includes/IP.php

   1 <?php
   2 /*
   3  * @Author "Ashar Voultoiz" <hashar@altern.org>
   4  * @License GPL v2 or later
   5  */
   6
   7 // Some regex definition to "play" with IP address and IP address blocks
   8
   9 // An IP is made of 4 bytes from x00 to xFF which is d0 to d255
  10 define( 'RE_IP_BYTE', '(25[0-5]|2[0-4][0-9]|1[0-9][0-9]|0?[0-9]?[0-9])');
  11 define( 'RE_IP_ADD' , RE_IP_BYTE . '\.' . RE_IP_BYTE . '\.' . RE_IP_BYTE . '\.' . RE_IP_BYTE );
  12 // An IPv4 block is an IP address and a prefix (d1 to d32)
  13 define( 'RE_IP_PREFIX', '(3[0-2]|[12]?\d)');
  14 define( 'RE_IP_BLOCK', RE_IP_ADD . '\/' . RE_IP_PREFIX);
  15 // For IPv6 canonicalization (NOT for strict validation; these are quite lax!)
  16 define( 'RE_IPV6_WORD', '([0-9A-Fa-f]{1,4})' );
  17 define( 'RE_IPV6_GAP', ':(?:0+:)*(?::(?:0+:)*)?' );
  18 define( 'RE_IPV6_V4_PREFIX', '0*' . RE_IPV6_GAP . '(?:ffff:)?' );
  19 // An IPv6 block is an IP address and a prefix (d1 to d128)
  20 define( 'RE_IPV6_PREFIX', '(12[0-8]|1[01][0-9]|[1-9]?\d)');
  21 // An IPv6 IP is made up of 8 octets. However abbreviations like "::" can be used. This is lax!
  22 define( 'RE_IPV6_ADD', '(:(:' . RE_IPV6_WORD . '){1,7}|' . RE_IPV6_WORD . '(:{1,2}' . RE_IPV6_WORD . '|::$){1,7})' );
  23 define( 'RE_IPV6_BLOCK', RE_IPV6_ADD . '\/' . RE_IPV6_PREFIX );
  24 // This might be useful for regexps used elsewhere, matches any IPv6 or IPv6 address or network
  25 define( 'IP_ADDRESS_STRING',
  26         '(?:' .
  27                 RE_IP_ADD . '(\/' . RE_IP_PREFIX . '|)' .
  28         '|' .
  29                 RE_IPV6_ADD . '(\/' . RE_IPV6_PREFIX . '|)' .
  30         ')' );
  31
  32 /**
  33  * A collection of public static functions to play with IP address
  34  * and IP blocks.
  35  */
  36 class IP {
  37         /**
  38          * Given a string, determine if it as valid IP
  39          * Unlike isValid(), this looks for networks too
  40          * @param $ip IP address.
  41          * @return string
  42          */
  43         public static function isIPAddress( $ip ) {
  44                 if ( !$ip ) return false;
  45                 if ( is_array( $ip ) ) {
  46                   throw new MWException( "invalid value passed to " . __METHOD__ );
  47                 }
  48                 // IPv6 IPs with two "::" strings are ambiguous and thus invalid
  49                 return preg_match( '/^' . IP_ADDRESS_STRING . '$/', $ip) && ( substr_count($ip, '::') < 2 );
  50         }
  51
  52         public static function isIPv6( $ip ) {
  53                 if ( !$ip ) return false;
  54                 if( is_array( $ip ) ) {
  55                   throw new MWException( "invalid value passed to " . __METHOD__ );
  56                 }
  57                 // IPv6 IPs with two "::" strings are ambiguous and thus invalid
  58                 return preg_match( '/^' . RE_IPV6_ADD . '(\/' . RE_IPV6_PREFIX . '|)$/', $ip) && ( substr_count($ip, '::') < 2);
  59         }
  60
  61         public static function isIPv4( $ip ) {
  62                 if ( !$ip ) return false;
  63                 return preg_match( '/^' . RE_IP_ADD . '(\/' . RE_IP_PREFIX . '|)$/', $ip);
  64         }
  65
  66         /**
  67          * Given an IP address in dotted-quad notation, returns an IPv6 octet.
