Reader.php
288 lines
| 1 | <?php |
| 2 | |
| 3 | declare (strict_types=1); |
| 4 | namespace IAWP\MaxMind\Db; |
| 5 | |
| 6 | use ArgumentCountError; |
| 7 | use BadMethodCallException; |
| 8 | use Exception; |
| 9 | use InvalidArgumentException; |
| 10 | use IAWP\MaxMind\Db\Reader\Decoder; |
| 11 | use IAWP\MaxMind\Db\Reader\InvalidDatabaseException; |
| 12 | use IAWP\MaxMind\Db\Reader\Metadata; |
| 13 | use IAWP\MaxMind\Db\Reader\Util; |
| 14 | use UnexpectedValueException; |
| 15 | /** |
| 16 | * Instances of this class provide a reader for the MaxMind DB format. IP |
| 17 | * addresses can be looked up using the get method. |
| 18 | */ |
| 19 | class Reader |
| 20 | { |
| 21 | /** |
| 22 | * @var int |
| 23 | */ |
| 24 | private static $DATA_SECTION_SEPARATOR_SIZE = 16; |
| 25 | /** |
| 26 | * @var string |
| 27 | */ |
| 28 | private static $METADATA_START_MARKER = "\xab\xcd\xefMaxMind.com"; |
| 29 | /** |
| 30 | * @var int |
| 31 | */ |
| 32 | private static $METADATA_START_MARKER_LENGTH = 14; |
| 33 | /** |
| 34 | * @var int |
| 35 | */ |
| 36 | private static $METADATA_MAX_SIZE = 131072; |
| 37 | // 128 * 1024 = 128KiB |
| 38 | /** |
| 39 | * @var Decoder |
| 40 | */ |
| 41 | private $decoder; |
| 42 | /** |
| 43 | * @var resource |
| 44 | */ |
| 45 | private $fileHandle; |
| 46 | /** |
| 47 | * @var int |
| 48 | */ |
| 49 | private $fileSize; |
| 50 | /** |
| 51 | * @var int |
| 52 | */ |
| 53 | private $ipV4Start; |
| 54 | /** |
| 55 | * @var Metadata |
| 56 | */ |
| 57 | private $metadata; |
| 58 | /** |
| 59 | * Constructs a Reader for the MaxMind DB format. The file passed to it must |
| 60 | * be a valid MaxMind DB file such as a GeoIp2 database file. |
| 61 | * |
| 62 | * @param string $database |
| 63 | * the MaxMind DB file to use |
| 64 | * |
| 65 | * @throws InvalidArgumentException for invalid database path or unknown arguments |
| 66 | * @throws InvalidDatabaseException |
| 67 | * if the database is invalid or there is an error reading |
| 68 | * from it |
| 69 | */ |
| 70 | public function __construct(string $database) |
| 71 | { |
| 72 | if (\func_num_args() !== 1) { |
| 73 | throw new ArgumentCountError(\sprintf('%s() expects exactly 1 parameter, %d given', __METHOD__, \func_num_args())); |
| 74 | } |
| 75 | $fileHandle = @\fopen($database, 'rb'); |
| 76 | if ($fileHandle === \false) { |
| 77 | throw new InvalidArgumentException("The file \"{$database}\" does not exist or is not readable."); |
| 78 | } |
| 79 | $this->fileHandle = $fileHandle; |
| 80 | $fileSize = @\filesize($database); |
| 81 | if ($fileSize === \false) { |
| 82 | throw new UnexpectedValueException("Error determining the size of \"{$database}\"."); |
| 83 | } |
| 84 | $this->fileSize = $fileSize; |
| 85 | $start = $this->findMetadataStart($database); |
| 86 | $metadataDecoder = new Decoder($this->fileHandle, $start); |
| 87 | [$metadataArray] = $metadataDecoder->decode($start); |
| 88 | $this->metadata = new Metadata($metadataArray); |
| 89 | $this->decoder = new Decoder($this->fileHandle, $this->metadata->searchTreeSize + self::$DATA_SECTION_SEPARATOR_SIZE); |
| 90 | $this->ipV4Start = $this->ipV4StartNode(); |
| 91 | } |
| 92 | /** |
| 93 | * Retrieves the record for the IP address. |
| 94 | * |
| 95 | * @param string $ipAddress |
| 96 | * the IP address to look up |
| 97 | * |
| 98 | * @throws BadMethodCallException if this method is called on a closed database |
| 99 | * @throws InvalidArgumentException if something other than a single IP address is passed to the method |
| 100 | * @throws InvalidDatabaseException |
| 101 | * if the database is invalid or there is an error reading |
| 102 | * from it |
| 103 | * |
| 104 | * @return mixed the record for the IP address |
| 105 | */ |
| 106 | public function get(string $ipAddress) |
| 107 | { |
| 108 | if (\func_num_args() !== 1) { |
