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