  68          * See http://www.answers.com/topic/ipv4-compatible-address
  69          * IPs with the first 92 bits as zeros are reserved from IPv6
  70          * @param $ip quad-dotted IP address.
  71          * @return string
  72          */
  73         public static function IPv4toIPv6( $ip ) {
  74                 if ( !$ip ) return null;
  75                 // Convert only if needed
  76                 if ( self::isIPv6( $ip ) ) return $ip;
  77                 // IPv4 CIDRs
  78                 if ( strpos( $ip, '/' ) !== false ) {
  79                         $parts = explode( '/', $ip, 2 );
  80                         if ( count( $parts ) != 2 ) {
  81                                 return false;
  82                         }
  83                         $network = self::toUnsigned( $parts[0] );
  84                         if ( $network !== false && is_numeric( $parts[1] ) && $parts[1] >= 0 && $parts[1] <= 32 ) {
  85                                 $bits = $parts[1] + 96;
  86                                 return self::toOctet( $network ) . "/$bits";
  87                         } else {
  88                                 return false;
  89                         }
  90                 }
  91                 return self::toOctet( self::toUnsigned( $ip ) );
  92         }
  93
  94         /**
  95          * Given an IPv6 address in octet notation, returns an unsigned integer.
  96          * @param $ip octet ipv6 IP address.
  97          * @return string
  98          */
  99         public static function toUnsigned6( $ip ) {
 100                 if ( !$ip ) return null;
 101         $ip = explode(':', self::sanitizeIP( $ip ) );
 102         $r_ip = '';
 103         foreach ($ip as $v) {
 104                 $r_ip .= str_pad( $v, 4, 0, STR_PAD_LEFT );
 105         }
 106         $r_ip = wfBaseConvert( $r_ip, 16, 10 );
 107         return $r_ip;
 108         }
 109
 110         /**
 111          * Given an IPv6 address in octet notation, returns the expanded octet.
 112          * IPv4 IPs will be trimmed, thats it...
 113          * @param $ip octet ipv6 IP address.
 114          * @return string
 115          */
 116         public static function sanitizeIP( $ip ) {
 117                 $ip = trim( $ip );
 118                 if ( $ip === '' ) return null;
 119                 // Trim and return IPv4 addresses
 120                 if ( self::isIPv4($ip) ) return $ip;
 121                 // Only IPv6 addresses can be expanded
 122                 if ( !self::isIPv6($ip) ) return $ip;
 123                 // Remove any whitespaces, convert to upper case
 124                 $ip = strtoupper( $ip );
 125                 // Expand zero abbreviations
 126                 if ( strpos( $ip, '::' ) !== false ) {
 127                 $ip = str_replace('::', str_repeat(':0', 8 - substr_count($ip, ':')) . ':', $ip);
 128         }
 129         // For IPs that start with "::", correct the final IP so that it starts with '0' and not ':'
 130         if ( $ip[0] == ':' ) $ip = "0$ip";
 131         // Remove leading zereos from each bloc as needed
 132         $ip = preg_replace( '/(^|:)0+' . RE_IPV6_WORD . '/', '$1$2', $ip );
 133         return $ip;
 134         }
 135
 136         /**
 137          * Given an unsigned integer, returns an IPv6 address in octet notation
 138          * @param $ip integer IP address.
