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                 $ip = trim( $ip );
 118                 if ( $ip === '' ) return null;
 119                 // Trim and return IPv4 addresses
 120                 if ( self::isIPv4($ip) ) return $ip;
 121                 // Only IPv6 addresses can be expanded
 122                 if ( !self::isIPv6($ip) ) return $ip;
 123                 // Remove any whitespaces, convert to upper case
 124                 $ip = strtoupper( $ip );
 125                 // Expand zero abbreviations
 126                 if ( strpos( $ip, '::' ) !== false ) {
 127                 $ip = str_replace('::', str_repeat(':0', 8 - substr_count($ip, ':')) . ':', $ip);
 128         }
 129         // For IPs that start with "::", correct the final IP so that it starts with '0' and not ':'
 130         if ( $ip[0] == ':' ) $ip = "0$ip";
 131         // Remove leading zereos from each bloc as needed
 132         $ip = preg_replace( '/(^|:)0+' . RE_IPV6_WORD . '/', '$1$2', $ip );
 133         return $ip;
 134         }
 135
 136         /**
 137          * Given an unsigned integer, returns an IPv6 address in octet notation
 138          * @param $ip integer IP address.
 139          * @return string
 140          */
 141         public static function toOctet( $ip_int ) {
 142                 // Convert to padded uppercase hex
 143                 $ip_hex = wfBaseConvert($ip_int, 10, 16, 32, false);
 144                 // Separate into 8 octets
 145                 $ip_oct = substr( $ip_hex, 0, 4 );
 146                 for ($n=1; $n < 8; $n++) {
 147                         $ip_oct .= ':' . substr($ip_hex, 4*$n, 4);
 148                 }
 149                 // NO leading zeroes
 150                 $ip_oct = preg_replace( '/(^|:)0+' . RE_IPV6_WORD . '/', '$1$2', $ip_oct );
 151         return $ip_oct;
 152         }
 153
 154         /**
 155          * Convert an IPv4 or IPv6 hexadecimal representation back to readable format
 156          */
 157         public static function formatHex( $hex ) {
 158                 if ( substr( $hex, 0, 3 ) == 'v6-' ) {
 159                         return self::hexToOctet( $hex );
 160                 } else {
 161                         return self::hexToQuad( $hex );
 162                 }
 163         }
 164
 165         /**
 166          * Given a hexadecimal number, returns to an IPv6 address in octet notation
 167          * @param $ip string hex IP
 168          * @return string
 169          */
 170         public static function hextoOctet( $ip_hex ) {
 171                 // Convert to padded uppercase hex
 172                 $ip_hex = str_pad( strtoupper($ip_hex), 32, '0');
 173                 // Separate into 8 octets
 174                 $ip_oct = substr( $ip_hex, 0, 4 );
 175                 for ($n=1; $n < 8; $n++) {
 176                         $ip_oct .= ':' . substr($ip_hex, 4*$n, 4);
 177                 }
 178                 // NO leading zeroes
 179                 $ip_oct = preg_replace( '/(^|:)0+' . RE_IPV6_WORD . '/', '$1$2', $ip_oct );
 180         return $ip_oct;
 181         }
 182
 183         /**
 184          * Converts a hexadecimal number to an IPv4 address in octet notation
 185          * @param $ip string Hex IP
 186          * @return string
 187          */
 188         public static function hexToQuad( $ip ) {
 189                 // Converts a hexadecimal IP to nnn.nnn.nnn.nnn format
 190                 $s = '';
 191                 for ( $i = 0; $i < 4; $i++ ) {
 192                         if ( $s !== '' ) {
 193                                 $s .= '.';
 194                         }
 195                         $s .= base_convert( substr( $ip, $i * 2, 2 ), 16, 10 );
 196                 }
 197                 return $s;
 198         }
 199
 200         /**
 201          * Convert a network specification in IPv6 CIDR notation to an integer network and a number of bits
 202          * @return array(string, int)
 203          */
 204         public static function parseCIDR6( $range ) {
 205                 # Expand any IPv6 IP
 206                 $parts = explode( '/', IP::sanitizeIP( $range ), 2 );
 207                 if ( count( $parts ) != 2 ) {
 208                         return array( false, false );
 209                 }
 210                 $network = self::toUnsigned6( $parts[0] );
 211                 if ( $network !== false && is_numeric( $parts[1] ) && $parts[1] >= 0 && $parts[1] <= 128 ) {
 212                         $bits = $parts[1];
 213                         if ( $bits == 0 ) {
 214                                 $network = 0;
 215                         } else {
 216                         # Native 32 bit functions WONT work here!!!
