case 3: |
| 161 | $bytes = \str_pad($bytes, 4, "\x00", \STR_PAD_LEFT); |
| 162 | break; |
| 163 | case 4: |
| 164 | break; |
| 165 | default: |
| 166 | throw new InvalidDatabaseException("The MaxMind DB file's data section contains bad data (unknown data type or corrupt data)"); |
| 167 | } |
| 168 | [, $int] = \unpack('l', $this->maybeSwitchByteOrder($bytes)); |
| 169 | return $int; |
| 170 | } |
| 171 | private function decodeMap(int $size, int $offset) : array |
| 172 | { |
| 173 | $map = []; |
| 174 | for ($i = 0; $i < $size; ++$i) { |
| 175 | [$key, $offset] = $this->decode($offset); |
| 176 | [$value, $offset] = $this->decode($offset); |
| 177 | $map[$key] = $value; |
| 178 | } |
| 179 | return [$map, $offset]; |
| 180 | } |
| 181 | private function decodePointer(int $ctrlByte, int $offset) : array |
| 182 | { |
| 183 | $pointerSize = ($ctrlByte >> 3 & 0x3) + 1; |
| 184 | $buffer = Util::read($this->fileStream, $offset, $pointerSize); |
| 185 | $offset = $offset + $pointerSize; |
| 186 | switch ($pointerSize) { |
| 187 | case 1: |
| 188 | $packed = \chr($ctrlByte & 0x7) . $buffer; |
| 189 | [, $pointer] = \unpack('n', $packed); |
| 190 | $pointer += $this->pointerBase; |
| 191 | break; |
| 192 | case 2: |
| 193 | $packed = "\x00" . \chr($ctrlByte & 0x7) . $buffer; |
| 194 | [, $pointer] = \unpack('N', $packed); |
| 195 | $pointer += $this->pointerBase + 2048; |
| 196 | break; |
| 197 | case 3: |
| 198 | $packed = \chr($ctrlByte & 0x7) . $buffer; |
| 199 | // It is safe to use 'N' here, even on 32 bit machines as the |
| 200 | // first bit is 0. |
| 201 | [, $pointer] = \unpack('N', $packed); |
| 202 | $pointer += $this->pointerBase + 526336; |
| 203 | break; |
| 204 | case 4: |
| 205 | // We cannot use unpack here as we might overflow on 32 bit |
| 206 | // machines |
| 207 | $pointerOffset = $this->decodeUint($buffer, $pointerSize); |
| 208 | $pointerBase = $this->pointerBase; |
| 209 | if (\PHP_INT_MAX - $pointerBase >= $pointerOffset) { |
| 210 | $pointer = $pointerOffset + $pointerBase; |
| 211 | } else { |
| 212 | throw new RuntimeException('The database offset is too large to be represented on your platform.'); |
| 213 | } |
| 214 | break; |
| 215 | default: |
| 216 | throw new InvalidDatabaseException('Unexpected pointer size ' . $pointerSize); |
| 217 | } |
| 218 | return [$pointer, $offset]; |
| 219 | } |
| 220 | // @phpstan-ignore-next-line |
| 221 | private function decodeUint(string $bytes, int $byteLength) |
| 222 | { |
| 223 | if ($byteLength === 0) { |
| 224 | return 0; |
| 225 | } |
| 226 | $integer = 0; |
| 227 | // PHP integers are signed. PHP_INT_SIZE - 1 is the number of |
| 228 | // complete bytes that can be converted to an integer. However, |
| 229 | // we can convert another byte if the leading bit is zero. |
| 230 | $useRealInts = $byteLength <= \PHP_INT_SIZE - 1 || $byteLength === \PHP_INT_SIZE && (\ord($bytes[0]) & 0x80) === 0; |
| 231 | for ($i = 0; $i < $byteLength; ++$i) { |
| 232 | $part = \ord($bytes[$i]); |
| 233 | // We only use gmp or bcmath if the final value is too big |
| 234 | if ($useRealInts) { |
| 235 | $integer = ($integer << 8) + $part; |
| 236 | } elseif (\extension_loaded('gmp')) { |
| 237 | $integer = \gmp_strval(\gmp_add(\gmp_mul((string) $integer, '256'), $part)); |
| 238 | } elseif (\extension_loaded('bcmath')) { |
| 239 | $integer = \bcadd(\bcmul((string) $integer, '256'), (string) $part); |
| 240 | } else { |
| 241 | throw new RuntimeException('The gmp or bcmath extension must be installed to read this database.'); |
| 242 | } |
| 243 | } |
| 244 | return $integer; |
| 245 | } |
| 246 | private function sizeFromCtrlByte(int $ctrlByte, int $offset) : array |
| 247 | { |
| 248 | $size = $ctrlByte & 0x1f; |
| 249 | if ($size < 29) { |
| 250 | return [$size, $offset]; |
| 251 | } |
| 252 | $bytesToRead = $size - 28; |
| 253 | $bytes = Util::read($this->fileStream, $offset, $bytesToRead); |
| 254 | if ($size === 29) { |
| 255 | $size = 29 + \ord($bytes); |
| 256 | } elseif ($size === 30) { |
| 257 | [, $adjust] = \unpack('n', $bytes); |
| 258 | $size = 285 + $adjust; |
| 259 | } else { |
| 260 | [, $adjust] = \unpack('N', "\x00" . $bytes); |
| 261 | $size = $adjust + 65821; |
| 262 | } |
| 263 | return [$size, $offset + $bytesToRead]; |
| 264 | } |
| 265 | private function maybeSwitchByteOrder(string $bytes) : string |
| 266 | { |
| 267 | return $this->switchByteOrder ? \strrev($bytes) : $bytes; |
| 268 | } |
| 269 | private function isPlatformLittleEndian() : bool |
| 270 | { |
| 271 | $testint = 0xff; |
| 272 | $packed = \pack('S', $testint); |
| 273 | return $testint === \current(\unpack('v', $packed)); |
| 274 | } |
| 275 | } |
| 276 |