includes/utils/IP.php

   1 <?php
   2 /**
   3  * Functions and constants to play with IP addresses and ranges
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  * http://www.gnu.org/copyleft/gpl.html
  19  *
  20  * @file
  21  * @author Antoine Musso "<hashar at free dot fr>", Aaron Schulz
  22  */
  23
  24 // Some regex definition to "play" with IP address and IP address blocks
  25
  26 // An IPv4 address is made of 4 bytes from x00 to xFF which is d0 to d255
  27 define( 'RE_IP_BYTE', '(25[0-5]|2[0-4][0-9]|1[0-9][0-9]|0?[0-9]?[0-9])' );
  28 define( 'RE_IP_ADD', RE_IP_BYTE . '\.' . RE_IP_BYTE . '\.' . RE_IP_BYTE . '\.' . RE_IP_BYTE );
  29 // An IPv4 block is an IP address and a prefix (d1 to d32)
  30 define( 'RE_IP_PREFIX', '(3[0-2]|[12]?\d)' );
  31 define( 'RE_IP_BLOCK', RE_IP_ADD . '\/' . RE_IP_PREFIX );
  32
  33 // An IPv6 address is made up of 8 words (each x0000 to xFFFF).
  34 // However, the "::" abbreviation can be used on consecutive x0000 words.
  35 define( 'RE_IPV6_WORD', '([0-9A-Fa-f]{1,4})' );
  36 define( 'RE_IPV6_PREFIX', '(12[0-8]|1[01][0-9]|[1-9]?\d)' );
  37 define( 'RE_IPV6_ADD',
  38         '(?:' . // starts with "::" (including "::")
  39                 ':(?::|(?::' . RE_IPV6_WORD . '){1,7})' .
  40         '|' . // ends with "::" (except "::")
  41                 RE_IPV6_WORD . '(?::' . RE_IPV6_WORD . '){0,6}::' .
  42         '|' . // contains one "::" in the middle (the ^ makes the test fail if none found)
  43                 RE_IPV6_WORD . '(?::((?(-1)|:))?' . RE_IPV6_WORD . '){1,6}(?(-2)|^)' .
  44         '|' . // contains no "::"
  45                 RE_IPV6_WORD . '(?::' . RE_IPV6_WORD . '){7}' .
  46         ')'
  47 );
  48 // An IPv6 block is an IP address and a prefix (d1 to d128)
  49 define( 'RE_IPV6_BLOCK', RE_IPV6_ADD . '\/' . RE_IPV6_PREFIX );
  50 // For IPv6 canonicalization (NOT for strict validation; these are quite lax!)
  51 define( 'RE_IPV6_GAP', ':(?:0+:)*(?::(?:0+:)*)?' );
  52 define( 'RE_IPV6_V4_PREFIX', '0*' . RE_IPV6_GAP . '(?:ffff:)?' );
  53
  54 // This might be useful for regexps used elsewhere, matches any IPv6 or IPv6 address or network
  55 define( 'IP_ADDRESS_STRING',
  56         '(?:' .
  57                 RE_IP_ADD . '(?:\/' . RE_IP_PREFIX . ')?' . // IPv4
  58         '|' .
  59                 RE_IPV6_ADD . '(?:\/' . RE_IPV6_PREFIX . ')?' . // IPv6
  60         ')'
  61 );
  62
  63 /**
  64  * A collection of public static functions to play with IP address
  65  * and IP blocks.
  66  */
  67 class IP {
  68         /** @var IPSet */
  69         private static $ipSet = null;
  70
  71         /**
  72          * Determine if a string is as valid IP address or network (CIDR prefix).
  73          * SIIT IPv4-translated addresses are rejected.
  74          * Note: canonicalize() tries to convert translated addresses to IPv4.
