includes/IP.php

   1 <?php
   2 /**
   3  * Functions and constants to play with IP addresses and ranges
   4  *
   5  * @file
   6  * @author "Ashar Voultoiz" <hashar@altern.org>
   7  * @license GPL v2 or later
   8  */
   9
  10 // Some regex definition to "play" with IP address and IP address blocks
  11
  12 // An IP is made of 4 bytes from x00 to xFF which is d0 to d255
  13 define( 'RE_IP_BYTE', '(25[0-5]|2[0-4][0-9]|1[0-9][0-9]|0?[0-9]?[0-9])');
  14 define( 'RE_IP_ADD' , RE_IP_BYTE . '\.' . RE_IP_BYTE . '\.' . RE_IP_BYTE . '\.' . RE_IP_BYTE );
  15 // An IPv4 block is an IP address and a prefix (d1 to d32)
  16 define( 'RE_IP_PREFIX', '(3[0-2]|[12]?\d)');
  17 define( 'RE_IP_BLOCK', RE_IP_ADD . '\/' . RE_IP_PREFIX);
  18 // For IPv6 canonicalization (NOT for strict validation; these are quite lax!)
  19 define( 'RE_IPV6_WORD', '([0-9A-Fa-f]{1,4})' );
  20 define( 'RE_IPV6_GAP', ':(?:0+:)*(?::(?:0+:)*)?' );
  21 define( 'RE_IPV6_V4_PREFIX', '0*' . RE_IPV6_GAP . '(?:ffff:)?' );
  22 // An IPv6 block is an IP address and a prefix (d1 to d128)
  23 define( 'RE_IPV6_PREFIX', '(12[0-8]|1[01][0-9]|[1-9]?\d)');
  24 // An IPv6 IP is made up of 8 octets. However abbreviations like "::" can be used.
  25 // This is lax! Number of octets/double colons validation not done.
  26 define( 'RE_IPV6_ADD',
  27         '(' .
  28                 ':(:' . RE_IPV6_WORD . '){1,7}' . // IPs that start with ":"
  29         '|' .
  30                 RE_IPV6_WORD . '(:{1,2}' . RE_IPV6_WORD . '|::$){1,7}' . // IPs that don't start with ":"
  31         ')'
  32 );
  33 define( 'RE_IPV6_BLOCK', RE_IPV6_ADD . '\/' . RE_IPV6_PREFIX );
  34 // This might be useful for regexps used elsewhere, matches any IPv6 or IPv6 address or network
  35 define( 'IP_ADDRESS_STRING',
  36         '(?:' .
  37                 RE_IP_ADD . '(\/' . RE_IP_PREFIX . '|)' . // IPv4
  38         '|' .
  39                 RE_IPV6_ADD . '(\/' . RE_IPV6_PREFIX . '|)' . // IPv6
  40         ')'
  41 );
  42
  43 /**
  44  * A collection of public static functions to play with IP address
  45  * and IP blocks.
  46  */
  47 class IP {
  48         /**
  49          * Given a string, determine if it as valid IP
  50          * Unlike isValid(), this looks for networks too
  51          * @param $ip IP address.
  52          * @return string
  53          */
  54         public static function isIPAddress( $ip ) {
  55                 if ( !$ip ) return false;
  56                 if ( is_array( $ip ) ) {
  57                   throw new MWException( "invalid value passed to " . __METHOD__ );
  58                 }
  59                 // IPv6 IPs with two "::" strings are ambiguous and thus invalid
  60                 return preg_match( '/^' . IP_ADDRESS_STRING . '$/', $ip) && ( substr_count($ip, '::') < 2 );
  61         }
  62
  63         public static function isIPv6( $ip ) {
  64                 if ( !$ip ) return false;
  65                 if( is_array( $ip ) ) {
  66                         throw new MWException( "invalid value passed to " . __METHOD__ );
  67                 }
  68                 $doubleColons = substr_count($ip, '::');
  69                 // IPv6 IPs with two "::" strings are ambiguous and thus invalid
  70                 return preg_match( '/^' . RE_IPV6_ADD . '(\/' . RE_IPV6_PREFIX . '|)$/', $ip)
  71                         && ( $doubleColons == 1 || substr_count($ip,':') == 7 );
  72         }
  73
  74         public static function isIPv4( $ip ) {
  75                 if ( !$ip ) return false;
  76                 return preg_match( '/^' . RE_IP_ADD . '(\/' . RE_IP_PREFIX . '|)$/', $ip);
  77         }
  78
  79         /**
  80          * Given an IP address in dotted-quad notation, returns an IPv6 octet.