 217                         # Convert to a padded binary number
 218                                 $network = wfBaseConvert( $network, 10, 2, 128 );
 219                         # Truncate the last (128-$bits) bits and replace them with zeros
 220                                 $network = str_pad( substr( $network, 0, $bits ), 128, 0, STR_PAD_RIGHT );
 221                         # Convert back to an integer
 222                                 $network = wfBaseConvert( $network, 2, 10 );
 223                         }
 224                 } else {
 225                         $network = false;
 226                         $bits = false;
 227                 }
 228                 return array( $network, $bits );
 229         }
 230
 231         /**
 232          * Given a string range in a number of formats, return the start and end of
 233          * the range in hexadecimal. For IPv6.
 234          *
 235          * Formats are:
 236          *     2001:0db8:85a3::7344/96                                   CIDR
 237          *     2001:0db8:85a3::7344 - 2001:0db8:85a3::7344   Explicit range
 238          *     2001:0db8:85a3::7344/96                                   Single IP
 239          * @return array(string, int)
 240          */
 241         public static function parseRange6( $range ) {
 242                 # Expand any IPv6 IP
 243                 $range = IP::sanitizeIP( $range );
 244                 if ( strpos( $range, '/' ) !== false ) {
 245                         # CIDR
 246                         list( $network, $bits ) = self::parseCIDR6( $range );
 247                         if ( $network === false ) {
 248                                 $start = $end = false;
 249                         } else {
 250                                 $start = wfBaseConvert( $network, 10, 16, 32, false );
 251                                 # Turn network to binary (again)
 252                                 $end = wfBaseConvert( $network, 10, 2, 128 );
 253                                 # Truncate the last (128-$bits) bits and replace them with ones
 254                                 $end = str_pad( substr( $end, 0, $bits ), 128, 1, STR_PAD_RIGHT );
 255                                 # Convert to hex
 256                                 $end = wfBaseConvert( $end, 2, 16, 32, false );
 257                                 # see toHex() comment
 258                                 $start = "v6-$start"; $end = "v6-$end";
 259                         }
 260                 } elseif ( strpos( $range, '-' ) !== false ) {
 261                         # Explicit range
 262                         list( $start, $end ) = array_map( 'trim', explode( '-', $range, 2 ) );
 263                         $start = self::toUnsigned6( $start ); $end = self::toUnsigned6( $end );
 264                         if ( $start > $end ) {
 265                                 $start = $end = false;
 266                         } else {
 267                                 $start = wfBaseConvert( $start, 10, 16, 32, false );
 268                                 $end = wfBaseConvert( $end, 10, 16, 32, false );
 269                         }
 270                         # see toHex() comment
 271                         $start = "v6-$start"; $end = "v6-$end";
 272                 } else {
 273                         # Single IP
 274                         $start = $end = self::toHex( $range );
 275                 }
 276                 if ( $start === false || $end === false ) {
 277                         return array( false, false );
 278                 } else {
 279                         return array( $start, $end );
 280                 }
 281     }
 282
 283         /**
 284          * Validate an IP address.
 285          * @return boolean True if it is valid.
 286          */
 287         public static function isValid( $ip ) {
 288                 return ( preg_match( '/^' . RE_IP_ADD . '$/', $ip) || preg_match( '/^' . RE_IPV6_ADD . '$/', $ip) );
 289         }
 290
 291         /**
 292          * Validate an IP Block.
 293          * @return boolean True if it is valid.
 294          */
 295         public static function isValidBlock( $ipblock ) {
 296                 return ( count(self::toArray($ipblock)) == 1 + 5 );
 297         }
 298
 299         /**
 300          * Determine if an IP address really is an IP address, and if it is public,
 301          * i.e. not RFC 1918 or similar
 302          * Comes from ProxyTools.php
 303          */
 304         public static function isPublic( $ip ) {
 305                 $n = self::toUnsigned( $ip );
 306                 if ( !$n ) {
 307                         return false;
 308                 }
 309
 310                 // ip2long accepts incomplete addresses, as well as some addresses
 311                 // followed by garbage characters. Check that it's really valid.