 139          * @return string
 140          */
 141         public static function toOctet( $ip_int ) {
 142                 // Convert to padded uppercase hex
 143                 $ip_hex = wfBaseConvert($ip_int, 10, 16, 32, false);
 144                 // Separate into 8 octets
 145                 $ip_oct = substr( $ip_hex, 0, 4 );
 146                 for ($n=1; $n < 8; $n++) {
 147                         $ip_oct .= ':' . substr($ip_hex, 4*$n, 4);
 148                 }
 149                 // NO leading zeroes
 150                 $ip_oct = preg_replace( '/(^|:)0+' . RE_IPV6_WORD . '/', '$1$2', $ip_oct );
 151         return $ip_oct;
 152         }
 153
 154         /**
 155          * Given a hexadecimal number, returns to an IPv6 address in octet notation
 156          * @param $ip string hex IP
 157          * @return string
 158          */
 159         public static function HextoOctet( $ip_hex ) {
 160                 // Convert to padded uppercase hex
 161                 $ip_hex = str_pad( strtoupper($ip_hex), 32, '0');
 162                 // Separate into 8 octets
 163                 $ip_oct = substr( $ip_hex, 0, 4 );
 164                 for ($n=1; $n < 8; $n++) {
 165                         $ip_oct .= ':' . substr($ip_hex, 4*$n, 4);
 166                 }
 167                 // NO leading zeroes
 168                 $ip_oct = preg_replace( '/(^|:)0+' . RE_IPV6_WORD . '/', '$1$2', $ip_oct );
 169         return $ip_oct;
 170         }
 171
 172         /**
 173          * Converts a hexadecimal number to an IPv4 address in octet notation
 174          * @param $ip string Hex IP
 175          * @return string
 176          */
 177         public static function hexToQuad( $ip ) {
 178                 // Converts a hexadecimal IP to nnn.nnn.nnn.nnn format
 179                 $dec = wfBaseConvert( $ip, 16, 10 );
 180                 $parts[3] = $dec % 256;
 181                 $dec /= 256;
 182                 $parts[2] = $dec % 256;
 183                 $dec /= 256;
 184                 $parts[1] = $dec % 256;
 185                 $parts[0] = $dec / 256;
 186                 return implode( '.', array_reverse( $parts ) );
 187         }
 188
 189         /**
 190          * Convert a network specification in IPv6 CIDR notation to an integer network and a number of bits
 191          * @return array(string, int)
 192          */
 193         public static function parseCIDR6( $range ) {
 194                 # Expand any IPv6 IP
 195                 $parts = explode( '/', IP::sanitizeIP( $range ), 2 );
 196                 if ( count( $parts ) != 2 ) {
 197                         return array( false, false );
 198                 }
 199                 $network = self::toUnsigned6( $parts[0] );
 200                 if ( $network !== false && is_numeric( $parts[1] ) && $parts[1] >= 0 && $parts[1] <= 128 ) {
 201                         $bits = $parts[1];
 202                         if ( $bits == 0 ) {
 203                                 $network = 0;
 204                         } else {
 205                         # Native 32 bit functions WONT work here!!!
 206                         # Convert to a padded binary number
 207                                 $network = wfBaseConvert( $network, 10, 2, 128 );
 208                         # Truncate the last (128-$bits) bits and replace them with zeros
 209                                 $network = str_pad( substr( $network, 0, $bits ), 128, 0, STR_PAD_RIGHT );
 210                         # Convert back to an integer
 211                                 $network = wfBaseConvert( $network, 2, 10 );
 212                         }
 213                 } else {
 214                         $network = false;
 215                         $bits = false;
 216                 }
 217                 return array( $network, $bits );
 218         }
 219
 220         /**
 221          * Given a string range in a number of formats, return the start and end of
 222          * the range in hexadecimal. For IPv6.