  81          * See http://www.answers.com/topic/ipv4-compatible-address
  82          * IPs with the first 92 bits as zeros are reserved from IPv6
  83          * @param $ip quad-dotted IP address.
  84          * @return string
  85          */
  86         public static function IPv4toIPv6( $ip ) {
  87                 if ( !$ip ) return null;
  88                 // Convert only if needed
  89                 if ( self::isIPv6( $ip ) ) return $ip;
  90                 // IPv4 CIDRs
  91                 if ( strpos( $ip, '/' ) !== false ) {
  92                         $parts = explode( '/', $ip, 2 );
  93                         if ( count( $parts ) != 2 ) {
  94                                 return false;
  95                         }
  96                         $network = self::toUnsigned( $parts[0] );
  97                         if ( $network !== false && is_numeric( $parts[1] ) && $parts[1] >= 0 && $parts[1] <= 32 ) {
  98                                 $bits = $parts[1] + 96;
  99                                 return self::toOctet( $network ) . "/$bits";
 100                         } else {
 101                                 return false;
 102                         }
 103                 }
 104                 return self::toOctet( self::toUnsigned( $ip ) );
 105         }
 106
 107         /**
 108          * Given an IPv6 address in octet notation, returns an unsigned integer.
 109          * @param $ip octet ipv6 IP address.
 110          * @return string
 111          */
 112         public static function toUnsigned6( $ip ) {
 113                 if ( !$ip ) return null;
 114                 $ip = explode(':', self::sanitizeIP( $ip ) );
 115                 $r_ip = '';
 116                 foreach ($ip as $v) {
 117                         $r_ip .= str_pad( $v, 4, 0, STR_PAD_LEFT );
 118                 }
 119                 $r_ip = wfBaseConvert( $r_ip, 16, 10 );
 120                 return $r_ip;
 121         }
 122
 123         /**
 124          * Given an IPv6 address in octet notation, returns the expanded octet.
 125          * IPv4 IPs will be trimmed, thats it...
 126          * @param $ip octet ipv6 IP address.
 127          * @return string
 128          */
 129         public static function sanitizeIP( $ip ) {
 130                 $ip = trim( $ip );
 131                 if ( $ip === '' ) return null;
 132                 // Trim and return IPv4 addresses
 133                 if ( self::isIPv4($ip) ) return $ip;
 134                 // Only IPv6 addresses can be expanded
 135                 if ( !self::isIPv6($ip) ) return $ip;
 136                 // Remove any whitespaces, convert to upper case
 137                 $ip = strtoupper( $ip );
 138                 // Expand zero abbreviations
 139                 $abbrevPos = strpos( $ip, '::' );
 140                 if ( $abbrevPos !== false ) {
 141                         // If the '::' is at the beginning...
 142                         if( $abbrevPos == 0 ) {
 143                                 $repeat = '0:'; $extra = ''; $pad = 9; // 7+2 (due to '::')
 144                         // If the '::' is at the end...
 145                         } else if( $abbrevPos == (strlen($ip)-2) ) {
 146                                 $repeat = ':0'; $extra = ''; $pad = 9; // 7+2 (due to '::')
 147                         // If the '::' is at the end...
 148                         } else {
 149                                 $repeat = ':0'; $extra = ':'; $pad = 8; // 6+2 (due to '::')
 150                         }
 151                         $ip = str_replace('::', str_repeat($repeat, $pad-substr_count($ip,':')).$extra, $ip);
 152                 }
 153                 // Remove leading zereos from each bloc as needed
 154                 $ip = preg_replace( '/(^|:)0+' . RE_IPV6_WORD . '/', '$1$2', $ip );
 155                 return $ip;
 156         }
 157
 158         /**
 159          * Given an unsigned integer, returns an IPv6 address in octet notation
 160          * @param $ip_int integer IP address.
