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                 if ( is_array( $ip ) ) {
  41                   throw new MWException( "invalid value passed to " . __METHOD__ );
  42                 }
  43                 // IPv6 IPs with two "::" strings are ambiguous and thus invalid
  44                 return preg_match( '/^' . IP_ADDRESS_STRING . '$/', $ip) && ( substr_count($ip, '::') < 2 );
  45         }
  46
  47         public static function isIPv6( $ip ) {
  48                 if ( !$ip ) return false;
  49                 if( is_array( $ip ) ) {
  50                   throw new MWException( "invalid value passed to " . __METHOD__ );
  51                 }
  52                 // IPv6 IPs with two "::" strings are ambiguous and thus invalid
  53                 return preg_match( '/^' . RE_IPV6_ADD . '(\/' . RE_IPV6_PREFIX . '|)$/', $ip) && ( substr_count($ip, '::') < 2);
  54         }
  55
  56         public static function isIPv4( $ip ) {
  57                 if ( !$ip ) return false;
  58                 return preg_match( '/^' . RE_IP_ADD . '(\/' . RE_IP_PREFIX . '|)$/', $ip);
  59         }
  60
  61         /**
  62          * Given an IP address in dotted-quad notation, returns an IPv6 octet.
  63          * See http://www.answers.com/topic/ipv4-compatible-address
  64          * IPs with the first 92 bits as zeros are reserved from IPv6
  65          * @param $ip quad-dotted IP address.
  66          * @return string
  67          */
  68         public static function IPv4toIPv6( $ip ) {
  69                 if ( !$ip ) return null;
  70                 // Convert only if needed
  71                 if ( self::isIPv6( $ip ) ) return $ip;
  72                 // IPv4 CIDRs
  73                 if ( strpos( $ip, '/' ) !== false ) {
  74                         $parts = explode( '/', $ip, 2 );
  75                         if ( count( $parts ) != 2 ) {
  76                                 return false;
  77                         }
  78                         $network = self::toUnsigned( $parts[0] );
  79                         if ( $network !== false && is_numeric( $parts[1] ) && $parts[1] >= 0 && $parts[1] <= 32 ) {
  80                                 $bits = $parts[1] + 96;
  81                                 return self::toOctet( $network ) . "/$bits";
  82                         } else {
  83                                 return false;
  84                         }
  85                 }
  86                 return self::toOctet( self::toUnsigned( $ip ) );
  87         }
  88
  89         /**
  90          * Given an IPv6 address in octet notation, returns an unsigned integer.
  91          * @param $ip octet ipv6 IP address.
  92          * @return string
  93          */
  94         public static function toUnsigned6( $ip ) {
  95                 if ( !$ip ) return null;
  96         $ip = explode(':', self::sanitizeIP( $ip ) );
  97         $r_ip = '';
  98         foreach ($ip as $v) {
  99                 $r_ip .= str_pad( $v, 4, 0, STR_PAD_LEFT );
 100         }
 101         $r_ip = wfBaseConvert( $r_ip, 16, 10 );
 102         return $r_ip;
 103         }
 104
 105         /**
 106          * Given an IPv6 address in octet notation, returns the expanded octet.
 107          * IPv4 IPs will be trimmed, thats it...
 108          * @param $ip octet ipv6 IP address.
 109          * @return string
 110          */
 111         public static function sanitizeIP( $ip ) {
 112                 if ( !$ip ) return null;
 113                 // Trim and return IPv4 addresses
 114                 if ( self::isIPv4($ip) ) return trim($ip);
 115                 // Only IPv6 addresses can be expanded
 116                 if ( !self::isIPv6($ip) ) return $ip;
 117                 // Remove any whitespaces, convert to upper case
 118                 $ip = strtoupper( trim($ip) );
 119                 // Expand zero abbreviations
 120                 if ( strpos( $ip, '::' ) !== false ) {
 121                 $ip = str_replace('::', str_repeat(':0', 8 - substr_count($ip, ':')) . ':', $ip);
 122         }
 123         // For IPs that start with "::", correct the final IP so that it starts with '0' and not ':'
 124         if ( $ip[0] == ':' ) $ip = "0$ip";
 125         // Remove leading zereos from each bloc as needed
 126         $ip = preg_replace( '/(^|:)0+' . RE_IPV6_WORD . '/', '$1$2', $ip );
 127         return $ip;
 128         }
 129
 130         /**
 131          * Given an unsigned integer, returns an IPv6 address in octet notation
 132          * @param $ip integer IP address.