  75          *
  76          * @param string $ip possible IP address
  77          * @return Boolean
  78          */
  79         public static function isIPAddress( $ip ) {
  80                 return (bool)preg_match( '/^' . IP_ADDRESS_STRING . '$/', $ip );
  81         }
  82
  83         /**
  84          * Given a string, determine if it as valid IP in IPv6 only.
  85          * Note: Unlike isValid(), this looks for networks too.
  86          *
  87          * @param string $ip possible IP address
  88          * @return Boolean
  89          */
  90         public static function isIPv6( $ip ) {
  91                 return (bool)preg_match( '/^' . RE_IPV6_ADD . '(?:\/' . RE_IPV6_PREFIX . ')?$/', $ip );
  92         }
  93
  94         /**
  95          * Given a string, determine if it as valid IP in IPv4 only.
  96          * Note: Unlike isValid(), this looks for networks too.
  97          *
  98          * @param string $ip possible IP address
  99          * @return Boolean
 100          */
 101         public static function isIPv4( $ip ) {
 102                 return (bool)preg_match( '/^' . RE_IP_ADD . '(?:\/' . RE_IP_PREFIX . ')?$/', $ip );
 103         }
 104
 105         /**
 106          * Validate an IP address. Ranges are NOT considered valid.
 107          * SIIT IPv4-translated addresses are rejected.
 108          * Note: canonicalize() tries to convert translated addresses to IPv4.
 109          *
 110          * @param $ip String
 111          * @return Boolean: True if it is valid.
 112          */
 113         public static function isValid( $ip ) {
 114                 return ( preg_match( '/^' . RE_IP_ADD . '$/', $ip )
 115                         || preg_match( '/^' . RE_IPV6_ADD . '$/', $ip ) );
 116         }
 117
 118         /**
 119          * Validate an IP Block (valid address WITH a valid prefix).
 120          * SIIT IPv4-translated addresses are rejected.
 121          * Note: canonicalize() tries to convert translated addresses to IPv4.
 122          *
 123          * @param $ipblock String
 124          * @return Boolean: True if it is valid.
 125          */
 126         public static function isValidBlock( $ipblock ) {
 127                 return ( preg_match( '/^' . RE_IPV6_BLOCK . '$/', $ipblock )
 128                         || preg_match( '/^' . RE_IP_BLOCK . '$/', $ipblock ) );
 129         }
 130
 131         /**
 132          * Convert an IP into a verbose, uppercase, normalized form.
 133          * IPv6 addresses in octet notation are expanded to 8 words.
 134          * IPv4 addresses are just trimmed.
 135          *
 136          * @param string $ip IP address in quad or octet form (CIDR or not).
 137          * @return String
 138          */
 139         public static function sanitizeIP( $ip ) {
 140                 $ip = trim( $ip );
 141                 if ( $ip === '' ) {
 142                         return null;
 143                 }
 144                 if ( self::isIPv4( $ip ) || !self::isIPv6( $ip ) ) {
 145                         return $ip; // nothing else to do for IPv4 addresses or invalid ones
 146                 }
 147                 // Remove any whitespaces, convert to upper case
 148                 $ip = strtoupper( $ip );
 149                 // Expand zero abbreviations
 150                 $abbrevPos = strpos( $ip, '::' );
 151                 if ( $abbrevPos !== false ) {
 152                         // We know this is valid IPv6. Find the last index of the
 153                         // address before any CIDR number (e.g. "a:b:c::/24").
 154                         $CIDRStart = strpos( $ip, "/" );
 155                         $addressEnd = ( $CIDRStart !== false )
 156                                 ? $CIDRStart - 1
 157                                 : strlen( $ip ) - 1;
 158                         // If the '::' is at the beginning...
 159                         if ( $abbrevPos == 0 ) {
 160                                 $repeat = '0:';
 161                                 $extra = ( $ip == '::' ) ? '0' : ''; // for the address '::'
 162                                 $pad = 9; // 7+2 (due to '::')
 163                         // If the '::' is at the end...
