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 thus 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 thus 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 .= str_pad( $v, 4, 0, STR_PAD_LEFT );
  93         }
  94         $r_ip = wfBaseConvert( $r_ip, 16, 10 );
  95         return $r_ip;
  96         }
  97
  98         /**
  99          * Given an IPv6 address in octet notation, returns the expanded octet.
 100          * IPv4 IPs will be trimmed, thats it...
 101          * @param $ip octet ipv6 IP address.
 102          * @return string
 103          */
 104         public static function sanitizeIP( $ip ) {
 105                 if ( !$ip ) return null;
 106                 // Trim and return IPv4 addresses
 107                 if ( self::isIPv4($ip) ) return trim($ip);
 108                 // Only IPv6 addresses can be expanded
 109                 if ( !self::isIPv6($ip) ) return $ip;
 110                 // Remove any whitespaces, convert to upper case
 111                 $ip = strtoupper( trim($ip) );
 112                 // Expand zero abbreviations
 113                 if ( strpos( $ip, '::' ) !== false ) {
 114                 $ip = str_replace('::', str_repeat(':0', 8 - substr_count($ip, ':')) . ':', $ip);
 115         }
 116         // For IPs that start with "::", correct the final IP so that it starts with '0' and not ':'
 117         if ( $ip[0] == ':' ) $ip = "0$ip";
 118         // Remove leading zereos from each bloc as needed
 119         $ip = preg_replace( '/(^|:)0+' . RE_IPV6_WORD . '/', '$1$2', $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 to padded uppercase hex
 130                 $ip_hex = wfBaseConvert($ip_int, 10, 16, 32, false);
 131                 // Seperate into 8 octets
 132                 $ip_oct = substr( $ip_hex, 0, 4 );
 133                 for ($n=1; $n < 8; $n++) {
 134                         $ip_oct .= ':' . substr($ip_hex, 4*$n, 4);
 135                 }
 136                 // NO leading zeroes
 137                 $ip_oct = preg_replace( '/(^|:)0+' . RE_IPV6_WORD . '/', '$1$2', $ip_oct );
 138         return $ip_oct;
 139         }
 140
 141         /**
 142          * Convert a network specification in IPv6 CIDR notation to an integer network and a number of bits
 143          * @return array(string, int)
 144          */
 145         public static function parseCIDR6( $range ) {
 146                 # Expand any IPv6 IP
 147                 $parts = explode( '/', IP::sanitizeIP( $range ), 2 );
 148                 if ( count( $parts ) != 2 ) {
 149                         return array( false, false );
 150                 }
 151                 $network = self::toUnsigned6( $parts[0] );
 152                 if ( $network !== false && is_numeric( $parts[1] ) && $parts[1] >= 0 && $parts[1] <= 128 ) {
 153                         $bits = $parts[1];
 154                         if ( $bits == 0 ) {
 155                                 $network = 0;
 156                         } else {
 157                         # Native 32 bit functions WONT work here!!!
 158                         # Convert to a padded binary number
 159                                 $network = wfBaseConvert( $network, 10, 2, 128 );
 160                         # Truncate the last (128-$bits) bits and replace them with zeros
 161                                 $network = str_pad( substr( $network, 0, $bits ), 128, 0, STR_PAD_RIGHT );
 162                         # Convert back to an integer
 163                                 $network = wfBaseConvert( $network, 2, 10 );
 164                         }
 165                 } else {
 166                         $network = false;
 167                         $bits = false;
 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                                 # see toHex() comment
 199                                 $start = "ipv6-$start"; $end = "ipv6-$end";
 200                         }
 201                 } elseif ( strpos( $range, '-' ) !== false ) {
 202                         # Explicit range
 203                         list( $start, $end ) = array_map( 'trim', explode( '-', $range, 2 ) );
 204                         $start = self::toUnsigned6( $start ); $end = self::toUnsigned6( $end );
 205                         if ( $start > $end ) {
 206                                 $start = $end = false;
 207                         } else {
 208                                 $start = wfBaseConvert( $start, 10, 16, 32, false );
 209                                 $end = wfBaseConvert( $end, 10, 16, 32, false );
 210                         }
 211                         # see toHex() comment
 212                         $start = "ipv6-$start"; $end = "ipv6-$end";
 213                 } else {
 214                         # Single IP
 215                         $start = $end = self::toHex( $range );
 216                 }
 217                 if ( $start === false || $end === false ) {
 218                         return array( false, false );
 219                 } else {
 220                         return array( $start, $end );
 221                 }
 222     }
 223
 224         /**
 225          * Validate an IP address.
 226          * @return boolean True if it is valid.
 227          */
 228         public static function isValid( $ip ) {
 229                 return ( preg_match( '/^' . RE_IP_ADD . '$/', $ip) || preg_match( '/^' . RE_IPV6_ADD . '$/', $ip) );
 230         }
 231
 232         /**
 233          * Validate an IP Block.