 133          * @return string
 134          */
 135         public static function toOctet( $ip_int ) {
 136                 // Convert to padded uppercase hex
 137                 $ip_hex = wfBaseConvert($ip_int, 10, 16, 32, false);
 138                 // Seperate into 8 octets
 139                 $ip_oct = substr( $ip_hex, 0, 4 );
 140                 for ($n=1; $n < 8; $n++) {
 141                         $ip_oct .= ':' . substr($ip_hex, 4*$n, 4);
 142                 }
 143                 // NO leading zeroes
 144                 $ip_oct = preg_replace( '/(^|:)0+' . RE_IPV6_WORD . '/', '$1$2', $ip_oct );
 145         return $ip_oct;
 146         }
 147
 148         /**
 149          * Convert a network specification in IPv6 CIDR notation to an integer network and a number of bits
 150          * @return array(string, int)
 151          */
 152         public static function parseCIDR6( $range ) {
 153                 # Expand any IPv6 IP
 154                 $parts = explode( '/', IP::sanitizeIP( $range ), 2 );
 155                 if ( count( $parts ) != 2 ) {
 156                         return array( false, false );
 157                 }
 158                 $network = self::toUnsigned6( $parts[0] );
 159                 if ( $network !== false && is_numeric( $parts[1] ) && $parts[1] >= 0 && $parts[1] <= 128 ) {
 160                         $bits = $parts[1];
 161                         if ( $bits == 0 ) {
 162                                 $network = 0;
 163                         } else {
 164                         # Native 32 bit functions WONT work here!!!
 165                         # Convert to a padded binary number
 166                                 $network = wfBaseConvert( $network, 10, 2, 128 );
 167                         # Truncate the last (128-$bits) bits and replace them with zeros
 168                                 $network = str_pad( substr( $network, 0, $bits ), 128, 0, STR_PAD_RIGHT );
 169                         # Convert back to an integer
 170                                 $network = wfBaseConvert( $network, 2, 10 );
 171                         }
 172                 } else {
 173                         $network = false;
 174                         $bits = false;
 175                 }
 176                 return array( $network, $bits );
 177         }
 178
 179         /**
 180          * Given a string range in a number of formats, return the start and end of
 181          * the range in hexadecimal. For IPv6.
 182          *
 183          * Formats are:
 184          *     2001:0db8:85a3::7344/96                                   CIDR
 185          *     2001:0db8:85a3::7344 - 2001:0db8:85a3::7344   Explicit range
 186          *     2001:0db8:85a3::7344/96                                   Single IP
 187          * @return array(string, int)
 188          */
 189         public static function parseRange6( $range ) {
 190                 # Expand any IPv6 IP
 191                 $range = IP::sanitizeIP( $range );
 192                 if ( strpos( $range, '/' ) !== false ) {
 193                         # CIDR
 194                         list( $network, $bits ) = self::parseCIDR6( $range );
 195                         if ( $network === false ) {
 196                                 $start = $end = false;
 197                         } else {
 198                                 $start = wfBaseConvert( $network, 10, 16, 32, false );
 199                                 # Turn network to binary (again)
 200                                 $end = wfBaseConvert( $network, 10, 2, 128 );
 201                                 # Truncate the last (128-$bits) bits and replace them with ones
 202                                 $end = str_pad( substr( $end, 0, $bits ), 128, 1, STR_PAD_RIGHT );
 203                                 # Convert to hex
 204                                 $end = wfBaseConvert( $end, 2, 16, 32, false );
 205                                 # see toHex() comment
 206                                 $start = "v6-$start"; $end = "v6-$end";
 207                         }
 208                 } elseif ( strpos( $range, '-' ) !== false ) {
 209                         # Explicit range
 210                         list( $start, $end ) = array_map( 'trim', explode( '-', $range, 2 ) );
 211                         $start = self::toUnsigned6( $start ); $end = self::toUnsigned6( $end );
 212                         if ( $start > $end ) {
 213                                 $start = $end = false;
 214                         } else {
 215                                 $start = wfBaseConvert( $start, 10, 16, 32, false );
 216                                 $end = wfBaseConvert( $end, 10, 16, 32, false );
 217                         }
 218                         # see toHex() comment
 219                         $start = "v6-$start"; $end = "v6-$end";
 220                 } else {
 221                         # Single IP
 222                         $start = $end = self::toHex( $range );
 223                 }
 224                 if ( $start === false || $end === false ) {
 225                         return array( false, false );
 226                 } else {
 227                         return array( $start, $end );
 228                 }
 229     }
 230
 231         /**
 232          * Validate an IP address.
 233          * @return boolean True if it is valid.
 234          */
 235         public static function isValid( $ip ) {
 236                 return ( preg_match( '/^' . RE_IP_ADD . '$/', $ip) || preg_match( '/^' . RE_IPV6_ADD . '$/', $ip) );
 237         }
 238
 239         /**
 240          * Validate an IP Block.
 241          * @return boolean True if it is valid.
