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',
  26         '(?:' .
  27                 RE_IP_ADD . '(\/' . RE_IP_PREFIX . '|)' .
  28         '|' .
  29                 RE_IPV6_ADD . '(\/' . RE_IPV6_PREFIX . '|)' .
  30         ')' );
  31
  32 /**
  33  * A collection of public static functions to play with IP address
  34  * and IP blocks.
  35  */
  36 class IP {
  37         /**
  38          * Given a string, determine if it as valid IP
  39          * Unlike isValid(), this looks for networks too
  40          * @param $ip IP address.
  41          * @return string
  42          */
  43         public static function isIPAddress( $ip ) {
  44                 if ( !$ip ) return false;
  45                 if ( is_array( $ip ) ) {
  46                   throw new MWException( "invalid value passed to " . __METHOD__ );
  47                 }
  48                 // IPv6 IPs with two "::" strings are ambiguous and thus invalid
  49                 return preg_match( '/^' . IP_ADDRESS_STRING . '$/', $ip) && ( substr_count($ip, '::') < 2 );
  50         }
  51
  52         public static function isIPv6( $ip ) {
  53                 if ( !$ip ) return false;
  54                 if( is_array( $ip ) ) {
  55                   throw new MWException( "invalid value passed to " . __METHOD__ );
  56                 }
  57                 // IPv6 IPs with two "::" strings are ambiguous and thus invalid
  58                 return preg_match( '/^' . RE_IPV6_ADD . '(\/' . RE_IPV6_PREFIX . '|)$/', $ip) && ( substr_count($ip, '::') < 2);
  59         }
  60
  61         public static function isIPv4( $ip ) {
  62                 if ( !$ip ) return false;
  63                 return preg_match( '/^' . RE_IP_ADD . '(\/' . RE_IP_PREFIX . '|)$/', $ip);
  64         }
  65
  66         /**
  67          * Given an IP address in dotted-quad notation, returns an IPv6 octet.
  68          * See http://www.answers.com/topic/ipv4-compatible-address
  69          * IPs with the first 92 bits as zeros are reserved from IPv6
  70          * @param $ip quad-dotted IP address.
  71          * @return string
  72          */
  73         public static function IPv4toIPv6( $ip ) {
  74                 if ( !$ip ) return null;
  75                 // Convert only if needed
  76                 if ( self::isIPv6( $ip ) ) return $ip;
  77                 // IPv4 CIDRs
  78                 if ( strpos( $ip, '/' ) !== false ) {
  79                         $parts = explode( '/', $ip, 2 );
  80                         if ( count( $parts ) != 2 ) {
  81                                 return false;
  82                         }
  83                         $network = self::toUnsigned( $parts[0] );
  84                         if ( $network !== false && is_numeric( $parts[1] ) && $parts[1] >= 0 && $parts[1] <= 32 ) {
  85                                 $bits = $parts[1] + 96;
  86                                 return self::toOctet( $network ) . "/$bits";
  87                         } else {
  88                                 return false;
  89                         }
  90                 }
  91                 return self::toOctet( self::toUnsigned( $ip ) );
  92         }
  93
  94         /**
  95          * Given an IPv6 address in octet notation, returns an unsigned integer.
  96          * @param $ip octet ipv6 IP address.
  97          * @return string
  98          */
  99         public static function toUnsigned6( $ip ) {
 100                 if ( !$ip ) return null;
 101         $ip = explode(':', self::sanitizeIP( $ip ) );
 102         $r_ip = '';
 103         foreach ($ip as $v) {
 104                 $r_ip .= str_pad( $v, 4, 0, STR_PAD_LEFT );
 105         }
 106         $r_ip = wfBaseConvert( $r_ip, 16, 10 );
 107         return $r_ip;
 108         }
 109
 110         /**
 111          * Given an IPv6 address in octet notation, returns the expanded octet.
 112          * IPv4 IPs will be trimmed, thats it...
 113          * @param $ip octet ipv6 IP address.
