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 an unsigned integer, returns 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 $ip string Hex IP
 175          * @return string
 176          */
 177         public static function hexToIP($ip)
 178         {
 179                 // Converts a hexadecimal IP to nnn.nnn.nnn.nnn format
 180                 $dec = wfBaseConvert($ip, 16, 10);
 181                 $parts[3] = $dec % 256;
 182                 $dec /= 256;
 183                 $parts[2] = $dec % 256;
 184                 $dec /= 256;
 185                 $parts[1] = $dec % 256;
 186                 $parts[0] = $dec / 256;
 187                 return implode('.', array_reverse($parts));
 188         }
 189
 190         /**
 191          * Convert a network specification in IPv6 CIDR notation to an integer network and a number of bits
 192          * @return array(string, int)
 193          */
 194         public static function parseCIDR6( $range ) {
 195                 # Expand any IPv6 IP
 196                 $parts = explode( '/', IP::sanitizeIP( $range ), 2 );
 197                 if ( count( $parts ) != 2 ) {
 198                         return array( false, false );
 199                 }
 200                 $network = self::toUnsigned6( $parts[0] );
 201                 if ( $network !== false && is_numeric( $parts[1] ) && $parts[1] >= 0 && $parts[1] <= 128 ) {
 202                         $bits = $parts[1];
 203                         if ( $bits == 0 ) {
 204                                 $network = 0;
 205                         } else {
 206                         # Native 32 bit functions WONT work here!!!
 207                         # Convert to a padded binary number
 208                                 $network = wfBaseConvert( $network, 10, 2, 128 );
 209                         # Truncate the last (128-$bits) bits and replace them with zeros
 210                                 $network = str_pad( substr( $network, 0, $bits ), 128, 0, STR_PAD_RIGHT );
 211                         # Convert back to an integer
 212                                 $network = wfBaseConvert( $network, 2, 10 );
 213                         }
 214                 } else {
 215                         $network = false;
 216                         $bits = false;
 217                 }
 218                 return array( $network, $bits );
 219         }
 220
 221         /**
 222          * Given a string range in a number of formats, return the start and end of
 223          * the range in hexadecimal. For IPv6.
 224          *
 225          * Formats are:
 226          *     2001:0db8:85a3::7344/96                                   CIDR
 227          *     2001:0db8:85a3::7344 - 2001:0db8:85a3::7344   Explicit range
 228          *     2001:0db8:85a3::7344/96                                   Single IP
 229          * @return array(string, int)
 230          */
 231         public static function parseRange6( $range ) {
 232                 # Expand any IPv6 IP
 233                 $range = IP::sanitizeIP( $range );
 234                 if ( strpos( $range, '/' ) !== false ) {
 235                         # CIDR
 236                         list( $network, $bits ) = self::parseCIDR6( $range );
 237                         if ( $network === false ) {
 238                                 $start = $end = false;
 239                         } else {
 240                                 $start = wfBaseConvert( $network, 10, 16, 32, false );
 241                                 # Turn network to binary (again)
 242                                 $end = wfBaseConvert( $network, 10, 2, 128 );
 243                                 # Truncate the last (128-$bits) bits and replace them with ones
 244                                 $end = str_pad( substr( $end, 0, $bits ), 128, 1, STR_PAD_RIGHT );
 245                                 # Convert to hex
 246                                 $end = wfBaseConvert( $end, 2, 16, 32, false );
 247                                 # see toHex() comment
 248                                 $start = "v6-$start"; $end = "v6-$end";
 249                         }
 250                 } elseif ( strpos( $range, '-' ) !== false ) {
 251                         # Explicit range
 252                         list( $start, $end ) = array_map( 'trim', explode( '-', $range, 2 ) );
 253                         $start = self::toUnsigned6( $start ); $end = self::toUnsigned6( $end );
 254                         if ( $start > $end ) {
 255                                 $start = $end = false;
 256                         } else {
 257                                 $start = wfBaseConvert( $start, 10, 16, 32, false );
 258                                 $end = wfBaseConvert( $end, 10, 16, 32, false );
 259                         }
 260                         # see toHex() comment
 261                         $start = "v6-$start"; $end = "v6-$end";
 262                 } else {
 263                         # Single IP
 264                         $start = $end = self::toHex( $range );
 265                 }
 266                 if ( $start === false || $end === false ) {
 267                         return array( false, false );
 268                 } else {
 269                         return array( $start, $end );
 270                 }
 271     }
 272
 273         /**
 274          * Validate an IP address.
 275          * @return boolean True if it is valid.
 276          */
 277         public static function isValid( $ip ) {
 278                 return ( preg_match( '/^' . RE_IP_ADD . '$/', $ip) || preg_match( '/^' . RE_IPV6_ADD . '$/', $ip) );
 279         }
 280
 281         /**
 282          * Validate an IP Block.
 283          * @return boolean True if it is valid.