 164                         } elseif ( $abbrevPos == ( $addressEnd - 1 ) ) {
 165                                 $repeat = ':0';
 166                                 $extra = '';
 167                                 $pad = 9; // 7+2 (due to '::')
 168                         // If the '::' is in the middle...
 169                         } else {
 170                                 $repeat = ':0';
 171                                 $extra = ':';
 172                                 $pad = 8; // 6+2 (due to '::')
 173                         }
 174                         $ip = str_replace( '::',
 175                                 str_repeat( $repeat, $pad - substr_count( $ip, ':' ) ) . $extra,
 176                                 $ip
 177                         );
 178                 }
 179                 // Remove leading zeros from each bloc as needed
 180                 $ip = preg_replace( '/(^|:)0+(' . RE_IPV6_WORD . ')/', '$1$2', $ip );
 181
 182                 return $ip;
 183         }
 184
 185         /**
 186          * Prettify an IP for display to end users.
 187          * This will make it more compact and lower-case.
 188          *
 189          * @param $ip string
 190          * @return string
 191          */
 192         public static function prettifyIP( $ip ) {
 193                 $ip = self::sanitizeIP( $ip ); // normalize (removes '::')
 194                 if ( self::isIPv6( $ip ) ) {
 195                         // Split IP into an address and a CIDR
 196                         if ( strpos( $ip, '/' ) !== false ) {
 197                                 list( $ip, $cidr ) = explode( '/', $ip, 2 );
 198                         } else {
 199                                 list( $ip, $cidr ) = array( $ip, '' );
 200                         }
 201                         // Get the largest slice of words with multiple zeros
 202                         $offset = 0;
 203                         $longest = $longestPos = false;
 204                         while ( preg_match(
 205                                 '!(?:^|:)0(?::0)+(?:$|:)!', $ip, $m, PREG_OFFSET_CAPTURE, $offset
 206                         ) ) {
 207                                 list( $match, $pos ) = $m[0]; // full match
 208                                 if ( strlen( $match ) > strlen( $longest ) ) {
 209                                         $longest = $match;
 210                                         $longestPos = $pos;
 211                                 }
 212                                 $offset = ( $pos + strlen( $match ) ); // advance
 213                         }
 214                         if ( $longest !== false ) {
 215                                 // Replace this portion of the string with the '::' abbreviation
 216                                 $ip = substr_replace( $ip, '::', $longestPos, strlen( $longest ) );
 217                         }
 218                         // Add any CIDR back on
 219                         if ( $cidr !== '' ) {
 220                                 $ip = "{$ip}/{$cidr}";
 221                         }
 222                         // Convert to lower case to make it more readable
 223                         $ip = strtolower( $ip );
 224                 }
 225
 226                 return $ip;
 227         }
 228
 229         /**
 230          * Given a host/port string, like one might find in the host part of a URL
 231          * per RFC 2732, split the hostname part and the port part and return an
 232          * array with an element for each. If there is no port part, the array will
 233          * have false in place of the port. If the string was invalid in some way,
 234          * false is returned.
 235          *
 236          * This was easy with IPv4 and was generally done in an ad-hoc way, but
 237          * with IPv6 it's somewhat more complicated due to the need to parse the
 238          * square brackets and colons.
 239          *
 240          * A bare IPv6 address is accepted despite the lack of square brackets.
 241          *
 242          * @param string $both The string with the host and port
 243          * @return array
 244          */
 245         public static function splitHostAndPort( $both ) {
 246                 if ( substr( $both, 0, 1 ) === '[' ) {
 247                         if ( preg_match( '/^\[(' . RE_IPV6_ADD . ')\](?::(?P<port>\d+))?$/', $both, $m ) ) {
 248                                 if ( isset( $m['port'] ) ) {
 249                                         return array( $m[1], intval( $m['port'] ) );
 250                                 } else {
 251                                         return array( $m[1], false );
 252                                 }
 253                         } else {
 254                                 // Square bracket found but no IPv6
 255                                 return false;
 256                         }
 257                 }
 258                 $numColons = substr_count( $both, ':' );
 259                 if ( $numColons >= 2 ) {
 260                         // Is it a bare IPv6 address?