 242          */
 243         public static function isValidBlock( $ipblock ) {
 244                 return ( count(self::toArray($ipblock)) == 1 + 5 );
 245         }
 246
 247         /**
 248          * Determine if an IP address really is an IP address, and if it is public,
 249          * i.e. not RFC 1918 or similar
 250          * Comes from ProxyTools.php
 251          */
 252         public static function isPublic( $ip ) {
 253                 $n = self::toUnsigned( $ip );
 254                 if ( !$n ) {
 255                         return false;
 256                 }
 257
 258                 // ip2long accepts incomplete addresses, as well as some addresses
 259                 // followed by garbage characters. Check that it's really valid.
 260                 if( $ip != long2ip( $n ) ) {
 261                         return false;
 262                 }
 263
 264                 static $privateRanges = false;
 265                 if ( !$privateRanges ) {
 266                         $privateRanges = array(
 267                                 array( '10.0.0.0',    '10.255.255.255' ),   # RFC 1918 (private)
 268                                 array( '172.16.0.0',  '172.31.255.255' ),   #     "
 269                                 array( '192.168.0.0', '192.168.255.255' ),  #     "
 270                                 array( '0.0.0.0',     '0.255.255.255' ),    # this network
 271                                 array( '127.0.0.0',   '127.255.255.255' ),  # loopback
 272                         );
 273                 }
 274
 275                 foreach ( $privateRanges as $r ) {
 276                         $start = self::toUnsigned( $r[0] );
 277                         $end = self::toUnsigned( $r[1] );
 278                         if ( $n >= $start && $n <= $end ) {
 279                                 return false;
 280                         }
 281                 }
 282                 return true;
 283         }
 284
 285         /**
 286          * Split out an IP block as an array of 4 bytes and a mask,
 287          * return false if it can't be determined
 288          *
 289          * @param $ip string A quad dotted/octet IP address
 290          * @return array
 291          */
 292         public static function toArray( $ipblock ) {
 293                 $matches = array();
 294                 if( preg_match( '/^' . RE_IP_ADD . '(?:\/(?:'.RE_IP_PREFIX.'))?' . '$/', $ipblock, $matches ) ) {
 295                         return $matches;
 296                 } else if ( preg_match( '/^' . RE_IPV6_ADD . '(?:\/(?:'.RE_IPV6_PREFIX.'))?' . '$/', $ipblock, $matches ) ) {
 297                         return $matches;
 298                 } else {
 299                         return false;
 300                 }
 301         }
 302
 303         /**
 304          * Return a zero-padded hexadecimal representation of an IP address.
 305          *
 306          * Hexadecimal addresses are used because they can easily be extended to
 307          * IPv6 support. To separate the ranges, the return value from this
 308          * function for an IPv6 address will be prefixed with "v6-", a non-
 309          * hexadecimal string which sorts after the IPv4 addresses.
 310          *
 311          * @param $ip Quad dotted/octet IP address.
 312          * @return hexidecimal
 313          */
 314         public static function toHex( $ip ) {
 315                 $n = self::toUnsigned( $ip );
 316                 if ( $n !== false ) {
 317                         $n = ( self::isIPv6($ip) ) ? "v6-" . wfBaseConvert( $n, 10, 16, 32, false ) : wfBaseConvert( $n, 10, 16, 8, false );
 318                 }
 319                 return $n;
 320         }
 321
 322         /**
 323          * Given an IP address in dotted-quad/octet notation, returns an unsigned integer.
 324          * Like ip2long() except that it actually works and has a consistent error return value.
 325          * Comes from ProxyTools.php
 326          * @param $ip Quad dotted IP address.
 327          * @return integer
 328          */
 329         public static function toUnsigned( $ip ) {
 330                 // Use IPv6 functions if needed
 331                 if ( self::isIPv6( $ip ) ) {
 332                         return self::toUnsigned6( $ip );
 333                 }
 334                 if ( $ip == '255.255.255.255' ) {
 335                         $n = -1;
 336                 } else {
 337                         $n = ip2long( $ip );
 338                         if ( $n == -1 || $n === false ) { # Return value on error depends on PHP version
 339                                 $n = false;
 340                         }
 341                 }
 342                 if ( $n < 0 ) {
 343                         $n += pow( 2, 32 );
 344                 }
 345                 return $n;
 346         }
 347
 348         /**
 349          * Convert a dotted-quad IP to a signed integer
 350          * Returns false on failure
 351          */
 352         public static function toSigned( $ip ) {
 353                 if ( $ip == '255.255.255.255' ) {
 354                         $n = -1;
 355                 } else {
 356                         $n = ip2long( $ip );
 357                         if ( $n == -1 ) {
 358                                 $n = false;
 359                         }
 360                 }
 361                 return $n;
 362         }
 363
 364         /**
 365          * Convert a network specification in CIDR notation to an integer network and a number of bits
 366          * @return array(string, int)
 367          */
 368         public static function parseCIDR( $range ) {
 369                 $parts = explode( '/', $range, 2 );
 370                 if ( count( $parts ) != 2 ) {
 371                         return array( false, false );
 372                 }
 373                 $network = self::toSigned( $parts[0] );
 374                 if ( $network !== false && is_numeric( $parts[1] ) && $parts[1] >= 0 && $parts[1] <= 32 ) {
 375                         $bits = $parts[1];
 376                         if ( $bits == 0 ) {
 377                                 $network = 0;
 378                         } else {
 379                                 $network &= ~((1 << (32 - $bits)) - 1);
 380                         }
 381                         # Convert to unsigned
 382                         if ( $network < 0 ) {
 383                                 $network += pow( 2, 32 );
 384                         }
 385                 } else {
 386                         $network = false;
 387                         $bits = false;
 388                 }
 389                 return array( $network, $bits );
 390         }
 391
 392         /**
 393          * Given a string range in a number of formats, return the start and end of
 394          * the range in hexadecimal.