 223          *
 224          * Formats are:
 225          *     2001:0db8:85a3::7344/96                                   CIDR
 226          *     2001:0db8:85a3::7344 - 2001:0db8:85a3::7344   Explicit range
 227          *     2001:0db8:85a3::7344/96                                   Single IP
 228          * @return array(string, int)
 229          */
 230         public static function parseRange6( $range ) {
 231                 # Expand any IPv6 IP
 232                 $range = IP::sanitizeIP( $range );
 233                 if ( strpos( $range, '/' ) !== false ) {
 234                         # CIDR
 235                         list( $network, $bits ) = self::parseCIDR6( $range );
 236                         if ( $network === false ) {
 237                                 $start = $end = false;
 238                         } else {
 239                                 $start = wfBaseConvert( $network, 10, 16, 32, false );
 240                                 # Turn network to binary (again)
 241                                 $end = wfBaseConvert( $network, 10, 2, 128 );
 242                                 # Truncate the last (128-$bits) bits and replace them with ones
 243                                 $end = str_pad( substr( $end, 0, $bits ), 128, 1, STR_PAD_RIGHT );
 244                                 # Convert to hex
 245                                 $end = wfBaseConvert( $end, 2, 16, 32, false );
 246                                 # see toHex() comment
 247                                 $start = "v6-$start"; $end = "v6-$end";
 248                         }
 249                 } elseif ( strpos( $range, '-' ) !== false ) {
 250                         # Explicit range
 251                         list( $start, $end ) = array_map( 'trim', explode( '-', $range, 2 ) );
 252                         $start = self::toUnsigned6( $start ); $end = self::toUnsigned6( $end );
 253                         if ( $start > $end ) {
 254                                 $start = $end = false;
 255                         } else {
 256                                 $start = wfBaseConvert( $start, 10, 16, 32, false );
 257                                 $end = wfBaseConvert( $end, 10, 16, 32, false );
 258                         }
 259                         # see toHex() comment
 260                         $start = "v6-$start"; $end = "v6-$end";
 261                 } else {
 262                         # Single IP
 263                         $start = $end = self::toHex( $range );
 264                 }
 265                 if ( $start === false || $end === false ) {
 266                         return array( false, false );
 267                 } else {
 268                         return array( $start, $end );
 269                 }
 270     }
 271
 272         /**
 273          * Validate an IP address.
 274          * @return boolean True if it is valid.
 275          */
 276         public static function isValid( $ip ) {
 277                 return ( preg_match( '/^' . RE_IP_ADD . '$/', $ip) || preg_match( '/^' . RE_IPV6_ADD . '$/', $ip) );
 278         }
 279
 280         /**
 281          * Validate an IP Block.
 282          * @return boolean True if it is valid.
 283          */
 284         public static function isValidBlock( $ipblock ) {
 285                 return ( count(self::toArray($ipblock)) == 1 + 5 );
 286         }
 287
 288         /**
 289          * Determine if an IP address really is an IP address, and if it is public,
 290          * i.e. not RFC 1918 or similar
 291          * Comes from ProxyTools.php
 292          */
 293         public static function isPublic( $ip ) {
 294                 $n = self::toUnsigned( $ip );
 295                 if ( !$n ) {
 296                         return false;
 297                 }
 298
 299                 // ip2long accepts incomplete addresses, as well as some addresses
 300                 // followed by garbage characters. Check that it's really valid.
 301                 if( $ip != long2ip( $n ) ) {
 302                         return false;
 303                 }
 304
 305                 static $privateRanges = false;
 306                 if ( !$privateRanges ) {
 307                         $privateRanges = array(
 308                                 array( '10.0.0.0',    '10.255.255.255' ),   # RFC 1918 (private)
 309                                 array( '172.16.0.0',  '172.31.255.255' ),   #     "
 310                                 array( '192.168.0.0', '192.168.255.255' ),  #     "
 311                                 array( '0.0.0.0',     '0.255.255.255' ),    # this network
 312                                 array( '127.0.0.0',   '127.255.255.255' ),  # loopback
 313                         );
 314                 }
 315
 316                 foreach ( $privateRanges as $r ) {
 317                         $start = self::toUnsigned( $r[0] );
 318                         $end = self::toUnsigned( $r[1] );
 319                         if ( $n >= $start && $n <= $end ) {
 320                                 return false;
 321                         }
 322                 }
 323                 return true;
 324         }
 325
 326         /**
 327          * Split out an IP block as an array of 4 bytes and a mask,
 328          * return false if it can't be determined
 329          *
 330          * @param $ip string A quad dotted/octet IP address
 331          * @return array
 332          */
 333         public static function toArray( $ipblock ) {
 334                 $matches = array();
 335                 if( preg_match( '/^' . RE_IP_ADD . '(?:\/(?:'.RE_IP_PREFIX.'))?' . '$/', $ipblock, $matches ) ) {
 336                         return $matches;
 337                 } else if ( preg_match( '/^' . RE_IPV6_ADD . '(?:\/(?:'.RE_IPV6_PREFIX.'))?' . '$/', $ipblock, $matches ) ) {
 338                         return $matches;
 339                 } else {
 340                         return false;
 341                 }
 342         }
 343
 344         /**
 345          * Return a zero-padded hexadecimal representation of an IP address.