 261                         if ( preg_match( '/^' . RE_IPV6_ADD . '$/', $both ) ) {
 262                                 return array( $both, false );
 263                         } else {
 264                                 // Not valid IPv6, but too many colons for anything else
 265                                 return false;
 266                         }
 267                 }
 268                 if ( $numColons >= 1 ) {
 269                         // Host:port?
 270                         $bits = explode( ':', $both );
 271                         if ( preg_match( '/^\d+/', $bits[1] ) ) {
 272                                 return array( $bits[0], intval( $bits[1] ) );
 273                         } else {
 274                                 // Not a valid port
 275                                 return false;
 276                         }
 277                 }
 278
 279                 // Plain hostname
 280                 return array( $both, false );
 281         }
 282
 283         /**
 284          * Given a host name and a port, combine them into host/port string like
 285          * you might find in a URL. If the host contains a colon, wrap it in square
 286          * brackets like in RFC 2732. If the port matches the default port, omit
 287          * the port specification
 288          *
 289          * @param $host string
 290          * @param $port int
 291          * @param $defaultPort bool|int
 292          * @return string
 293          */
 294         public static function combineHostAndPort( $host, $port, $defaultPort = false ) {
 295                 if ( strpos( $host, ':' ) !== false ) {
 296                         $host = "[$host]";
 297                 }
 298                 if ( $defaultPort !== false && $port == $defaultPort ) {
 299                         return $host;
 300                 } else {
 301                         return "$host:$port";
 302                 }
 303         }
 304
 305         /**
 306          * Convert an IPv4 or IPv6 hexadecimal representation back to readable format
 307          *
 308          * @param string $hex number, with "v6-" prefix if it is IPv6
 309          * @return String: quad-dotted (IPv4) or octet notation (IPv6)
 310          */
 311         public static function formatHex( $hex ) {
 312                 if ( substr( $hex, 0, 3 ) == 'v6-' ) { // IPv6
 313                         return self::hexToOctet( substr( $hex, 3 ) );
 314                 } else { // IPv4
 315                         return self::hexToQuad( $hex );
 316                 }
 317         }
 318
 319         /**
 320          * Converts a hexadecimal number to an IPv6 address in octet notation
 321          *
 322          * @param $ip_hex String: pure hex (no v6- prefix)
 323          * @return String (of format a:b:c:d:e:f:g:h)
 324          */
 325         public static function hexToOctet( $ip_hex ) {
 326                 // Pad hex to 32 chars (128 bits)
 327                 $ip_hex = str_pad( strtoupper( $ip_hex ), 32, '0', STR_PAD_LEFT );
 328                 // Separate into 8 words
 329                 $ip_oct = substr( $ip_hex, 0, 4 );
 330                 for ( $n = 1; $n < 8; $n++ ) {
 331                         $ip_oct .= ':' . substr( $ip_hex, 4 * $n, 4 );
 332                 }
 333                 // NO leading zeroes
 334                 $ip_oct = preg_replace( '/(^|:)0+(' . RE_IPV6_WORD . ')/', '$1$2', $ip_oct );
 335
 336                 return $ip_oct;
 337         }
 338
 339         /**
 340          * Converts a hexadecimal number to an IPv4 address in quad-dotted notation
 341          *
 342          * @param $ip_hex String: pure hex
 343          * @return String (of format a.b.c.d)
 344          */
 345         public static function hexToQuad( $ip_hex ) {
 346                 // Pad hex to 8 chars (32 bits)
 347                 $ip_hex = str_pad( strtoupper( $ip_hex ), 8, '0', STR_PAD_LEFT );
 348                 // Separate into four quads
 349                 $s = '';
 350                 for ( $i = 0; $i < 4; $i++ ) {
 351                         if ( $s !== '' ) {
 352                                 $s .= '.';
 353                         }
