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', RE_IP_ADD . '(\/' . RE_IP_PREFIX . '|)|' . RE_IPV6_ADD . '(\/' . RE_IPV6_PREFIX . '|)');
  26
  27 /**
  28  * A collection of public static functions to play with IP address
  29  * and IP blocks.
  30  */
  31 class IP {
  32         /**
  33          * Given a string, determine if it as valid IP
  34          * Unlike isValid(), this looks for networks too
  35          * @param $ip IP address.
  36          * @return string
  37          */
  38         public static function isIPAddress( $ip ) {
  39                 if ( !$ip ) return false;
  40                 // IPv6 IPs with two "::" strings are ambiguous and thus invalid
  41                 return preg_match( '/^' . IP_ADDRESS_STRING . '$/', $ip) && ( substr_count($ip, '::') < 2 );
  42         }
  43
  44         public static function isIPv6( $ip ) {
  45                 if ( !$ip ) return false;
  46                 // IPv6 IPs with two "::" strings are ambiguous and thus invalid
  47                 return preg_match( '/^' . RE_IPV6_ADD . '(\/' . RE_IPV6_PREFIX . '|)$/', $ip) && ( substr_count($ip, '::') < 2);
  48         }
  49
  50         public static function isIPv4( $ip ) {
  51                 if ( !$ip ) return false;
  52                 return preg_match( '/^' . RE_IP_ADD . '(\/' . RE_IP_PREFIX . '|)$/', $ip);
  53         }
  54
  55         /**
  56          * Given an IP address in dotted-quad notation, returns an IPv6 octet.
  57          * See http://www.answers.com/topic/ipv4-compatible-address
  58          * IPs with the first 92 bits as zeros are reserved from IPv6
  59          * @param $ip quad-dotted IP address.
  60          * @return string
  61          */
  62         public static function IPv4toIPv6( $ip ) {
  63                 if ( !$ip ) return null;
  64                 // Convert only if needed
  65                 if ( self::isIPv6( $ip ) ) return $ip;
  66                 // IPv4 CIDRs
  67                 if ( strpos( $ip, '/' ) !== false ) {
  68                         $parts = explode( '/', $ip, 2 );
  69                         if ( count( $parts ) != 2 ) {
  70                                 return false;
  71                         }
  72                         $network = self::toUnsigned( $parts[0] );
  73                         if ( $network !== false && is_numeric( $parts[1] ) && $parts[1] >= 0 && $parts[1] <= 32 ) {
  74                                 $bits = $parts[1] + 96;
  75                                 return self::toOctet( $network ) . "/$bits";
  76                         } else {
  77                                 return false;
  78                         }
  79                 }
  80                 return self::toOctet( self::toUnsigned( $ip ) );
  81         }
  82
  83         /**
  84          * Given an IPv6 address in octet notation, returns an unsigned integer.
  85          * @param $ip octet ipv6 IP address.
  86          * @return string
  87          */
  88         public static function toUnsigned6( $ip ) {
  89                 if ( !$ip ) return null;
  90         $ip = explode(':', self::sanitizeIP( $ip ) );
  91         $r_ip = '';
  92         foreach ($ip as $v) {
  93                 $r_ip .= str_pad( $v, 4, 0, STR_PAD_LEFT );
  94         }
  95         $r_ip = wfBaseConvert( $r_ip, 16, 10 );
  96         return $r_ip;
  97         }
  98
  99         /**
 100          * Given an IPv6 address in octet notation, returns the expanded octet.
 101          * IPv4 IPs will be trimmed, thats it...
 102          * @param $ip octet ipv6 IP address.