 114          * @return string
 115          */
 116         public static function sanitizeIP( $ip ) {
 117                 if ( !$ip ) return null;
 118                 // Trim and return IPv4 addresses
 119                 if ( self::isIPv4($ip) ) return trim($ip);
 120                 // Only IPv6 addresses can be expanded
 121                 if ( !self::isIPv6($ip) ) return $ip;
 122                 // Remove any whitespaces, convert to upper case
 123                 $ip = strtoupper( trim($ip) );
 124                 // Expand zero abbreviations
 125                 if ( strpos( $ip, '::' ) !== false ) {
 126                 $ip = str_replace('::', str_repeat(':0', 8 - substr_count($ip, ':')) . ':', $ip);
 127         }
 128         // For IPs that start with "::", correct the final IP so that it starts with '0' and not ':'
 129         if ( $ip[0] == ':' ) $ip = "0$ip";
 130         // Remove leading zereos from each bloc as needed
 131         $ip = preg_replace( '/(^|:)0+' . RE_IPV6_WORD . '/', '$1$2', $ip );
 132         return $ip;
 133         }
 134
 135         /**
 136          * Given an unsigned integer, returns an IPv6 address in octet notation
 137          * @param $ip integer IP address.
 138          * @return string
 139          */
 140         public static function toOctet( $ip_int ) {
 141                 // Convert to padded uppercase hex
 142                 $ip_hex = wfBaseConvert($ip_int, 10, 16, 32, false);
 143                 // Seperate into 8 octets
 144                 $ip_oct = substr( $ip_hex, 0, 4 );
 145                 for ($n=1; $n < 8; $n++) {
 146                         $ip_oct .= ':' . substr($ip_hex, 4*$n, 4);
 147                 }
 148                 // NO leading zeroes
 149                 $ip_oct = preg_replace( '/(^|:)0+' . RE_IPV6_WORD . '/', '$1$2', $ip_oct );
 150         return $ip_oct;
 151         }
 152
 153         /**
 154          * Convert a network specification in IPv6 CIDR notation to an integer network and a number of bits
 155          * @return array(string, int)
 156          */
 157         public static function parseCIDR6( $range ) {
 158                 # Expand any IPv6 IP
 159                 $parts = explode( '/', IP::sanitizeIP( $range ), 2 );
 160                 if ( count( $parts ) != 2 ) {
 161                         return array( false, false );
 162                 }
 163                 $network = self::toUnsigned6( $parts[0] );
 164                 if ( $network !== false && is_numeric( $parts[1] ) && $parts[1] >= 0 && $parts[1] <= 128 ) {
 165                         $bits = $parts[1];
 166                         if ( $bits == 0 ) {
 167                                 $network = 0;
 168                         } else {
 169                         # Native 32 bit functions WONT work here!!!
 170                         # Convert to a padded binary number
 171                                 $network = wfBaseConvert( $network, 10, 2, 128 );
 172                         # Truncate the last (128-$bits) bits and replace them with zeros
 173                                 $network = str_pad( substr( $network, 0, $bits ), 128, 0, STR_PAD_RIGHT );
 174                         # Convert back to an integer
 175                                 $network = wfBaseConvert( $network, 2, 10 );
 176                         }
 177                 } else {
 178                         $network = false;
 179                         $bits = false;
 180                 }
 181                 return array( $network, $bits );
 182         }
 183
 184         /**
 185          * Given a string range in a number of formats, return the start and end of
 186          * the range in hexadecimal. For IPv6.