 354                         $s .= base_convert( substr( $ip_hex, $i * 2, 2 ), 16, 10 );
 355                 }
 356
 357                 return $s;
 358         }
 359
 360         /**
 361          * Determine if an IP address really is an IP address, and if it is public,
 362          * i.e. not RFC 1918 or similar
 363          *
 364          * @param $ip String
 365          * @return Boolean
 366          */
 367         public static function isPublic( $ip ) {
 368                 static $privateSet = null;
 369                 if ( !$privateSet ) {
 370                         $privateSet = new IPSet( array(
 371                                 '10.0.0.0/8', # RFC 1918 (private)
 372                                 '172.16.0.0/12', # RFC 1918 (private)
 373                                 '192.168.0.0/16', # RFC 1918 (private)
 374                                 '0.0.0.0/8', # this network
 375                                 '127.0.0.0/8', # loopback
 376                                 'fc00::/7', # RFC 4193 (local)
 377                                 '0:0:0:0:0:0:0:1', # loopback
 378                         ) );
 379                 }
 380                 return !$privateSet->match( $ip );
 381         }
 382
 383         /**
 384          * Return a zero-padded upper case hexadecimal representation of an IP address.
 385          *
 386          * Hexadecimal addresses are used because they can easily be extended to
 387          * IPv6 support. To separate the ranges, the return value from this
 388          * function for an IPv6 address will be prefixed with "v6-", a non-
 389          * hexadecimal string which sorts after the IPv4 addresses.
 390          *
 391          * @param string $ip quad dotted/octet IP address.
 392          * @return String|bool false on failure
 393          */
 394         public static function toHex( $ip ) {
 395                 if ( self::isIPv6( $ip ) ) {
 396                         $n = 'v6-' . self::IPv6ToRawHex( $ip );
 397                 } else {
 398                         $n = self::toUnsigned( $ip );
 399                         if ( $n !== false ) {
 400                                 $n = wfBaseConvert( $n, 10, 16, 8, false );
 401                         }
 402                 }
 403
 404                 return $n;
 405         }
 406
 407         /**
 408          * Given an IPv6 address in octet notation, returns a pure hex string.
 409          *
 410          * @param string $ip octet ipv6 IP address.
 411          * @return String|bool pure hex (uppercase); false on failure
 412          */
 413         private static function IPv6ToRawHex( $ip ) {
 414                 $ip = self::sanitizeIP( $ip );
 415                 if ( !$ip ) {
 416                         return false;
 417                 }
 418                 $r_ip = '';
 419                 foreach ( explode( ':', $ip ) as $v ) {
 420                         $r_ip .= str_pad( $v, 4, 0, STR_PAD_LEFT );
 421                 }
 422
 423                 return $r_ip;
 424         }
 425
 426         /**
 427          * Given an IP address in dotted-quad/octet notation, returns an unsigned integer.
 428          * Like ip2long() except that it actually works and has a consistent error return value.
 429          *
 430          * @param string $ip quad dotted IP address.
 431          * @return Mixed: string/int/false
 432          */
 433         public static function toUnsigned( $ip ) {
 434                 if ( self::isIPv6( $ip ) ) {
 435                         $n = wfBaseConvert( self::IPv6ToRawHex( $ip ), 16, 10 );
 436                 } else {
 437                         // Bug 60035: an IP with leading 0's fails in ip2long sometimes (e.g. *.08)
 438                         $ip = preg_replace( '/(?<=\.)0+(?=[1-9])/', '', $ip );
 439                         $n = ip2long( $ip );
 440                         if ( $n < 0 ) {
 441                                 $n += pow( 2, 32 );
 442                                 # On 32-bit platforms (and on Windows), 2^32 does not fit into an int,
 443                                 # so $n becomes a float. We convert it to string instead.