 103          * @return string
 104          */
 105         public static function sanitizeIP( $ip ) {
 106                 if ( !$ip ) return null;
 107                 // Trim and return IPv4 addresses
 108                 if ( self::isIPv4($ip) ) return trim($ip);
 109                 // Only IPv6 addresses can be expanded
 110                 if ( !self::isIPv6($ip) ) return $ip;
 111                 // Remove any whitespaces, convert to upper case
 112                 $ip = strtoupper( trim($ip) );
 113                 // Expand zero abbreviations
 114                 if ( strpos( $ip, '::' ) !== false ) {
 115                 $ip = str_replace('::', str_repeat(':0', 8 - substr_count($ip, ':')) . ':', $ip);
 116         }
 117         // For IPs that start with "::", correct the final IP so that it starts with '0' and not ':'
 118         if ( $ip[0] == ':' ) $ip = "0$ip";
 119         // Remove leading zereos from each bloc as needed
 120         $ip = preg_replace( '/(^|:)0+' . RE_IPV6_WORD . '/', '$1$2', $ip );
 121         return $ip;
 122         }
 123
 124         /**
 125          * Given an unsigned integer, returns an IPv6 address in octet notation
 126          * @param $ip integer IP address.
 127          * @return string
 128          */
 129         public static function toOctet( $ip_int ) {
 130                 // Convert to padded uppercase hex
 131                 $ip_hex = wfBaseConvert($ip_int, 10, 16, 32, false);
 132                 // Seperate into 8 octets
 133                 $ip_oct = substr( $ip_hex, 0, 4 );
 134                 for ($n=1; $n < 8; $n++) {
 135                         $ip_oct .= ':' . substr($ip_hex, 4*$n, 4);
 136                 }
 137                 // NO leading zeroes
 138                 $ip_oct = preg_replace( '/(^|:)0+' . RE_IPV6_WORD . '/', '$1$2', $ip_oct );
 139         return $ip_oct;
 140         }
 141
 142         /**
 143          * Convert a network specification in IPv6 CIDR notation to an integer network and a number of bits
 144          * @return array(string, int)
 145          */
 146         public static function parseCIDR6( $range ) {
 147                 # Expand any IPv6 IP
 148                 $parts = explode( '/', IP::sanitizeIP( $range ), 2 );
 149                 if ( count( $parts ) != 2 ) {
 150                         return array( false, false );
 151                 }
 152                 $network = self::toUnsigned6( $parts[0] );
 153                 if ( $network !== false && is_numeric( $parts[1] ) && $parts[1] >= 0 && $parts[1] <= 128 ) {
 154                         $bits = $parts[1];
 155                         if ( $bits == 0 ) {
 156                                 $network = 0;
 157                         } else {
 158                         # Native 32 bit functions WONT work here!!!
 159                         # Convert to a padded binary number
 160                                 $network = wfBaseConvert( $network, 10, 2, 128 );
 161                         # Truncate the last (128-$bits) bits and replace them with zeros
 162                                 $network = str_pad( substr( $network, 0, $bits ), 128, 0, STR_PAD_RIGHT );
 163                         # Convert back to an integer
 164                                 $network = wfBaseConvert( $network, 2, 10 );
 165                         }
 166                 } else {
 167                         $network = false;
 168                         $bits = false;
 169                 }
 170                 return array( $network, $bits );
 171         }
 172
 173         /**
 174          * Given a string range in a number of formats, return the start and end of
 175          * the range in hexadecimal. For IPv6.