| 109 | throw new ArgumentCountError(\sprintf('%s() expects exactly 1 parameter, %d given', __METHOD__, \func_num_args())); |
| 110 | } |
| 111 | [$record] = $this->getWithPrefixLen($ipAddress); |
| 112 | return $record; |
| 113 | } |
| 114 | /** |
| 115 | * Retrieves the record for the IP address and its associated network prefix length. |
| 116 | * |
| 117 | * @param string $ipAddress |
| 118 | * the IP address to look up |
| 119 | * |
| 120 | * @throws BadMethodCallException if this method is called on a closed database |
| 121 | * @throws InvalidArgumentException if something other than a single IP address is passed to the method |
| 122 | * @throws InvalidDatabaseException |
| 123 | * if the database is invalid or there is an error reading |
| 124 | * from it |
| 125 | * |
| 126 | * @return array an array where the first element is the record and the |
| 127 | * second the network prefix length for the record |
| 128 | */ |
| 129 | public function getWithPrefixLen(string $ipAddress) : array |
| 130 | { |
| 131 | if (\func_num_args() !== 1) { |
| 132 | throw new ArgumentCountError(\sprintf('%s() expects exactly 1 parameter, %d given', __METHOD__, \func_num_args())); |
| 133 | } |
| 134 | if (!\is_resource($this->fileHandle)) { |
| 135 | throw new BadMethodCallException('Attempt to read from a closed MaxMind DB.'); |
| 136 | } |
| 137 | [$pointer, $prefixLen] = $this->findAddressInTree($ipAddress); |
| 138 | if ($pointer === 0) { |
| 139 | return [null, $prefixLen]; |
| 140 | } |
| 141 | return [$this->resolveDataPointer($pointer), $prefixLen]; |
| 142 | } |
| 143 | private function findAddressInTree(string $ipAddress) : array |
| 144 | { |
| 145 | $packedAddr = @\inet_pton($ipAddress); |
| 146 | if ($packedAddr === \false) { |
| 147 | throw new InvalidArgumentException("The value \"{$ipAddress}\" is not a valid IP address."); |
| 148 | } |
| 149 | $rawAddress = \unpack('C*', $packedAddr); |
| 150 | $bitCount = \count($rawAddress) * 8; |
| 151 | // The first node of the tree is always node 0, at the beginning of the |
| 152 | // value |
| 153 | $node = 0; |
| 154 | $metadata = $this->metadata; |
| 155 | // Check if we are looking up an IPv4 address in an IPv6 tree. If this |
| 156 | // is the case, we can skip over the first 96 nodes. |
| 157 | if ($metadata->ipVersion === 6) { |
| 158 | if ($bitCount === 32) { |
| 159 | $node = $this->ipV4Start; |
| 160 | } |
| 161 | } elseif ($metadata->ipVersion === 4 && $bitCount === 128) { |
| 162 | throw new InvalidArgumentException("Error looking up {$ipAddress}. You attempted to look up an" . ' IPv6 address in an IPv4-only database.'); |
| 163 | } |
| 164 | $nodeCount = $metadata->nodeCount; |
| 165 | for ($i = 0; $i < $bitCount && $node < $nodeCount; ++$i) { |
| 166 | $tempBit = 0xff & $rawAddress[($i >> 3) + 1]; |
| 167 | $bit = 1 & $tempBit >> 7 - $i % 8; |
| 168 | $node = $this->readNode($node, $bit); |
| 169 | } |
| 170 | if ($node === $nodeCount) { |
| 171 | // Record is empty |
| 172 | return [0, $i]; |
| 173 | } |
| 174 | if ($node > $nodeCount) { |
| 175 | // Record is a data pointer |
| 176 | return [$node, $i]; |
| 177 | } |
| 178 | throw new InvalidDatabaseException('Invalid or corrupt database. Maximum search depth reached without finding a leaf node'); |
| 179 | } |
| 180 | private function ipV4StartNode() : int |
| 181 | { |
| 182 | // If we have an IPv4 database, the start node is the first node |
| 183 | if ($this->metadata->ipVersion === 4) { |
| 184 | return 0; |
| 185 | } |
| 186 | $node = 0; |
| 187 | for ($i = 0; $i < 96 && $node < $this->metadata->nodeCount; ++$i) { |
| 188 | $node = $this->readNode($node, 0); |
| 189 | } |
| 190 | return $node; |
| 191 | } |
| 192 | private function readNode(int $nodeNumber, int $index) : int |
| 193 | { |
| 194 | $baseOffset = $nodeNumber * $this->metadata->nodeByteSize; |
| 195 | switch ($this->metadata->recordSize) { |