 444                                 if ( is_float( $n ) ) {
 445                                         $n = (string)$n;
 446                                 }
 447                         }
 448                 }
 449
 450                 return $n;
 451         }
 452
 453         /**
 454          * Convert a network specification in CIDR notation
 455          * to an integer network and a number of bits
 456          *
 457          * @param string $range IP with CIDR prefix
 458          * @return array(int or string, int)
 459          */
 460         public static function parseCIDR( $range ) {
 461                 if ( self::isIPv6( $range ) ) {
 462                         return self::parseCIDR6( $range );
 463                 }
 464                 $parts = explode( '/', $range, 2 );
 465                 if ( count( $parts ) != 2 ) {
 466                         return array( false, false );
 467                 }
 468                 list( $network, $bits ) = $parts;
 469                 $network = ip2long( $network );
 470                 if ( $network !== false && is_numeric( $bits ) && $bits >= 0 && $bits <= 32 ) {
 471                         if ( $bits == 0 ) {
 472                                 $network = 0;
 473                         } else {
 474                                 $network &= ~( ( 1 << ( 32 - $bits ) ) - 1 );
 475                         }
 476                         # Convert to unsigned
 477                         if ( $network < 0 ) {
 478                                 $network += pow( 2, 32 );
 479                         }
 480                 } else {
 481                         $network = false;
 482                         $bits = false;
 483                 }
 484
 485                 return array( $network, $bits );
 486         }
 487
 488         /**
 489          * Given a string range in a number of formats,
 490          * return the start and end of the range in hexadecimal.
 491          *
 492          * Formats are:
 493          *     1.2.3.4/24          CIDR
 494          *     1.2.3.4 - 1.2.3.5   Explicit range
 495          *     1.2.3.4             Single IP
 496          *
 497          *     2001:0db8:85a3::7344/96                       CIDR
 498          *     2001:0db8:85a3::7344 - 2001:0db8:85a3::7344   Explicit range
 499          *     2001:0db8:85a3::7344                          Single IP
 500          * @param string $range IP range
 501          * @return array(string, string)
 502          */
 503         public static function parseRange( $range ) {
 504                 // CIDR notation
 505                 if ( strpos( $range, '/' ) !== false ) {
 506                         if ( self::isIPv6( $range ) ) {
 507                                 return self::parseRange6( $range );
 508                         }
 509                         list( $network, $bits ) = self::parseCIDR( $range );
 510                         if ( $network === false ) {
 511                                 $start = $end = false;
 512                         } else {
 513                                 $start = sprintf( '%08X', $network );
 514                                 $end = sprintf( '%08X', $network + pow( 2, ( 32 - $bits ) ) - 1 );
 515                         }
 516                 // Explicit range
 517                 } elseif ( strpos( $range, '-' ) !== false ) {
 518                         list( $start, $end ) = array_map( 'trim', explode( '-', $range, 2 ) );
 519                         if ( self::isIPv6( $start ) && self::isIPv6( $end ) ) {
 520                                 return self::parseRange6( $range );
 521                         }
 522                         if ( self::isIPv4( $start ) && self::isIPv4( $end ) ) {
 523                                 $start = self::toHex( $start );
 524                                 $end = self::toHex( $end );
 525                                 if ( $start > $end ) {
 526                                         $start = $end = false;
 527                                 }
 528                         } else {
 529                                 $start = $end = false;
 530                         }
 531                 } else {
 532                         # Single IP
 533                         $start = $end = self::toHex( $range );
 534                 }
 535                 if ( $start === false || $end === false ) {
 536                         return array( false, false );
 537                 } else {
 538                         return array( $start, $end );
 539                 }
 540         }
 541
 542         /**
 543          * Convert a network specification in IPv6 CIDR notation to an
 544          * integer network and a number of bits
 545          *
 546          * @param $range
 547          *
 548          * @return array(string, int)
 549          */
 550         private static function parseCIDR6( $range ) {
 551                 # Explode into <expanded IP,range>
 552                 $parts = explode( '/', IP::sanitizeIP( $range ), 2 );
 553                 if ( count( $parts ) != 2 ) {
 554                         return array( false, false );
 555                 }
 556                 list( $network, $bits ) = $parts;
 557                 $network = self::IPv6ToRawHex( $network );
 558                 if ( $network !== false && is_numeric( $bits ) && $bits >= 0 && $bits <= 128 ) {
 559                         if ( $bits == 0 ) {
 560                                 $network = "0";
 561                         } else {
 562                                 # Native 32 bit functions WONT work here!!!