 234          * @return boolean True if it is valid.
 235          */
 236         public static function isValidBlock( $ipblock ) {
 237                 return ( count(self::toArray($ipblock)) == 1 + 5 );
 238         }
 239
 240         /**
 241          * Determine if an IP address really is an IP address, and if it is public,
 242          * i.e. not RFC 1918 or similar
 243          * Comes from ProxyTools.php
 244          */
 245         public static function isPublic( $ip ) {
 246                 $n = self::toUnsigned( $ip );
 247                 if ( !$n ) {
 248                         return false;
 249                 }
 250
 251                 // ip2long accepts incomplete addresses, as well as some addresses
 252                 // followed by garbage characters. Check that it's really valid.
 253                 if( $ip != long2ip( $n ) ) {
 254                         return false;
 255                 }
 256
 257                 static $privateRanges = false;
 258                 if ( !$privateRanges ) {
 259                         $privateRanges = array(
 260                                 array( '10.0.0.0',    '10.255.255.255' ),   # RFC 1918 (private)
 261                                 array( '172.16.0.0',  '172.31.255.255' ),   #     "
 262                                 array( '192.168.0.0', '192.168.255.255' ),  #     "
 263                                 array( '0.0.0.0',     '0.255.255.255' ),    # this network
 264                                 array( '127.0.0.0',   '127.255.255.255' ),  # loopback
 265                         );
 266                 }
 267
 268                 foreach ( $privateRanges as $r ) {
 269                         $start = self::toUnsigned( $r[0] );
 270                         $end = self::toUnsigned( $r[1] );
 271                         if ( $n >= $start && $n <= $end ) {
 272                                 return false;
 273                         }
 274                 }
 275                 return true;
 276         }
 277
 278         /**
 279          * Split out an IP block as an array of 4 bytes and a mask,
 280          * return false if it can't be determined
 281          *
 282          * @parameter $ip string A quad dotted/octet IP address
 283          * @return array
 284          */
 285         public static function toArray( $ipblock ) {
 286                 $matches = array();
 287                 if( preg_match( '/^' . RE_IP_ADD . '(?:\/(?:'.RE_IP_PREFIX.'))?' . '$/', $ipblock, $matches ) ) {
 288                         return $matches;
 289                 } else if ( preg_match( '/^' . RE_IPV6_ADD . '(?:\/(?:'.RE_IPV6_PREFIX.'))?' . '$/', $ipblock, $matches ) ) {
 290                         return $matches;
 291                 } else {
 292                         return false;
 293                 }
 294         }
 295
 296         /**
 297          * Return a zero-padded hexadecimal representation of an IP address.
 298          *
 299          * Hexadecimal addresses are used because they can easily be extended to
 300          * IPv6 support. To separate the ranges, the return value from this
 301          * function for an IPv6 address will be prefixed with "v6-", a non-
 302          * hexadecimal string which sorts after the IPv4 addresses.
 303          *
 304          * @param $ip Quad dotted/octet IP address.
 305          * @return hexidecimal
 306          */
 307         public static function toHex( $ip ) {
 308                 $n = self::toUnsigned( $ip );
 309                 if ( $n !== false ) {
 310                         $n = ( self::isIPv6($ip) ) ? "v6-" . wfBaseConvert( $n, 10, 16, 32, false ) : wfBaseConvert( $n, 10, 16, 8, false );
 311                 }
 312                 return $n;
 313         }
 314
 315         /**
 316          * Given an IP address in dotted-quad/octet notation, returns an unsigned integer.
 317          * Like ip2long() except that it actually works and has a consistent error return value.
 318          * Comes from ProxyTools.php
 319          * @param $ip Quad dotted IP address.
 320          * @return integer
 321          */
 322         public static function toUnsigned( $ip ) {
 323                 // Use IPv6 functions if needed
 324                 if ( self::isIPv6( $ip ) ) {
 325                         return self::toUnsigned6( $ip );
 326                 }
 327                 if ( $ip == '255.255.255.255' ) {
 328                         $n = -1;
 329                 } else {
 330                         $n = ip2long( $ip );
 331                         if ( $n == -1 || $n === false ) { # Return value on error depends on PHP version
 332                                 $n = false;
 333                         }
 334                 }
 335                 if ( $n < 0 ) {
 336                         $n += pow( 2, 32 );
 337                 }
 338                 return $n;
 339         }
 340
 341         /**
 342          * Convert a dotted-quad IP to a signed integer
 343          * Returns false on failure
 344          */
 345         public static function toSigned( $ip ) {
 346                 if ( $ip == '255.255.255.255' ) {
 347                         $n = -1;
 348                 } else {
 349                         $n = ip2long( $ip );
 350                         if ( $n == -1 ) {
 351                                 $n = false;
 352                         }
 353                 }
 354                 return $n;
 355         }
 356
 357         /**
 358          * Convert a network specification in CIDR notation to an integer network and a number of bits
 359          * @return array(string, int)
 360          */
 361         public static function parseCIDR( $range ) {
 362                 $parts = explode( '/', $range, 2 );
 363                 if ( count( $parts ) != 2 ) {
 364                         return array( false, false );
 365                 }
 366                 $network = self::toSigned( $parts[0] );
 367                 if ( $network !== false && is_numeric( $parts[1] ) && $parts[1] >= 0 && $parts[1] <= 32 ) {
 368                         $bits = $parts[1];
 369                         if ( $bits == 0 ) {
 370                                 $network = 0;
 371                         } else {
 372                                 $network &= ~((1 << (32 - $bits)) - 1);
 373                         }
 374                         # Convert to unsigned
 375                         if ( $network < 0 ) {
 376                                 $network += pow( 2, 32 );
 377                         }
 378                 } else {
 379                         $network = false;
 380                         $bits = false;
 381                 }
 382                 return array( $network, $bits );
 383         }
 384
 385         /**
 386          * Given a string range in a number of formats, return the start and end of
 387          * the range in hexadecimal.