 161          * @return string
 162          */
 163         public static function toOctet( $ip_int ) {
 164                 // Convert to padded uppercase hex
 165                 $ip_hex = wfBaseConvert($ip_int, 10, 16, 32, false);
 166                 // Separate into 8 octets
 167                 $ip_oct = substr( $ip_hex, 0, 4 );
 168                 for ($n=1; $n < 8; $n++) {
 169                         $ip_oct .= ':' . substr($ip_hex, 4*$n, 4);
 170                 }
 171                 // NO leading zeroes
 172                 $ip_oct = preg_replace( '/(^|:)0+' . RE_IPV6_WORD . '/', '$1$2', $ip_oct );
 173                 return $ip_oct;
 174         }
 175
 176         /**
 177          * Convert an IPv4 or IPv6 hexadecimal representation back to readable format
 178          */
 179         public static function formatHex( $hex ) {
 180                 if ( substr( $hex, 0, 3 ) == 'v6-' ) {
 181                         return self::hexToOctet( $hex );
 182                 } else {
 183                         return self::hexToQuad( $hex );
 184                 }
 185         }
 186
 187         /**
 188          * Given a hexadecimal number, returns to an IPv6 address in octet notation
 189          * @param $ip_hex string hex IP
 190          * @return string
 191          */
 192         public static function hextoOctet( $ip_hex ) {
 193                 // Convert to padded uppercase hex
 194                 $ip_hex = str_pad( strtoupper($ip_hex), 32, '0');
 195                 // Separate into 8 octets
 196                 $ip_oct = substr( $ip_hex, 0, 4 );
 197                 for ($n=1; $n < 8; $n++) {
 198                         $ip_oct .= ':' . substr($ip_hex, 4*$n, 4);
 199                 }
 200                 // NO leading zeroes
 201                 $ip_oct = preg_replace( '/(^|:)0+' . RE_IPV6_WORD . '/', '$1$2', $ip_oct );
 202                 return $ip_oct;
 203         }
 204
 205         /**
 206          * Converts a hexadecimal number to an IPv4 address in octet notation
 207          * @param $ip string Hex IP
 208          * @return string
 209          */
 210         public static function hexToQuad( $ip ) {
 211                 // Converts a hexadecimal IP to nnn.nnn.nnn.nnn format
 212                 $s = '';
 213                 for ( $i = 0; $i < 4; $i++ ) {
 214                         if ( $s !== '' ) {
 215                                 $s .= '.';
 216                         }
 217                         $s .= base_convert( substr( $ip, $i * 2, 2 ), 16, 10 );
 218                 }
 219                 return $s;
 220         }
 221
 222         /**
 223          * Convert a network specification in IPv6 CIDR notation to an integer network and a number of bits
 224          * @return array(string, int)
 225          */
 226         public static function parseCIDR6( $range ) {
 227                 # Expand any IPv6 IP
 228                 $parts = explode( '/', IP::sanitizeIP( $range ), 2 );
 229                 if ( count( $parts ) != 2 ) {
 230                         return array( false, false );
 231                 }
 232                 $network = self::toUnsigned6( $parts[0] );
 233                 if ( $network !== false && is_numeric( $parts[1] ) && $parts[1] >= 0 && $parts[1] <= 128 ) {
 234                         $bits = $parts[1];
 235                         if ( $bits == 0 ) {
 236                                 $network = 0;
 237                         } else {
 238                         # Native 32 bit functions WONT work here!!!
 239                         # Convert to a padded binary number
 240                                 $network = wfBaseConvert( $network, 10, 2, 128 );
 241                         # Truncate the last (128-$bits) bits and replace them with zeros
 242                                 $network = str_pad( substr( $network, 0, $bits ), 128, 0, STR_PAD_RIGHT );
 243                         # Convert back to an integer
 244                                 $network = wfBaseConvert( $network, 2, 10 );
 245                         }
 246                 } else {
 247                         $network = false;
 248                         $bits = false;
 249                 }
 250                 return array( $network, $bits );
 251         }
 252
 253         /**
 254          * Given a string range in a number of formats, return the start and end of
 255          * the range in hexadecimal. For IPv6.