 563                                 # Convert to a padded binary number
 564                                 $network = wfBaseConvert( $network, 16, 2, 128 );
 565                                 # Truncate the last (128-$bits) bits and replace them with zeros
 566                                 $network = str_pad( substr( $network, 0, $bits ), 128, 0, STR_PAD_RIGHT );
 567                                 # Convert back to an integer
 568                                 $network = wfBaseConvert( $network, 2, 10 );
 569                         }
 570                 } else {
 571                         $network = false;
 572                         $bits = false;
 573                 }
 574
 575                 return array( $network, (int)$bits );
 576         }
 577
 578         /**
 579          * Given a string range in a number of formats, return the
 580          * start and end of the range in hexadecimal. For IPv6.
 581          *
 582          * Formats are:
 583          *     2001:0db8:85a3::7344/96                       CIDR
 584          *     2001:0db8:85a3::7344 - 2001:0db8:85a3::7344   Explicit range
 585          *     2001:0db8:85a3::7344/96                       Single IP
 586          *
 587          * @param $range
 588          *
 589          * @return array(string, string)
 590          */
 591         private static function parseRange6( $range ) {
 592                 # Expand any IPv6 IP
 593                 $range = IP::sanitizeIP( $range );
 594                 // CIDR notation...
 595                 if ( strpos( $range, '/' ) !== false ) {
 596                         list( $network, $bits ) = self::parseCIDR6( $range );
 597                         if ( $network === false ) {
 598                                 $start = $end = false;
 599                         } else {
 600                                 $start = wfBaseConvert( $network, 10, 16, 32, false );
 601                                 # Turn network to binary (again)
 602                                 $end = wfBaseConvert( $network, 10, 2, 128 );
 603                                 # Truncate the last (128-$bits) bits and replace them with ones
 604                                 $end = str_pad( substr( $end, 0, $bits ), 128, 1, STR_PAD_RIGHT );
 605                                 # Convert to hex
 606                                 $end = wfBaseConvert( $end, 2, 16, 32, false );
 607                                 # see toHex() comment
 608                                 $start = "v6-$start";
 609                                 $end = "v6-$end";
 610                         }
 611         // Explicit range notation...
 612                 } elseif ( strpos( $range, '-' ) !== false ) {
 613                         list( $start, $end ) = array_map( 'trim', explode( '-', $range, 2 ) );
 614                         $start = self::toHex( $start );
 615                         $end = self::toHex( $end );
 616                         if ( $start > $end ) {
 617                                 $start = $end = false;
 618                         }
 619                 } else {
 620                         # Single IP
 621                         $start = $end = self::toHex( $range );
 622                 }
 623                 if ( $start === false || $end === false ) {
 624                         return array( false, false );
 625                 } else {
 626                         return array( $start, $end );
 627                 }
 628         }
 629
 630         /**
 631          * Determine if a given IPv4/IPv6 address is in a given CIDR network
 632          *
 633          * @param string $addr the address to check against the given range.
 634          * @param string $range the range to check the given address against.
 635          * @return Boolean: whether or not the given address is in the given range.