 176          *
 177          * Formats are:
 178          *     2001:0db8:85a3::7344/96                                   CIDR
 179          *     2001:0db8:85a3::7344 - 2001:0db8:85a3::7344   Explicit range
 180          *     2001:0db8:85a3::7344/96                                   Single IP
 181          * @return array(string, int)
 182          */
 183         public static function parseRange6( $range ) {
 184                 # Expand any IPv6 IP
 185                 $range = IP::sanitizeIP( $range );
 186                 if ( strpos( $range, '/' ) !== false ) {
 187                         # CIDR
 188                         list( $network, $bits ) = self::parseCIDR6( $range );
 189                         if ( $network === false ) {
 190                                 $start = $end = false;
 191                         } else {
 192                                 $start = wfBaseConvert( $network, 10, 16, 32, false );
 193                                 # Turn network to binary (again)
 194                                 $end = wfBaseConvert( $network, 10, 2, 128 );
 195                                 # Truncate the last (128-$bits) bits and replace them with ones
 196                                 $end = str_pad( substr( $end, 0, $bits ), 128, 1, STR_PAD_RIGHT );
 197                                 # Convert to hex
 198                                 $end = wfBaseConvert( $end, 2, 16, 32, false );
 199                                 # see toHex() comment
 200                                 $start = "ipv6-$start"; $end = "ipv6-$end";
 201                         }
 202                 } elseif ( strpos( $range, '-' ) !== false ) {
 203                         # Explicit range
 204                         list( $start, $end ) = array_map( 'trim', explode( '-', $range, 2 ) );
 205                         $start = self::toUnsigned6( $start ); $end = self::toUnsigned6( $end );
 206                         if ( $start > $end ) {
 207                                 $start = $end = false;
 208                         } else {
 209                                 $start = wfBaseConvert( $start, 10, 16, 32, false );
 210                                 $end = wfBaseConvert( $end, 10, 16, 32, false );
 211                         }
 212                         # see toHex() comment
 213                         $start = "ipv6-$start"; $end = "ipv6-$end";
 214                 } else {
 215                         # Single IP
 216                         $start = $end = self::toHex( $range );
 217                 }
 218                 if ( $start === false || $end === false ) {
 219                         return array( false, false );
 220                 } else {
 221                         return array( $start, $end );
 222                 }
 223     }
 224
 225         /**
 226          * Validate an IP address.
 227          * @return boolean True if it is valid.
 228          */
 229         public static function isValid( $ip ) {
 230                 return ( preg_match( '/^' . RE_IP_ADD . '$/', $ip) || preg_match( '/^' . RE_IPV6_ADD . '$/', $ip) );
 231         }
 232
 233         /**
 234          * Validate an IP Block.
 235          * @return boolean True if it is valid.
 236          */
 237         public static function isValidBlock( $ipblock ) {
 238                 return ( count(self::toArray($ipblock)) == 1 + 5 );
 239         }
 240
 241         /**
 242          * Determine if an IP address really is an IP address, and if it is public,
 243          * i.e. not RFC 1918 or similar
 244          * Comes from ProxyTools.php
 245          */
 246         public static function isPublic( $ip ) {
 247                 $n = self::toUnsigned( $ip );
 248                 if ( !$n ) {
 249                         return false;
 250                 }
 251
 252                 // ip2long accepts incomplete addresses, as well as some addresses
 253                 // followed by garbage characters. Check that it's really valid.
 254                 if( $ip != long2ip( $n ) ) {
 255                         return false;
 256                 }
 257
 258                 static $privateRanges = false;
 259                 if ( !$privateRanges ) {
 260                         $privateRanges = array(
 261                                 array( '10.0.0.0',    '10.255.255.255' ),   # RFC 1918 (private)
 262                                 array( '172.16.0.0',  '172.31.255.255' ),   #     "
 263                                 array( '192.168.0.0', '192.168.255.255' ),  #     "
 264                                 array( '0.0.0.0',     '0.255.255.255' ),    # this network
 265                                 array( '127.0.0.0',   '127.255.255.255' ),  # loopback
 266                         );
 267                 }
 268
 269                 foreach ( $privateRanges as $r ) {
 270                         $start = self::toUnsigned( $r[0] );
 271                         $end = self::toUnsigned( $r[1] );
 272                         if ( $n >= $start && $n <= $end ) {
 273                                 return false;
 274                         }
 275                 }
 276                 return true;
 277         }
 278
 279         /**
 280          * Split out an IP block as an array of 4 bytes and a mask,
 281          * return false if it can't be determined
 282          *
 283          * @parameter $ip string A quad dotted/octet IP address
 284          * @return array
 285          */
 286         public static function toArray( $ipblock ) {
 287                 $matches = array();
 288                 if( preg_match( '/^' . RE_IP_ADD . '(?:\/(?:'.RE_IP_PREFIX.'))?' . '$/', $ipblock, $matches ) ) {
 289                         return $matches;
 290                 } else if ( preg_match( '/^' . RE_IPV6_ADD . '(?:\/(?:'.RE_IPV6_PREFIX.'))?' . '$/', $ipblock, $matches ) ) {
 291                         return $matches;
 292                 } else {
 293                         return false;
 294                 }
 295         }
 296
 297         /**
 298          * Return a zero-padded hexadecimal representation of an IP address.