 312                 if( $ip != long2ip( $n ) ) {
 313                         return false;
 314                 }
 315
 316                 static $privateRanges = false;
 317                 if ( !$privateRanges ) {
 318                         $privateRanges = array(
 319                                 array( '10.0.0.0',    '10.255.255.255' ),   # RFC 1918 (private)
 320                                 array( '172.16.0.0',  '172.31.255.255' ),   #     "
 321                                 array( '192.168.0.0', '192.168.255.255' ),  #     "
 322                                 array( '0.0.0.0',     '0.255.255.255' ),    # this network
 323                                 array( '127.0.0.0',   '127.255.255.255' ),  # loopback
 324                         );
 325                 }
 326
 327                 foreach ( $privateRanges as $r ) {
 328                         $start = self::toUnsigned( $r[0] );
 329                         $end = self::toUnsigned( $r[1] );
 330                         if ( $n >= $start && $n <= $end ) {
 331                                 return false;
 332                         }
 333                 }
 334                 return true;
 335         }
 336
 337         /**
 338          * Split out an IP block as an array of 4 bytes and a mask,
 339          * return false if it can't be determined
 340          *
 341          * @param $ip string A quad dotted/octet IP address
 342          * @return array
 343          */
 344         public static function toArray( $ipblock ) {
 345                 $matches = array();
 346                 if( preg_match( '/^' . RE_IP_ADD . '(?:\/(?:'.RE_IP_PREFIX.'))?' . '$/', $ipblock, $matches ) ) {
 347                         return $matches;
 348                 } else if ( preg_match( '/^' . RE_IPV6_ADD . '(?:\/(?:'.RE_IPV6_PREFIX.'))?' . '$/', $ipblock, $matches ) ) {
 349                         return $matches;
 350                 } else {
 351                         return false;
 352                 }
 353         }
 354
 355         /**
 356          * Return a zero-padded hexadecimal representation of an IP address.
 357          *
 358          * Hexadecimal addresses are used because they can easily be extended to
 359          * IPv6 support. To separate the ranges, the return value from this
 360          * function for an IPv6 address will be prefixed with "v6-", a non-
 361          * hexadecimal string which sorts after the IPv4 addresses.
 362          *
 363          * @param $ip Quad dotted/octet IP address.
 364          * @return hexidecimal
 365          */
 366         public static function toHex( $ip ) {
 367                 $n = self::toUnsigned( $ip );
 368                 if ( $n !== false ) {
 369                         $n = self::isIPv6($ip) ? "v6-" . wfBaseConvert( $n, 10, 16, 32, false ) : wfBaseConvert( $n, 10, 16, 8, false );
 370                 }
 371                 return $n;
 372         }
 373
 374         /**
 375          * Given an IP address in dotted-quad/octet notation, returns an unsigned integer.
 376          * Like ip2long() except that it actually works and has a consistent error return value.
 377          * Comes from ProxyTools.php
 378          * @param $ip Quad dotted IP address.
 379          * @return integer
 380          */
 381         public static function toUnsigned( $ip ) {
 382                 // Use IPv6 functions if needed
 383                 if ( self::isIPv6( $ip ) ) {
 384                         return self::toUnsigned6( $ip );
 385                 }
 386                 if ( $ip == '255.255.255.255' ) {
 387                         $n = -1;
 388                 } else {
 389                         $n = ip2long( $ip );
 390                         if ( $n == -1 || $n === false ) { # Return value on error depends on PHP version
 391                                 $n = false;
 392                         }
 393                 }
 394                 if ( $n < 0 ) {
 395                         $n += pow( 2, 32 );
 396                 }
 397                 return $n;
 398         }
 399
 400         /**
 401          * Convert a dotted-quad IP to a signed integer
 402          * Returns false on failure
 403          */
 404         public static function toSigned( $ip ) {
 405                 if ( $ip == '255.255.255.255' ) {
 406                         $n = -1;
 407                 } else {
 408                         $n = ip2long( $ip );
 409                         if ( $n == -1 ) {
 410                                 $n = false;
 411                         }
 412                 }
 413                 return $n;
 414         }
 415
 416         /**
 417          * Convert a network specification in CIDR notation to an integer network and a number of bits
 418          * @return array(string, int)
 419          */
 420         public static function parseCIDR( $range ) {
 421                 $parts = explode( '/', $range, 2 );
 422                 if ( count( $parts ) != 2 ) {
 423                         return array( false, false );
 424                 }
 425                 $network = self::toSigned( $parts[0] );
 426                 if ( $network !== false && is_numeric( $parts[1] ) && $parts[1] >= 0 && $parts[1] <= 32 ) {
 427                         $bits = $parts[1];
 428                         if ( $bits == 0 ) {
 429                                 $network = 0;
 430                         } else {
 431                                 $network &= ~((1 << (32 - $bits)) - 1);
 432                         }
 433                         # Convert to unsigned
 434                         if ( $network < 0 ) {
 435                                 $network += pow( 2, 32 );
 436                         }
 437                 } else {
 438                         $network = false;
 439                         $bits = false;
 440                 }
 441                 return array( $network, $bits );
 442         }
 443
 444         /**
 445          * Given a string range in a number of formats, return the start and end of
 446          * the range in hexadecimal.
