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