 299          *
 300          * Hexadecimal addresses are used because they can easily be extended to
 301          * IPv6 support. To separate the ranges, the return value from this
 302          * function for an IPv6 address will be prefixed with "v6-", a non-
 303          * hexadecimal string which sorts after the IPv4 addresses.
 304          *
 305          * @param $ip Quad dotted/octet IP address.
 306          * @return hexidecimal
 307          */
 308         public static function toHex( $ip ) {
 309                 $n = self::toUnsigned( $ip );
 310                 if ( $n !== false ) {
 311                         $n = ( self::isIPv6($ip) ) ? "v6-" . wfBaseConvert( $n, 10, 16, 32, false ) : wfBaseConvert( $n, 10, 16, 8, false );
 312                 }
 313                 return $n;
 314         }
 315
 316         /**
 317          * Given an IP address in dotted-quad/octet notation, returns an unsigned integer.
 318          * Like ip2long() except that it actually works and has a consistent error return value.
 319          * Comes from ProxyTools.php
 320          * @param $ip Quad dotted IP address.
 321          * @return integer
 322          */
 323         public static function toUnsigned( $ip ) {
 324                 // Use IPv6 functions if needed
 325                 if ( self::isIPv6( $ip ) ) {
 326                         return self::toUnsigned6( $ip );
 327                 }
 328                 if ( $ip == '255.255.255.255' ) {
 329                         $n = -1;
 330                 } else {
 331                         $n = ip2long( $ip );
 332                         if ( $n == -1 || $n === false ) { # Return value on error depends on PHP version
 333                                 $n = false;
 334                         }
 335                 }
 336                 if ( $n < 0 ) {
 337                         $n += pow( 2, 32 );
 338                 }
 339                 return $n;
 340         }
 341
 342         /**
 343          * Convert a dotted-quad IP to a signed integer
 344          * Returns false on failure
 345          */
 346         public static function toSigned( $ip ) {
 347                 if ( $ip == '255.255.255.255' ) {
 348                         $n = -1;
 349                 } else {
 350                         $n = ip2long( $ip );
 351                         if ( $n == -1 ) {
 352                                 $n = false;
 353                         }
 354                 }
 355                 return $n;
 356         }
 357
 358         /**
 359          * Convert a network specification in CIDR notation to an integer network and a number of bits
 360          * @return array(string, int)
 361          */
 362         public static function parseCIDR( $range ) {
 363                 $parts = explode( '/', $range, 2 );
 364                 if ( count( $parts ) != 2 ) {
 365                         return array( false, false );
 366                 }
 367                 $network = self::toSigned( $parts[0] );
 368                 if ( $network !== false && is_numeric( $parts[1] ) && $parts[1] >= 0 && $parts[1] <= 32 ) {
 369                         $bits = $parts[1];
 370                         if ( $bits == 0 ) {
 371                                 $network = 0;
 372                         } else {
 373                                 $network &= ~((1 << (32 - $bits)) - 1);
 374                         }
 375                         # Convert to unsigned
 376                         if ( $network < 0 ) {
 377                                 $network += pow( 2, 32 );
 378                         }
 379                 } else {
 380                         $network = false;
 381                         $bits = false;
 382                 }
 383                 return array( $network, $bits );
 384         }
 385
 386         /**
 387          * Given a string range in a number of formats, return the start and end of
 388          * the range in hexadecimal.