 187          *
 188          * Formats are:
 189          *     2001:0db8:85a3::7344/96                                   CIDR
 190          *     2001:0db8:85a3::7344 - 2001:0db8:85a3::7344   Explicit range
 191          *     2001:0db8:85a3::7344/96                                   Single IP
 192          * @return array(string, int)
 193          */
 194         public static function parseRange6( $range ) {
 195                 # Expand any IPv6 IP
 196                 $range = IP::sanitizeIP( $range );
 197                 if ( strpos( $range, '/' ) !== false ) {
 198                         # CIDR
 199                         list( $network, $bits ) = self::parseCIDR6( $range );
 200                         if ( $network === false ) {
 201                                 $start = $end = false;
 202                         } else {
 203                                 $start = wfBaseConvert( $network, 10, 16, 32, false );
 204                                 # Turn network to binary (again)
 205                                 $end = wfBaseConvert( $network, 10, 2, 128 );
 206                                 # Truncate the last (128-$bits) bits and replace them with ones
 207                                 $end = str_pad( substr( $end, 0, $bits ), 128, 1, STR_PAD_RIGHT );
 208                                 # Convert to hex
 209                                 $end = wfBaseConvert( $end, 2, 16, 32, false );
 210                                 # see toHex() comment
 211                                 $start = "v6-$start"; $end = "v6-$end";
 212                         }
 213                 } elseif ( strpos( $range, '-' ) !== false ) {
 214                         # Explicit range
 215                         list( $start, $end ) = array_map( 'trim', explode( '-', $range, 2 ) );
 216                         $start = self::toUnsigned6( $start ); $end = self::toUnsigned6( $end );
 217                         if ( $start > $end ) {
 218                                 $start = $end = false;
 219                         } else {
 220                                 $start = wfBaseConvert( $start, 10, 16, 32, false );
 221                                 $end = wfBaseConvert( $end, 10, 16, 32, false );
 222                         }
 223                         # see toHex() comment
 224                         $start = "v6-$start"; $end = "v6-$end";
 225                 } else {
 226                         # Single IP
 227                         $start = $end = self::toHex( $range );
 228                 }
 229                 if ( $start === false || $end === false ) {
 230                         return array( false, false );
 231                 } else {
 232                         return array( $start, $end );
 233                 }
 234     }
 235
 236         /**
 237          * Validate an IP address.
 238          * @return boolean True if it is valid.
 239          */
 240         public static function isValid( $ip ) {
 241                 return ( preg_match( '/^' . RE_IP_ADD . '$/', $ip) || preg_match( '/^' . RE_IPV6_ADD . '$/', $ip) );
 242         }
 243
 244         /**
 245          * Validate an IP Block.
 246          * @return boolean True if it is valid.
 247          */
 248         public static function isValidBlock( $ipblock ) {
 249                 return ( count(self::toArray($ipblock)) == 1 + 5 );
 250         }
 251
 252         /**
 253          * Determine if an IP address really is an IP address, and if it is public,
 254          * i.e. not RFC 1918 or similar
 255          * Comes from ProxyTools.php
 256          */
 257         public static function isPublic( $ip ) {
 258                 $n = self::toUnsigned( $ip );
 259                 if ( !$n ) {
 260                         return false;
 261                 }
 262
 263                 // ip2long accepts incomplete addresses, as well as some addresses
 264                 // followed by garbage characters. Check that it's really valid.
 265                 if( $ip != long2ip( $n ) ) {
 266                         return false;
 267                 }
 268
 269                 static $privateRanges = false;
 270                 if ( !$privateRanges ) {
 271                         $privateRanges = array(
 272                                 array( '10.0.0.0',    '10.255.255.255' ),   # RFC 1918 (private)
 273                                 array( '172.16.0.0',  '172.31.255.255' ),   #     "
 274                                 array( '192.168.0.0', '192.168.255.255' ),  #     "
 275                                 array( '0.0.0.0',     '0.255.255.255' ),    # this network
 276                                 array( '127.0.0.0',   '127.255.255.255' ),  # loopback
 277                         );
 278                 }
 279
 280                 foreach ( $privateRanges as $r ) {
 281                         $start = self::toUnsigned( $r[0] );
 282                         $end = self::toUnsigned( $r[1] );
 283                         if ( $n >= $start && $n <= $end ) {
 284                                 return false;
 285                         }
 286                 }
 287                 return true;
 288         }
 289
 290         /**
 291          * Split out an IP block as an array of 4 bytes and a mask,
 292          * return false if it can't be determined
 293          *
 294          * @param $ip string A quad dotted/octet IP address
 295          * @return array
 296          */
 297         public static function toArray( $ipblock ) {
 298                 $matches = array();
 299                 if( preg_match( '/^' . RE_IP_ADD . '(?:\/(?:'.RE_IP_PREFIX.'))?' . '$/', $ipblock, $matches ) ) {
 300                         return $matches;
 301                 } else if ( preg_match( '/^' . RE_IPV6_ADD . '(?:\/(?:'.RE_IPV6_PREFIX.'))?' . '$/', $ipblock, $matches ) ) {
 302                         return $matches;
 303                 } else {
 304                         return false;
 305                 }
 306         }
 307
 308         /**
 309          * Return a zero-padded hexadecimal representation of an IP address.