 346          *
 347          * Hexadecimal addresses are used because they can easily be extended to
 348          * IPv6 support. To separate the ranges, the return value from this
 349          * function for an IPv6 address will be prefixed with "v6-", a non-
 350          * hexadecimal string which sorts after the IPv4 addresses.
 351          *
 352          * @param $ip Quad dotted/octet IP address.
 353          * @return hexidecimal
 354          */
 355         public static function toHex( $ip ) {
 356                 $n = self::toUnsigned( $ip );
 357                 if ( $n !== false ) {
 358                         $n = self::isIPv6($ip) ? "v6-" . wfBaseConvert( $n, 10, 16, 32, false ) : wfBaseConvert( $n, 10, 16, 8, false );
 359                 }
 360                 return $n;
 361         }
 362
 363         /**
 364          * Given an IP address in dotted-quad/octet notation, returns an unsigned integer.
 365          * Like ip2long() except that it actually works and has a consistent error return value.
 366          * Comes from ProxyTools.php
 367          * @param $ip Quad dotted IP address.
 368          * @return integer
 369          */
 370         public static function toUnsigned( $ip ) {
 371                 // Use IPv6 functions if needed
 372                 if ( self::isIPv6( $ip ) ) {
 373                         return self::toUnsigned6( $ip );
 374                 }
 375                 if ( $ip == '255.255.255.255' ) {
 376                         $n = -1;
 377                 } else {
 378                         $n = ip2long( $ip );
 379                         if ( $n == -1 || $n === false ) { # Return value on error depends on PHP version
 380                                 $n = false;
 381                         }
 382                 }
 383                 if ( $n < 0 ) {
 384                         $n += pow( 2, 32 );
 385                 }
 386                 return $n;
 387         }
 388
 389         /**
 390          * Convert a dotted-quad IP to a signed integer
 391          * Returns false on failure
 392          */
 393         public static function toSigned( $ip ) {
 394                 if ( $ip == '255.255.255.255' ) {
 395                         $n = -1;
 396                 } else {
 397                         $n = ip2long( $ip );
 398                         if ( $n == -1 ) {
 399                                 $n = false;
 400                         }
 401                 }
 402                 return $n;
 403         }
 404
 405         /**
 406          * Convert a network specification in CIDR notation to an integer network and a number of bits
 407          * @return array(string, int)
 408          */
 409         public static function parseCIDR( $range ) {
 410                 $parts = explode( '/', $range, 2 );
 411                 if ( count( $parts ) != 2 ) {
 412                         return array( false, false );
 413                 }
 414                 $network = self::toSigned( $parts[0] );
 415                 if ( $network !== false && is_numeric( $parts[1] ) && $parts[1] >= 0 && $parts[1] <= 32 ) {
 416                         $bits = $parts[1];
 417                         if ( $bits == 0 ) {
 418                                 $network = 0;
 419                         } else {
 420                                 $network &= ~((1 << (32 - $bits)) - 1);
 421                         }
 422                         # Convert to unsigned
 423                         if ( $network < 0 ) {
 424                                 $network += pow( 2, 32 );
 425                         }
 426                 } else {
 427                         $network = false;
 428                         $bits = false;
 429                 }
 430                 return array( $network, $bits );
 431         }
 432
 433         /**
 434          * Given a string range in a number of formats, return the start and end of
 435          * the range in hexadecimal.