 256          *
 257          * Formats are:
 258          *     2001:0db8:85a3::7344/96                                   CIDR
 259          *     2001:0db8:85a3::7344 - 2001:0db8:85a3::7344   Explicit range
 260          *     2001:0db8:85a3::7344/96                                   Single IP
 261          * @return array(string, int)
 262          */
 263         public static function parseRange6( $range ) {
 264                 # Expand any IPv6 IP
 265                 $range = IP::sanitizeIP( $range );
 266                 if ( strpos( $range, '/' ) !== false ) {
 267                         # CIDR
 268                         list( $network, $bits ) = self::parseCIDR6( $range );
 269                         if ( $network === false ) {
 270                                 $start = $end = false;
 271                         } else {
 272                                 $start = wfBaseConvert( $network, 10, 16, 32, false );
 273                                 # Turn network to binary (again)
 274                                 $end = wfBaseConvert( $network, 10, 2, 128 );
 275                                 # Truncate the last (128-$bits) bits and replace them with ones
 276                                 $end = str_pad( substr( $end, 0, $bits ), 128, 1, STR_PAD_RIGHT );
 277                                 # Convert to hex
 278                                 $end = wfBaseConvert( $end, 2, 16, 32, false );
 279                                 # see toHex() comment
 280                                 $start = "v6-$start"; $end = "v6-$end";
 281                         }
 282                 } elseif ( strpos( $range, '-' ) !== false ) {
 283                         # Explicit range
 284                         list( $start, $end ) = array_map( 'trim', explode( '-', $range, 2 ) );
 285                         $start = self::toUnsigned6( $start ); $end = self::toUnsigned6( $end );
 286                         if ( $start > $end ) {
 287                                 $start = $end = false;
 288                         } else {
 289                                 $start = wfBaseConvert( $start, 10, 16, 32, false );
 290                                 $end = wfBaseConvert( $end, 10, 16, 32, false );
 291                         }
 292                         # see toHex() comment
 293                         $start = "v6-$start"; $end = "v6-$end";
 294                 } else {
 295                         # Single IP
 296                         $start = $end = self::toHex( $range );
 297                 }
 298                 if ( $start === false || $end === false ) {
 299                         return array( false, false );
 300                 } else {
 301                         return array( $start, $end );
 302                 }
 303         }
 304
 305         /**
 306          * Validate an IP address.
 307          * @return boolean True if it is valid.
 308          */
 309         public static function isValid( $ip ) {
 310                 return ( preg_match( '/^' . RE_IP_ADD . '$/', $ip) || preg_match( '/^' . RE_IPV6_ADD . '$/', $ip) );
 311         }
 312
 313         /**
 314          * Validate an IP Block.
 315          * @return boolean True if it is valid.
 316          */
 317         public static function isValidBlock( $ipblock ) {
 318                 return ( count(self::toArray($ipblock)) == 1 + 5 );
 319         }
 320
 321         /**
 322          * Determine if an IP address really is an IP address, and if it is public,
 323          * i.e. not RFC 1918 or similar
 324          * Comes from ProxyTools.php
 325          */
 326         public static function isPublic( $ip ) {
 327                 $n = self::toUnsigned( $ip );
 328                 if ( !$n ) {
 329                         return false;
 330                 }
 331
 332                 // ip2long accepts incomplete addresses, as well as some addresses
 333                 // followed by garbage characters. Check that it's really valid.
 334                 if( $ip != long2ip( $n ) ) {
 335                         return false;
 336                 }
 337
 338                 static $privateRanges = false;
 339                 if ( !$privateRanges ) {
 340                         $privateRanges = array(
 341                                 array( '10.0.0.0',    '10.255.255.255' ),   # RFC 1918 (private)
 342                                 array( '172.16.0.0',  '172.31.255.255' ),   #     "
 343                                 array( '192.168.0.0', '192.168.255.255' ),  #     "
 344                                 array( '0.0.0.0',     '0.255.255.255' ),    # this network
 345                                 array( '127.0.0.0',   '127.255.255.255' ),  # loopback
 346                         );
 347                 }
 348
 349                 foreach ( $privateRanges as $r ) {
 350                         $start = self::toUnsigned( $r[0] );
 351                         $end = self::toUnsigned( $r[1] );
 352                         if ( $n >= $start && $n <= $end ) {
 353                                 return false;
 354                         }
 355                 }
 356                 return true;
 357         }
 358
 359         /**
 360          * Split out an IP block as an array of 4 bytes and a mask,
 361          * return false if it can't be determined
 362          *
 363          * @param $ipblock string A quad dotted/octet IP address
 364          * @return array
 365          */
 366         public static function toArray( $ipblock ) {
 367                 $matches = array();
 368                 if( preg_match( '/^' . RE_IP_ADD . '(?:\/(?:'.RE_IP_PREFIX.'))?' . '$/', $ipblock, $matches ) ) {
 369                         return $matches;
 370                 } else if ( preg_match( '/^' . RE_IPV6_ADD . '(?:\/(?:'.RE_IPV6_PREFIX.'))?' . '$/', $ipblock, $matches ) ) {
 371                         return $matches;
 372                 } else {
 373                         return false;
 374                 }
 375         }
 376
 377         /**
 378          * Return a zero-padded hexadecimal representation of an IP address.
