includes/IP.php

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