 436          *
 437          * Formats are:
 438          *     1.2.3.4/24          CIDR
 439          *     1.2.3.4 - 1.2.3.5   Explicit range
 440          *     1.2.3.4             Single IP
 441          *
 442          *     2001:0db8:85a3::7344/96                                   CIDR
 443          *     2001:0db8:85a3::7344 - 2001:0db8:85a3::7344   Explicit range
 444          *     2001:0db8:85a3::7344                                      Single IP
 445          * @return array(string, int)
 446          */
 447         public static function parseRange( $range ) {
 448                 // Use IPv6 functions if needed
 449                 if ( self::isIPv6( $range ) ) {
 450                         return self::parseRange6( $range );
 451                 }
 452                 if ( strpos( $range, '/' ) !== false ) {
 453                         # CIDR
 454                         list( $network, $bits ) = self::parseCIDR( $range );
 455                         if ( $network === false ) {
 456                                 $start = $end = false;
 457                         } else {
 458                                 $start = sprintf( '%08X', $network );
 459                                 $end = sprintf( '%08X', $network + pow( 2, (32 - $bits) ) - 1 );
 460                         }
 461                 } elseif ( strpos( $range, '-' ) !== false ) {
 462                         # Explicit range
 463                         list( $start, $end ) = array_map( 'trim', explode( '-', $range, 2 ) );
 464                         if( self::isIPAddress( $start ) && self::isIPAddress( $end ) ) {
 465                                 $start = self::toUnsigned( $start ); $end = self::toUnsigned( $end );
 466                                 if ( $start > $end ) {
 467                                         $start = $end = false;
 468                                 } else {
 469                                         $start = sprintf( '%08X', $start );
 470                                         $end = sprintf( '%08X', $end );
 471                                 }
 472                         } else {
 473                                 $start = $end = false;
 474                         }
 475                 } else {
 476                         # Single IP
 477                         $start = $end = self::toHex( $range );
 478                 }
 479                 if ( $start === false || $end === false ) {
 480                         return array( false, false );
 481                 } else {
 482                         return array( $start, $end );
 483                 }
 484     }
 485
 486     /**
 487      * Determine if a given IPv4/IPv6 address is in a given CIDR network
 488      * @param $addr The address to check against the given range.
 489      * @param $range The range to check the given address against.
 490      * @return bool Whether or not the given address is in the given range.
 491      */
 492     public static function isInRange( $addr, $range ) {
 493     // Convert to IPv6 if needed
 494         $unsignedIP = self::toHex( $addr );
 495         list( $start, $end ) = self::parseRange( $range );
 496         return (($unsignedIP >= $start) && ($unsignedIP <= $end));
 497     }
 498
 499     /**
 500      * Convert some unusual representations of IPv4 addresses to their
 501      * canonical dotted quad representation.
 502      *
 503      * This currently only checks a few IPV4-to-IPv6 related cases.  More
 504      * unusual representations may be added later.
 505      *
 506      * @param $addr something that might be an IP address
 507      * @return valid dotted quad IPv4 address or null
 508      */
 509     public static function canonicalize( $addr ) {
 510                 if ( self::isValid( $addr ) )
 511                         return $addr;
 512
 513                 // Annoying IPv6 representations like ::ffff:1.2.3.4
 514                 if ( strpos($addr,':') !==false && strpos($addr,'.') !==false ) {
 515                         $addr = str_replace( '.', ':', $addr );
 516                         if( IP::isIPv6( $addr ) )
 517                                 return $addr;
 518                 }
 519
 520                 // IPv6 loopback address
 521                 $m = array();
 522                 if ( preg_match( '/^0*' . RE_IPV6_GAP . '1$/', $addr, $m ) )
 523                         return '127.0.0.1';
 524
 525                 // IPv4-mapped and IPv4-compatible IPv6 addresses
 526                 if ( preg_match( '/^' . RE_IPV6_V4_PREFIX . '(' . RE_IP_ADD . ')$/i', $addr, $m ) )
 527                     return $m[1];
 528                 if ( preg_match( '/^' . RE_IPV6_V4_PREFIX . RE_IPV6_WORD . ':' . RE_IPV6_WORD . '$/i', $addr, $m ) )
 529                     return long2ip( ( hexdec( $m[1] ) << 16 ) + hexdec( $m[2] ) );
 530
 531                 return null;  // give up
 532     }
 533 }