 447          *
 448          * Formats are:
 449          *     1.2.3.4/24          CIDR
 450          *     1.2.3.4 - 1.2.3.5   Explicit range
 451          *     1.2.3.4             Single IP
 452          *
 453          *     2001:0db8:85a3::7344/96                                   CIDR
 454          *     2001:0db8:85a3::7344 - 2001:0db8:85a3::7344   Explicit range
 455          *     2001:0db8:85a3::7344                                      Single IP
 456          * @return array(string, int)
 457          */
 458         public static function parseRange( $range ) {
 459                 // Use IPv6 functions if needed
 460                 if ( self::isIPv6( $range ) ) {
 461                         return self::parseRange6( $range );
 462                 }
 463                 if ( strpos( $range, '/' ) !== false ) {
 464                         # CIDR
 465                         list( $network, $bits ) = self::parseCIDR( $range );
 466                         if ( $network === false ) {
 467                                 $start = $end = false;
 468                         } else {
 469                                 $start = sprintf( '%08X', $network );
 470                                 $end = sprintf( '%08X', $network + pow( 2, (32 - $bits) ) - 1 );
 471                         }
 472                 } elseif ( strpos( $range, '-' ) !== false ) {
 473                         # Explicit range
 474                         list( $start, $end ) = array_map( 'trim', explode( '-', $range, 2 ) );
 475                         if( self::isIPAddress( $start ) && self::isIPAddress( $end ) ) {
 476                                 $start = self::toUnsigned( $start ); $end = self::toUnsigned( $end );
 477                                 if ( $start > $end ) {
 478                                         $start = $end = false;
 479                                 } else {
 480                                         $start = sprintf( '%08X', $start );
 481                                         $end = sprintf( '%08X', $end );
 482                                 }
 483                         } else {
 484                                 $start = $end = false;
 485                         }
 486                 } else {
 487                         # Single IP
 488                         $start = $end = self::toHex( $range );
 489                 }
 490                 if ( $start === false || $end === false ) {
 491                         return array( false, false );
 492                 } else {
 493                         return array( $start, $end );
 494                 }
 495     }
 496
 497     /**
 498      * Determine if a given IPv4/IPv6 address is in a given CIDR network
 499      * @param $addr The address to check against the given range.
 500      * @param $range The range to check the given address against.
 501      * @return bool Whether or not the given address is in the given range.
 502      */
 503     public static function isInRange( $addr, $range ) {
 504     // Convert to IPv6 if needed
 505         $hexIP = self::toHex( $addr );
 506         list( $start, $end ) = self::parseRange( $range );
 507         return (strcmp($hexIP, $start) >= 0 &&
 508                 strcmp($hexIP, $end) <= 0);
 509     }
 510
 511     /**
 512      * Convert some unusual representations of IPv4 addresses to their
 513      * canonical dotted quad representation.
 514      *
 515      * This currently only checks a few IPV4-to-IPv6 related cases.  More
 516      * unusual representations may be added later.
 517      *
 518      * @param $addr something that might be an IP address
 519      * @return valid dotted quad IPv4 address or null
 520      */
 521     public static function canonicalize( $addr ) {
 522                 if ( self::isValid( $addr ) )
 523                         return $addr;
 524
 525                 // Annoying IPv6 representations like ::ffff:1.2.3.4
 526                 if ( strpos($addr,':') !==false && strpos($addr,'.') !==false ) {
 527                         $addr = str_replace( '.', ':', $addr );
 528                         if( IP::isIPv6( $addr ) )
 529                                 return $addr;
 530                 }
 531
 532                 // IPv6 loopback address
 533                 $m = array();
 534                 if ( preg_match( '/^0*' . RE_IPV6_GAP . '1$/', $addr, $m ) )
 535                         return '127.0.0.1';
 536
 537                 // IPv4-mapped and IPv4-compatible IPv6 addresses
 538                 if ( preg_match( '/^' . RE_IPV6_V4_PREFIX . '(' . RE_IP_ADD . ')$/i', $addr, $m ) )
 539                     return $m[1];
 540                 if ( preg_match( '/^' . RE_IPV6_V4_PREFIX . RE_IPV6_WORD . ':' . RE_IPV6_WORD . '$/i', $addr, $m ) )
 541                     return long2ip( ( hexdec( $m[1] ) << 16 ) + hexdec( $m[2] ) );
 542
 543                 return null;  // give up
 544     }
 545 }