 310          *
 311          * Hexadecimal addresses are used because they can easily be extended to
 312          * IPv6 support. To separate the ranges, the return value from this
 313          * function for an IPv6 address will be prefixed with "v6-", a non-
 314          * hexadecimal string which sorts after the IPv4 addresses.
 315          *
 316          * @param $ip Quad dotted/octet IP address.
 317          * @return hexidecimal
 318          */
 319         public static function toHex( $ip ) {
 320                 $n = self::toUnsigned( $ip );
 321                 if ( $n !== false ) {
 322                         $n = ( self::isIPv6($ip) ) ? "v6-" . wfBaseConvert( $n, 10, 16, 32, false ) : wfBaseConvert( $n, 10, 16, 8, false );
 323                 }
 324                 return $n;
 325         }
 326
 327         /**
 328          * Given an IP address in dotted-quad/octet notation, returns an unsigned integer.
 329          * Like ip2long() except that it actually works and has a consistent error return value.
 330          * Comes from ProxyTools.php
 331          * @param $ip Quad dotted IP address.
 332          * @return integer
 333          */
 334         public static function toUnsigned( $ip ) {
 335                 // Use IPv6 functions if needed
 336                 if ( self::isIPv6( $ip ) ) {
 337                         return self::toUnsigned6( $ip );
 338                 }
 339                 if ( $ip == '255.255.255.255' ) {
 340                         $n = -1;
 341                 } else {
 342                         $n = ip2long( $ip );
 343                         if ( $n == -1 || $n === false ) { # Return value on error depends on PHP version
 344                                 $n = false;
 345                         }
 346                 }
 347                 if ( $n < 0 ) {
 348                         $n += pow( 2, 32 );
 349                 }
 350                 return $n;
 351         }
 352
 353         /**
 354          * Convert a dotted-quad IP to a signed integer
 355          * Returns false on failure
 356          */
 357         public static function toSigned( $ip ) {
 358                 if ( $ip == '255.255.255.255' ) {
 359                         $n = -1;
 360                 } else {
 361                         $n = ip2long( $ip );
 362                         if ( $n == -1 ) {
 363                                 $n = false;
 364                         }
 365                 }
 366                 return $n;
 367         }
 368
 369         /**
 370          * Convert a network specification in CIDR notation to an integer network and a number of bits
 371          * @return array(string, int)
 372          */
 373         public static function parseCIDR( $range ) {
 374                 $parts = explode( '/', $range, 2 );
 375                 if ( count( $parts ) != 2 ) {
 376                         return array( false, false );
 377                 }
 378                 $network = self::toSigned( $parts[0] );
 379                 if ( $network !== false && is_numeric( $parts[1] ) && $parts[1] >= 0 && $parts[1] <= 32 ) {
 380                         $bits = $parts[1];
 381                         if ( $bits == 0 ) {
 382                                 $network = 0;
 383                         } else {
 384                                 $network &= ~((1 << (32 - $bits)) - 1);
 385                         }
 386                         # Convert to unsigned
 387                         if ( $network < 0 ) {
 388                                 $network += pow( 2, 32 );
 389                         }
 390                 } else {
 391                         $network = false;
 392                         $bits = false;
 393                 }
 394                 return array( $network, $bits );
 395         }
 396
 397         /**
 398          * Given a string range in a number of formats, return the start and end of
 399          * the range in hexadecimal.