 379          *
 380          * Hexadecimal addresses are used because they can easily be extended to
 381          * IPv6 support. To separate the ranges, the return value from this
 382          * function for an IPv6 address will be prefixed with "v6-", a non-
 383          * hexadecimal string which sorts after the IPv4 addresses.
 384          *
 385          * @param $ip Quad dotted/octet IP address.
 386          * @return hexidecimal
 387          */
 388         public static function toHex( $ip ) {
 389                 $n = self::toUnsigned( $ip );
 390                 if ( $n !== false ) {
 391                         $n = self::isIPv6($ip) ? "v6-" . wfBaseConvert( $n, 10, 16, 32, false ) : wfBaseConvert( $n, 10, 16, 8, false );
 392                 }
 393                 return $n;
 394         }
 395
 396         /**
 397          * Given an IP address in dotted-quad/octet notation, returns an unsigned integer.
 398          * Like ip2long() except that it actually works and has a consistent error return value.
 399          * Comes from ProxyTools.php
 400          * @param $ip Quad dotted IP address.
 401          * @return integer
 402          */
 403         public static function toUnsigned( $ip ) {
 404                 // Use IPv6 functions if needed
 405                 if ( self::isIPv6( $ip ) ) {
 406                         return self::toUnsigned6( $ip );
 407                 }
 408                 if ( $ip == '255.255.255.255' ) {
 409                         $n = -1;
 410                 } else {
 411                         $n = ip2long( $ip );
 412                         if ( $n == -1 || $n === false ) { # Return value on error depends on PHP version
 413                                 $n = false;
 414                         }
 415                 }
 416                 if ( $n < 0 ) {
 417                         $n += pow( 2, 32 );
 418                 }
 419                 return $n;
 420         }
 421
 422         /**
 423          * Convert a dotted-quad IP to a signed integer
 424          * Returns false on failure
 425          */
 426         public static function toSigned( $ip ) {
 427                 if ( $ip == '255.255.255.255' ) {
 428                         $n = -1;
 429                 } else {
 430                         $n = ip2long( $ip );
 431                         if ( $n == -1 ) {
 432                                 $n = false;
 433                         }
 434                 }
 435                 return $n;
 436         }
 437
 438         /**
 439          * Convert a network specification in CIDR notation to an integer network and a number of bits
 440          * @return array(string, int)
 441          */
 442         public static function parseCIDR( $range ) {
 443                 $parts = explode( '/', $range, 2 );
 444                 if ( count( $parts ) != 2 ) {
 445                         return array( false, false );
 446                 }
 447                 $network = self::toSigned( $parts[0] );
 448                 if ( $network !== false && is_numeric( $parts[1] ) && $parts[1] >= 0 && $parts[1] <= 32 ) {
 449                         $bits = $parts[1];
 450                         if ( $bits == 0 ) {
 451                                 $network = 0;
 452                         } else {
 453                                 $network &= ~((1 << (32 - $bits)) - 1);
 454                         }
 455                         # Convert to unsigned
 456                         if ( $network < 0 ) {
 457                                 $network += pow( 2, 32 );
 458                         }
 459                 } else {
 460                         $network = false;
 461                         $bits = false;
 462                 }
 463                 return array( $network, $bits );
 464         }
 465
 466         /**
 467          * Given a string range in a number of formats, return the start and end of
 468          * the range in hexadecimal.