 284          */
 285         public static function isValidBlock( $ipblock ) {
 286                 return ( count(self::toArray($ipblock)) == 1 + 5 );
 287         }
 288
 289         /**
 290          * Determine if an IP address really is an IP address, and if it is public,
 291          * i.e. not RFC 1918 or similar
 292          * Comes from ProxyTools.php
 293          */
 294         public static function isPublic( $ip ) {
 295                 $n = self::toUnsigned( $ip );
 296                 if ( !$n ) {
 297                         return false;
 298                 }
 299
 300                 // ip2long accepts incomplete addresses, as well as some addresses
 301                 // followed by garbage characters. Check that it's really valid.
 302                 if( $ip != long2ip( $n ) ) {
 303                         return false;
 304                 }
 305
 306                 static $privateRanges = false;
 307                 if ( !$privateRanges ) {
 308                         $privateRanges = array(
 309                                 array( '10.0.0.0',    '10.255.255.255' ),   # RFC 1918 (private)
 310                                 array( '172.16.0.0',  '172.31.255.255' ),   #     "
 311                                 array( '192.168.0.0', '192.168.255.255' ),  #     "
 312                                 array( '0.0.0.0',     '0.255.255.255' ),    # this network
 313                                 array( '127.0.0.0',   '127.255.255.255' ),  # loopback
 314                         );
 315                 }
 316
 317                 foreach ( $privateRanges as $r ) {
 318                         $start = self::toUnsigned( $r[0] );
 319                         $end = self::toUnsigned( $r[1] );
 320                         if ( $n >= $start && $n <= $end ) {
 321                                 return false;
 322                         }
 323                 }
 324                 return true;
 325         }
 326
 327         /**
 328          * Split out an IP block as an array of 4 bytes and a mask,
 329          * return false if it can't be determined
 330          *
 331          * @param $ip string A quad dotted/octet IP address
 332          * @return array
 333          */
 334         public static function toArray( $ipblock ) {
 335                 $matches = array();
 336                 if( preg_match( '/^' . RE_IP_ADD . '(?:\/(?:'.RE_IP_PREFIX.'))?' . '$/', $ipblock, $matches ) ) {
 337                         return $matches;
 338                 } else if ( preg_match( '/^' . RE_IPV6_ADD . '(?:\/(?:'.RE_IPV6_PREFIX.'))?' . '$/', $ipblock, $matches ) ) {
 339                         return $matches;
 340                 } else {
 341                         return false;
 342                 }
 343         }
 344
 345         /**
 346          * Return a zero-padded hexadecimal representation of an IP address.
 347          *
 348          * Hexadecimal addresses are used because they can easily be extended to
 349          * IPv6 support. To separate the ranges, the return value from this
 350          * function for an IPv6 address will be prefixed with "v6-", a non-
 351          * hexadecimal string which sorts after the IPv4 addresses.
 352          *
 353          * @param $ip Quad dotted/octet IP address.
 354          * @return hexidecimal
 355          */
 356         public static function toHex( $ip ) {
 357                 $n = self::toUnsigned( $ip );
 358                 if ( $n !== false ) {
 359                         $n = self::isIPv6($ip) ? "v6-" . wfBaseConvert( $n, 10, 16, 32, false ) : wfBaseConvert( $n, 10, 16, 8, false );
 360                 }
 361                 return $n;
 362         }
 363
 364         /**
 365          * Given an IP address in dotted-quad/octet notation, returns an unsigned integer.
 366          * Like ip2long() except that it actually works and has a consistent error return value.
 367          * Comes from ProxyTools.php
 368          * @param $ip Quad dotted IP address.
 369          * @return integer
 370          */
 371         public static function toUnsigned( $ip ) {
 372                 // Use IPv6 functions if needed
 373                 if ( self::isIPv6( $ip ) ) {
 374                         return self::toUnsigned6( $ip );
 375                 }
 376                 if ( $ip == '255.255.255.255' ) {
 377                         $n = -1;
 378                 } else {
 379                         $n = ip2long( $ip );
 380                         if ( $n == -1 || $n === false ) { # Return value on error depends on PHP version
 381                                 $n = false;
 382                         }
 383                 }
 384                 if ( $n < 0 ) {
 385                         $n += pow( 2, 32 );
 386                 }
 387                 return $n;
 388         }
 389
 390         /**
 391          * Convert a dotted-quad IP to a signed integer
 392          * Returns false on failure
 393          */
 394         public static function toSigned( $ip ) {
 395                 if ( $ip == '255.255.255.255' ) {
 396                         $n = -1;
 397                 } else {
 398                         $n = ip2long( $ip );
 399                         if ( $n == -1 ) {
 400                                 $n = false;
 401                         }
 402                 }
 403                 return $n;
 404         }
 405
 406         /**
 407          * Convert a network specification in CIDR notation to an integer network and a number of bits
 408          * @return array(string, int)
 409          */
 410         public static function parseCIDR( $range ) {
 411                 $parts = explode( '/', $range, 2 );
 412                 if ( count( $parts ) != 2 ) {
 413                         return array( false, false );
 414                 }
 415                 $network = self::toSigned( $parts[0] );
 416                 if ( $network !== false && is_numeric( $parts[1] ) && $parts[1] >= 0 && $parts[1] <= 32 ) {
 417                         $bits = $parts[1];
 418                         if ( $bits == 0 ) {
 419                                 $network = 0;
 420                         } else {
 421                                 $network &= ~((1 << (32 - $bits)) - 1);
 422                         }
 423                         # Convert to unsigned
 424                         if ( $network < 0 ) {
 425                                 $network += pow( 2, 32 );
 426                         }
 427                 } else {
 428                         $network = false;
 429                         $bits = false;
 430                 }
 431                 return array( $network, $bits );
 432         }
 433
 434         /**
 435          * Given a string range in a number of formats, return the start and end of
 436          * the range in hexadecimal.