 395          *
 396          * Formats are:
 397          *     1.2.3.4/24          CIDR
 398          *     1.2.3.4 - 1.2.3.5   Explicit range
 399          *     1.2.3.4             Single IP
 400          *
 401          *     2001:0db8:85a3::7344/96                                   CIDR
 402          *     2001:0db8:85a3::7344 - 2001:0db8:85a3::7344   Explicit range
 403          *     2001:0db8:85a3::7344                                      Single IP
 404          * @return array(string, int)
 405          */
 406         public static function parseRange( $range ) {
 407                 // Use IPv6 functions if needed
 408                 if ( self::isIPv6( $range ) ) {
 409                         return self::parseRange6( $range );
 410                 }
 411                 if ( strpos( $range, '/' ) !== false ) {
 412                         # CIDR
 413                         list( $network, $bits ) = self::parseCIDR( $range );
 414                         if ( $network === false ) {
 415                                 $start = $end = false;
 416                         } else {
 417                                 $start = sprintf( '%08X', $network );
 418                                 $end = sprintf( '%08X', $network + pow( 2, (32 - $bits) ) - 1 );
 419                         }
 420                 } elseif ( strpos( $range, '-' ) !== false ) {
 421                         # Explicit range
 422                         list( $start, $end ) = array_map( 'trim', explode( '-', $range, 2 ) );
 423                         $start = self::toUnsigned( $start ); $end = self::toUnsigned( $end );
 424                         if ( $start > $end ) {
 425                                 $start = $end = false;
 426                         } else {
 427                                 $start = sprintf( '%08X', $start );
 428                                 $end = sprintf( '%08X', $end );
 429                         }
 430                 } else {
 431                         # Single IP
 432                         $start = $end = self::toHex( $range );
 433                 }
 434                 if ( $start === false || $end === false ) {
 435                         return array( false, false );
 436                 } else {
 437                         return array( $start, $end );
 438                 }
 439     }
 440
 441     /**
 442      * Determine if a given IPv4/IPv6 address is in a given CIDR network
 443      * @param $addr The address to check against the given range.
 444      * @param $range The range to check the given address against.
 445      * @return bool Whether or not the given address is in the given range.
 446      */
 447     public static function isInRange( $addr, $range ) {
 448     // Convert to IPv6 if needed
 449         $unsignedIP = self::toHex( $addr );
 450         list( $start, $end ) = self::parseRange( $range );
 451         return (($unsignedIP >= $start) && ($unsignedIP <= $end));
 452     }
 453
 454     /**
 455      * Convert some unusual representations of IPv4 addresses to their
 456      * canonical dotted quad representation.
 457      *
 458      * This currently only checks a few IPV4-to-IPv6 related cases.  More
 459      * unusual representations may be added later.
 460      *
 461      * @param $addr something that might be an IP address
 462      * @return valid dotted quad IPv4 address or null
 463      */
 464     public static function canonicalize( $addr ) {
 465         if ( self::isValid( $addr ) )
 466             return $addr;
 467
 468         // IPv6 loopback address
 469         $m = array();
 470         if ( preg_match( '/^0*' . RE_IPV6_GAP . '1$/', $addr, $m ) )
 471             return '127.0.0.1';
 472
 473         // IPv4-mapped and IPv4-compatible IPv6 addresses
 474         if ( preg_match( '/^' . RE_IPV6_V4_PREFIX . '(' . RE_IP_ADD . ')$/i', $addr, $m ) )
 475             return $m[1];
 476         if ( preg_match( '/^' . RE_IPV6_V4_PREFIX . RE_IPV6_WORD . ':' . RE_IPV6_WORD . '$/i', $addr, $m ) )
 477             return long2ip( ( hexdec( $m[1] ) << 16 ) + hexdec( $m[2] ) );
 478
 479         return null;  // give up
 480     }
 481 }
 482
 483 ?>