 469          *
 470          * Formats are:
 471          *     1.2.3.4/24          CIDR
 472          *     1.2.3.4 - 1.2.3.5   Explicit range
 473          *     1.2.3.4             Single IP
 474          *
 475          *     2001:0db8:85a3::7344/96                                   CIDR
 476          *     2001:0db8:85a3::7344 - 2001:0db8:85a3::7344   Explicit range
 477          *     2001:0db8:85a3::7344                                      Single IP
 478          * @return array(string, int)
 479          */
 480         public static function parseRange( $range ) {
 481                 // Use IPv6 functions if needed
 482                 if ( self::isIPv6( $range ) ) {
 483                         return self::parseRange6( $range );
 484                 }
 485                 if ( strpos( $range, '/' ) !== false ) {
 486                         # CIDR
 487                         list( $network, $bits ) = self::parseCIDR( $range );
 488                         if ( $network === false ) {
 489                                 $start = $end = false;
 490                         } else {
 491                                 $start = sprintf( '%08X', $network );
 492                                 $end = sprintf( '%08X', $network + pow( 2, (32 - $bits) ) - 1 );
 493                         }
 494                 } elseif ( strpos( $range, '-' ) !== false ) {
 495                         # Explicit range
 496                         list( $start, $end ) = array_map( 'trim', explode( '-', $range, 2 ) );
 497                         if( self::isIPAddress( $start ) && self::isIPAddress( $end ) ) {
 498                                 $start = self::toUnsigned( $start ); $end = self::toUnsigned( $end );
 499                                 if ( $start > $end ) {
 500                                         $start = $end = false;
 501                                 } else {
 502                                         $start = sprintf( '%08X', $start );
 503                                         $end = sprintf( '%08X', $end );
 504                                 }
 505                         } else {
 506                                 $start = $end = false;
 507                         }
 508                 } else {
 509                         # Single IP
 510                         $start = $end = self::toHex( $range );
 511                 }
 512                 if ( $start === false || $end === false ) {
 513                         return array( false, false );
 514                 } else {
 515                         return array( $start, $end );
 516                 }
 517         }
 518
 519         /**
 520          * Determine if a given IPv4/IPv6 address is in a given CIDR network
 521          * @param $addr The address to check against the given range.
 522          * @param $range The range to check the given address against.
 523          * @return bool Whether or not the given address is in the given range.
 524          */
 525         public static function isInRange( $addr, $range ) {
 526                 // Convert to IPv6 if needed
 527                 $hexIP = self::toHex( $addr );
 528                 list( $start, $end ) = self::parseRange( $range );
 529                 return (strcmp($hexIP, $start) >= 0 &&
 530                 strcmp($hexIP, $end) <= 0);
 531         }
 532
 533         /**
 534          * Convert some unusual representations of IPv4 addresses to their
 535          * canonical dotted quad representation.
 536          *
 537          * This currently only checks a few IPV4-to-IPv6 related cases.  More
 538          * unusual representations may be added later.
 539          *
 540          * @param $addr something that might be an IP address
 541          * @return valid dotted quad IPv4 address or null
 542          */
 543         public static function canonicalize( $addr ) {
 544                 if ( self::isValid( $addr ) )
 545                         return $addr;
 546
 547                 // Turn mapped addresses from ::ce:ffff:1.2.3.4 to 1.2.3.4
 548                 if ( strpos($addr,':') !==false && strpos($addr,'.') !==false ) {
 549                         $addr = substr( $addr, strrpos($addr,':')+1 );
 550                         if( self::isIPv4($addr) ) return $addr;
 551                 }
 552
 553                 // IPv6 loopback address
 554                 $m = array();
 555                 if ( preg_match( '/^0*' . RE_IPV6_GAP . '1$/', $addr, $m ) )
 556                         return '127.0.0.1';
 557
 558                 // IPv4-mapped and IPv4-compatible IPv6 addresses
 559                 if ( preg_match( '/^' . RE_IPV6_V4_PREFIX . '(' . RE_IP_ADD . ')$/i', $addr, $m ) )
 560                         return $m[1];
 561                 if ( preg_match( '/^' . RE_IPV6_V4_PREFIX . RE_IPV6_WORD . ':' . RE_IPV6_WORD . '$/i', $addr, $m ) )
 562                         return long2ip( ( hexdec( $m[1] ) << 16 ) + hexdec( $m[2] ) );
 563
 564                 return null;  // give up
 565         }
 566 }