| 196 | case 24: |
| 197 | $bytes = Util::read($this->fileHandle, $baseOffset + $index * 3, 3); |
| 198 | [, $node] = \unpack('N', "\x00" . $bytes); |
| 199 | return $node; |
| 200 | case 28: |
| 201 | $bytes = Util::read($this->fileHandle, $baseOffset + 3 * $index, 4); |
| 202 | if ($index === 0) { |
| 203 | $middle = (0xf0 & \ord($bytes[3])) >> 4; |
| 204 | } else { |
| 205 | $middle = 0xf & \ord($bytes[0]); |
| 206 | } |
| 207 | [, $node] = \unpack('N', \chr($middle) . \substr($bytes, $index, 3)); |
| 208 | return $node; |
| 209 | case 32: |
| 210 | $bytes = Util::read($this->fileHandle, $baseOffset + $index * 4, 4); |
| 211 | [, $node] = \unpack('N', $bytes); |
| 212 | return $node; |
| 213 | default: |
| 214 | throw new InvalidDatabaseException('Unknown record size: ' . $this->metadata->recordSize); |
| 215 | } |
| 216 | } |
| 217 | /** |
| 218 | * @return mixed |
| 219 | */ |
| 220 | private function resolveDataPointer(int $pointer) |
| 221 | { |
| 222 | $resolved = $pointer - $this->metadata->nodeCount + $this->metadata->searchTreeSize; |
| 223 | if ($resolved >= $this->fileSize) { |
| 224 | throw new InvalidDatabaseException("The MaxMind DB file's search tree is corrupt"); |
| 225 | } |
| 226 | [$data] = $this->decoder->decode($resolved); |
| 227 | return $data; |
| 228 | } |
| 229 | /* |
| 230 | * This is an extremely naive but reasonably readable implementation. There |
| 231 | * are much faster algorithms (e.g., Boyer-Moore) for this if speed is ever |
| 232 | * an issue, but I suspect it won't be. |
| 233 | */ |
| 234 | private function findMetadataStart(string $filename) : int |
| 235 | { |
| 236 | $handle = $this->fileHandle; |
| 237 | $fstat = \fstat($handle); |
| 238 | $fileSize = $fstat['size']; |
| 239 | $marker = self::$METADATA_START_MARKER; |
| 240 | $markerLength = self::$METADATA_START_MARKER_LENGTH; |
| 241 | $minStart = $fileSize - \min(self::$METADATA_MAX_SIZE, $fileSize); |
| 242 | for ($offset = $fileSize - $markerLength; $offset >= $minStart; --$offset) { |
| 243 | if (\fseek($handle, $offset) !== 0) { |
| 244 | break; |
| 245 | } |
| 246 | $value = \fread($handle, $markerLength); |
| 247 | if ($value === $marker) { |
| 248 | return $offset + $markerLength; |
| 249 | } |
| 250 | } |
| 251 | throw new InvalidDatabaseException("Error opening database file ({$filename}). " . 'Is this a valid MaxMind DB file?'); |
| 252 | } |
| 253 | /** |
| 254 | * @throws InvalidArgumentException if arguments are passed to the method |
| 255 | * @throws BadMethodCallException if the database has been closed |
| 256 | * |
| 257 | * @return Metadata object for the database |
| 258 | */ |
| 259 | public function metadata() : Metadata |
| 260 | { |
| 261 | if (\func_num_args()) { |
| 262 | throw new ArgumentCountError(\sprintf('%s() expects exactly 0 parameters, %d given', __METHOD__, \func_num_args())); |
| 263 | } |
| 264 | // Not technically required, but this makes it consistent with |
| 265 | // C extension and it allows us to change our implementation later. |
| 266 | if (!\is_resource($this->fileHandle)) { |
| 267 | throw new BadMethodCallException('Attempt to read from a closed MaxMind DB.'); |
| 268 | } |
| 269 | return clone $this->metadata; |
| 270 | } |
| 271 | /** |
| 272 | * Closes the MaxMind DB and returns resources to the system. |
| 273 | * |
| 274 | * @throws Exception |
| 275 | * if an I/O error occurs |
| 276 | */ |
| 277 | public function close() : void |
| 278 | { |
| 279 | if (\func_num_args()) { |
| 280 | throw new ArgumentCountError(\sprintf('%s() expects exactly 0 parameters, %d given', __METHOD__, \func_num_args())); |
| 281 | } |
| 282 | if (!\is_resource($this->fileHandle)) { |
| 283 | throw new BadMethodCallException('Attempt to close a closed MaxMind DB.'); |
| 284 | } |
| 285 | \fclose($this->fileHandle); |
| 286 | } |
| 287 | } |
| 288 |