 389          *
 390          * Formats are:
 391          *     1.2.3.4/24          CIDR
 392          *     1.2.3.4 - 1.2.3.5   Explicit range
 393          *     1.2.3.4             Single IP
 394          *
 395          *     2001:0db8:85a3::7344/96                                   CIDR
 396          *     2001:0db8:85a3::7344 - 2001:0db8:85a3::7344   Explicit range
 397          *     2001:0db8:85a3::7344/96                                   Single IP
 398          * @return array(string, int)
 399          */
 400         public static function parseRange( $range ) {
 401                 // Use IPv6 functions if needed
 402                 if ( self::isIPv6( $range ) ) {
 403                         return self::parseRange6( $range );
 404                 }
 405                 if ( strpos( $range, '/' ) !== false ) {
 406                         # CIDR
 407                         list( $network, $bits ) = self::parseCIDR( $range );
 408                         if ( $network === false ) {
 409                                 $start = $end = false;
 410                         } else {
 411                                 $start = sprintf( '%08X', $network );
 412                                 $end = sprintf( '%08X', $network + pow( 2, (32 - $bits) ) - 1 );
 413                         }
 414                 } elseif ( strpos( $range, '-' ) !== false ) {
 415                         # Explicit range
 416                         list( $start, $end ) = array_map( 'trim', explode( '-', $range, 2 ) );
 417                         $start = self::toUnsigned( $start ); $end = self::toUnsigned( $end );
 418                         if ( $start > $end ) {
 419                                 $start = $end = false;
 420                         } else {
 421                                 $start = sprintf( '%08X', $start );
 422                                 $end = sprintf( '%08X', $end );
 423                         }
 424                 } else {
 425                         # Single IP
 426                         $start = $end = self::toHex( $range );
 427                 }
 428                 if ( $start === false || $end === false ) {
 429                         return array( false, false );
 430                 } else {
 431                         return array( $start, $end );
 432                 }
 433     }
 434
 435     /**
 436      * Determine if a given IPv4/IPv6 address is in a given CIDR network
 437      * @param $addr The address to check against the given range.
 438      * @param $range The range to check the given address against.
 439      * @return bool Whether or not the given address is in the given range.
 440      */
 441     public static function isInRange( $addr, $range ) {
 442     // Convert to IPv6 if needed
 443         $unsignedIP = self::toHex( $addr );
 444         list( $start, $end ) = self::parseRange( $range );
 445         return (($unsignedIP >= $start) && ($unsignedIP <= $end));
 446     }
 447
 448     /**
 449      * Convert some unusual representations of IPv4 addresses to their
 450      * canonical dotted quad representation.
 451      *
 452      * This currently only checks a few IPV4-to-IPv6 related cases.  More
 453      * unusual representations may be added later.
 454      *
 455      * @param $addr something that might be an IP address
 456      * @return valid dotted quad IPv4 address or null
 457      */
 458     public static function canonicalize( $addr ) {
 459         if ( self::isValid( $addr ) )
 460             return $addr;
 461
 462         // IPv6 loopback address
 463         $m = array();
 464         if ( preg_match( '/^0*' . RE_IPV6_GAP . '1$/', $addr, $m ) )
 465             return '127.0.0.1';
 466
 467         // IPv4-mapped and IPv4-compatible IPv6 addresses
 468         if ( preg_match( '/^' . RE_IPV6_V4_PREFIX . '(' . RE_IP_ADD . ')$/i', $addr, $m ) )
 469             return $m[1];
 470         if ( preg_match( '/^' . RE_IPV6_V4_PREFIX . RE_IPV6_WORD . ':' . RE_IPV6_WORD . '$/i', $addr, $m ) )
 471             return long2ip( ( hexdec( $m[1] ) << 16 ) + hexdec( $m[2] ) );
 472
 473         return null;  // give up
 474     }
 475 }
 476
 477 ?>