 437          *
 438          * Formats are:
 439          *     1.2.3.4/24          CIDR
 440          *     1.2.3.4 - 1.2.3.5   Explicit range
 441          *     1.2.3.4             Single IP
 442          *
 443          *     2001:0db8:85a3::7344/96                                   CIDR
 444          *     2001:0db8:85a3::7344 - 2001:0db8:85a3::7344   Explicit range
 445          *     2001:0db8:85a3::7344                                      Single IP
 446          * @return array(string, int)
 447          */
 448         public static function parseRange( $range ) {
 449                 // Use IPv6 functions if needed
 450                 if ( self::isIPv6( $range ) ) {
 451                         return self::parseRange6( $range );
 452                 }
 453                 if ( strpos( $range, '/' ) !== false ) {
 454                         # CIDR
 455                         list( $network, $bits ) = self::parseCIDR( $range );
 456                         if ( $network === false ) {
 457                                 $start = $end = false;
 458                         } else {
 459                                 $start = sprintf( '%08X', $network );
 460                                 $end = sprintf( '%08X', $network + pow( 2, (32 - $bits) ) - 1 );
 461                         }
 462                 } elseif ( strpos( $range, '-' ) !== false ) {
 463                         # Explicit range
 464                         list( $start, $end ) = array_map( 'trim', explode( '-', $range, 2 ) );
 465                         if( self::isIPAddress( $start ) && self::isIPAddress( $end ) ) {
 466                                 $start = self::toUnsigned( $start ); $end = self::toUnsigned( $end );
 467                                 if ( $start > $end ) {
 468                                         $start = $end = false;
 469                                 } else {
 470                                         $start = sprintf( '%08X', $start );
 471                                         $end = sprintf( '%08X', $end );
 472                                 }
 473                         } else {
 474                                 $start = $end = false;
 475                         }
 476                 } else {
 477                         # Single IP
 478                         $start = $end = self::toHex( $range );
 479                 }
 480                 if ( $start === false || $end === false ) {
 481                         return array( false, false );
 482                 } else {
 483                         return array( $start, $end );
 484                 }
 485     }
 486
 487     /**
 488      * Determine if a given IPv4/IPv6 address is in a given CIDR network
 489      * @param $addr The address to check against the given range.
 490      * @param $range The range to check the given address against.
 491      * @return bool Whether or not the given address is in the given range.
 492      */
 493     public static function isInRange( $addr, $range ) {
 494     // Convert to IPv6 if needed
 495         $unsignedIP = self::toHex( $addr );
 496         list( $start, $end ) = self::parseRange( $range );
 497         return (($unsignedIP >= $start) && ($unsignedIP <= $end));
 498     }
 499
 500     /**
 501      * Convert some unusual representations of IPv4 addresses to their
 502      * canonical dotted quad representation.
 503      *
 504      * This currently only checks a few IPV4-to-IPv6 related cases.  More
 505      * unusual representations may be added later.
 506      *
 507      * @param $addr something that might be an IP address
 508      * @return valid dotted quad IPv4 address or null
 509      */
 510     public static function canonicalize( $addr ) {
 511                 if ( self::isValid( $addr ) )
 512                         return $addr;
 513
 514                 // Annoying IPv6 representations like ::ffff:1.2.3.4
 515                 if ( strpos($addr,':') !==false && strpos($addr,'.') !==false ) {
 516                         $addr = str_replace( '.', ':', $addr );
 517                         if( IP::isIPv6( $addr ) )
 518                                 return $addr;
 519                 }
 520
 521                 // IPv6 loopback address
 522                 $m = array();
 523                 if ( preg_match( '/^0*' . RE_IPV6_GAP . '1$/', $addr, $m ) )
 524                         return '127.0.0.1';
 525
 526                 // IPv4-mapped and IPv4-compatible IPv6 addresses
 527                 if ( preg_match( '/^' . RE_IPV6_V4_PREFIX . '(' . RE_IP_ADD . ')$/i', $addr, $m ) )
 528                     return $m[1];
 529                 if ( preg_match( '/^' . RE_IPV6_V4_PREFIX . RE_IPV6_WORD . ':' . RE_IPV6_WORD . '$/i', $addr, $m ) )
 530                     return long2ip( ( hexdec( $m[1] ) << 16 ) + hexdec( $m[2] ) );
 531
 532                 return null;  // give up
 533     }
 534 }