 400          *
 401          * Formats are:
 402          *     1.2.3.4/24          CIDR
 403          *     1.2.3.4 - 1.2.3.5   Explicit range
 404          *     1.2.3.4             Single IP
 405          *
 406          *     2001:0db8:85a3::7344/96                                   CIDR
 407          *     2001:0db8:85a3::7344 - 2001:0db8:85a3::7344   Explicit range
 408          *     2001:0db8:85a3::7344                                      Single IP
 409          * @return array(string, int)
 410          */
 411         public static function parseRange( $range ) {
 412                 // Use IPv6 functions if needed
 413                 if ( self::isIPv6( $range ) ) {
 414                         return self::parseRange6( $range );
 415                 }
 416                 if ( strpos( $range, '/' ) !== false ) {
 417                         # CIDR
 418                         list( $network, $bits ) = self::parseCIDR( $range );
 419                         if ( $network === false ) {
 420                                 $start = $end = false;
 421                         } else {
 422                                 $start = sprintf( '%08X', $network );
 423                                 $end = sprintf( '%08X', $network + pow( 2, (32 - $bits) ) - 1 );
 424                         }
 425                 } elseif ( strpos( $range, '-' ) !== false ) {
 426                         # Explicit range
 427                         list( $start, $end ) = array_map( 'trim', explode( '-', $range, 2 ) );
 428                         $start = self::toUnsigned( $start ); $end = self::toUnsigned( $end );
 429                         if ( $start > $end ) {
 430                                 $start = $end = false;
 431                         } else {
 432                                 $start = sprintf( '%08X', $start );
 433                                 $end = sprintf( '%08X', $end );
 434                         }
 435                 } else {
 436                         # Single IP
 437                         $start = $end = self::toHex( $range );
 438                 }
 439                 if ( $start === false || $end === false ) {
 440                         return array( false, false );
 441                 } else {
 442                         return array( $start, $end );
 443                 }
 444     }
 445
 446     /**
 447      * Determine if a given IPv4/IPv6 address is in a given CIDR network
 448      * @param $addr The address to check against the given range.
 449      * @param $range The range to check the given address against.
 450      * @return bool Whether or not the given address is in the given range.
 451      */
 452     public static function isInRange( $addr, $range ) {
 453     // Convert to IPv6 if needed
 454         $unsignedIP = self::toHex( $addr );
 455         list( $start, $end ) = self::parseRange( $range );
 456         return (($unsignedIP >= $start) && ($unsignedIP <= $end));
 457     }
 458
 459     /**
 460      * Convert some unusual representations of IPv4 addresses to their
 461      * canonical dotted quad representation.
 462      *
 463      * This currently only checks a few IPV4-to-IPv6 related cases.  More
 464      * unusual representations may be added later.
 465      *
 466      * @param $addr something that might be an IP address
 467      * @return valid dotted quad IPv4 address or null
 468      */
 469     public static function canonicalize( $addr ) {
 470         if ( self::isValid( $addr ) )
 471             return $addr;
 472
 473         // IPv6 loopback address
 474         $m = array();
 475         if ( preg_match( '/^0*' . RE_IPV6_GAP . '1$/', $addr, $m ) )
 476             return '127.0.0.1';
 477
 478         // IPv4-mapped and IPv4-compatible IPv6 addresses
 479         if ( preg_match( '/^' . RE_IPV6_V4_PREFIX . '(' . RE_IP_ADD . ')$/i', $addr, $m ) )
 480             return $m[1];
 481         if ( preg_match( '/^' . RE_IPV6_V4_PREFIX . RE_IPV6_WORD . ':' . RE_IPV6_WORD . '$/i', $addr, $m ) )
 482             return long2ip( ( hexdec( $m[1] ) << 16 ) + hexdec( $m[2] ) );
 483
 484         return null;  // give up
 485     }
 486 }
 487
 488