 388          *
 389          * Formats are:
 390          *     1.2.3.4/24          CIDR
 391          *     1.2.3.4 - 1.2.3.5   Explicit range
 392          *     1.2.3.4             Single IP
 393          *
 394          *     2001:0db8:85a3::7344/96                                   CIDR
 395          *     2001:0db8:85a3::7344 - 2001:0db8:85a3::7344   Explicit range
 396          *     2001:0db8:85a3::7344                                      Single IP
 397          * @return array(string, int)
 398          */
 399         public static function parseRange( $range ) {
 400                 // Use IPv6 functions if needed
 401                 if ( self::isIPv6( $range ) ) {
 402                         return self::parseRange6( $range );
 403                 }
 404                 if ( strpos( $range, '/' ) !== false ) {
 405                         # CIDR
 406                         list( $network, $bits ) = self::parseCIDR( $range );
 407                         if ( $network === false ) {
 408                                 $start = $end = false;
 409                         } else {
 410                                 $start = sprintf( '%08X', $network );
 411                                 $end = sprintf( '%08X', $network + pow( 2, (32 - $bits) ) - 1 );
 412                         }
 413                 } elseif ( strpos( $range, '-' ) !== false ) {
 414                         # Explicit range
 415                         list( $start, $end ) = array_map( 'trim', explode( '-', $range, 2 ) );
 416                         $start = self::toUnsigned( $start ); $end = self::toUnsigned( $end );
 417                         if ( $start > $end ) {
 418                                 $start = $end = false;
 419                         } else {
 420                                 $start = sprintf( '%08X', $start );
 421                                 $end = sprintf( '%08X', $end );
 422                         }
 423                 } else {
 424                         # Single IP
 425                         $start = $end = self::toHex( $range );
 426                 }
 427                 if ( $start === false || $end === false ) {
 428                         return array( false, false );
 429                 } else {
 430                         return array( $start, $end );
 431                 }
 432     }
 433
 434     /**
 435      * Determine if a given IPv4/IPv6 address is in a given CIDR network
 436      * @param $addr The address to check against the given range.
 437      * @param $range The range to check the given address against.
 438      * @return bool Whether or not the given address is in the given range.
 439      */
 440     public static function isInRange( $addr, $range ) {
 441     // Convert to IPv6 if needed
 442         $unsignedIP = self::toHex( $addr );
 443         list( $start, $end ) = self::parseRange( $range );
 444         return (($unsignedIP >= $start) && ($unsignedIP <= $end));
 445     }
 446
 447     /**
 448      * Convert some unusual representations of IPv4 addresses to their
 449      * canonical dotted quad representation.
 450      *
 451      * This currently only checks a few IPV4-to-IPv6 related cases.  More
 452      * unusual representations may be added later.
 453      *
 454      * @param $addr something that might be an IP address
 455      * @return valid dotted quad IPv4 address or null
 456      */
 457     public static function canonicalize( $addr ) {
 458         if ( self::isValid( $addr ) )
 459             return $addr;
 460
 461         // IPv6 loopback address
 462         $m = array();
 463         if ( preg_match( '/^0*' . RE_IPV6_GAP . '1$/', $addr, $m ) )
 464             return '127.0.0.1';
 465
 466         // IPv4-mapped and IPv4-compatible IPv6 addresses
 467         if ( preg_match( '/^' . RE_IPV6_V4_PREFIX . '(' . RE_IP_ADD . ')$/i', $addr, $m ) )
 468             return $m[1];
 469         if ( preg_match( '/^' . RE_IPV6_V4_PREFIX . RE_IPV6_WORD . ':' . RE_IPV6_WORD . '$/i', $addr, $m ) )
 470             return long2ip( ( hexdec( $m[1] ) << 16 ) + hexdec( $m[2] ) );
 471
 472         return null;  // give up
 473     }
 474 }
 475
 476 ?>