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