 636          */
 637         public static function isInRange( $addr, $range ) {
 638                 $hexIP = self::toHex( $addr );
 639                 list( $start, $end ) = self::parseRange( $range );
 640
 641                 return ( strcmp( $hexIP, $start ) >= 0 &&
 642                         strcmp( $hexIP, $end ) <= 0 );
 643         }
 644
 645         /**
 646          * Convert some unusual representations of IPv4 addresses to their
 647          * canonical dotted quad representation.
 648          *
 649          * This currently only checks a few IPV4-to-IPv6 related cases.  More
 650          * unusual representations may be added later.
 651          *
 652          * @param string $addr something that might be an IP address
 653          * @return String: valid dotted quad IPv4 address or null
 654          */
 655         public static function canonicalize( $addr ) {
 656                 // remove zone info (bug 35738)
 657                 $addr = preg_replace( '/\%.*/', '', $addr );
 658
 659                 if ( self::isValid( $addr ) ) {
 660                         return $addr;
 661                 }
 662                 // Turn mapped addresses from ::ce:ffff:1.2.3.4 to 1.2.3.4
 663                 if ( strpos( $addr, ':' ) !== false && strpos( $addr, '.' ) !== false ) {
 664                         $addr = substr( $addr, strrpos( $addr, ':' ) + 1 );
 665                         if ( self::isIPv4( $addr ) ) {
 666                                 return $addr;
 667                         }
 668                 }
 669                 // IPv6 loopback address
 670                 $m = array();
 671                 if ( preg_match( '/^0*' . RE_IPV6_GAP . '1$/', $addr, $m ) ) {
 672                         return '127.0.0.1';
 673                 }
 674                 // IPv4-mapped and IPv4-compatible IPv6 addresses
 675                 if ( preg_match( '/^' . RE_IPV6_V4_PREFIX . '(' . RE_IP_ADD . ')$/i', $addr, $m ) ) {
 676                         return $m[1];
 677                 }
 678                 if ( preg_match( '/^' . RE_IPV6_V4_PREFIX . RE_IPV6_WORD .
 679                         ':' . RE_IPV6_WORD . '$/i', $addr, $m )
 680                 ) {
 681                         return long2ip( ( hexdec( $m[1] ) << 16 ) + hexdec( $m[2] ) );
 682                 }
 683
 684                 return null; // give up
 685         }
 686
 687         /**
 688          * Gets rid of unneeded numbers in quad-dotted/octet IP strings
 689          * For example, 127.111.113.151/24 -> 127.111.113.0/24
 690          * @param string $range IP address to normalize
 691          * @return string
 692          */
 693         public static function sanitizeRange( $range ) {
 694                 list( /*...*/, $bits ) = self::parseCIDR( $range );
 695                 list( $start, /*...*/ ) = self::parseRange( $range );
 696                 $start = self::formatHex( $start );
 697                 if ( $bits === false ) {
 698                         return $start; // wasn't actually a range
 699                 }
 700
 701                 return "$start/$bits";
 702         }
 703
 704         /**
 705          * Checks if an IP is a trusted proxy provider.
 706          * Useful to tell if X-Forwarded-For data is possibly bogus.
 707          * Squid cache servers for the site are whitelisted.
 708          * @since 1.24
 709          *
 710          * @param string $ip
 711          * @return bool
 712          */
 713         public static function isTrustedProxy( $ip ) {
 714                 $trusted = self::isConfiguredProxy( $ip );
 715                 wfRunHooks( 'IsTrustedProxy', array( &$ip, &$trusted ) );
 716                 return $trusted;
 717         }
 718
 719         /**
 720          * Checks if an IP matches a proxy we've configured
 721          * @since 1.24
 722          *
 723          * @param string $ip
 724          * @return bool
 725          */
 726         public static function isConfiguredProxy( $ip ) {
 727                 global $wgSquidServers, $wgSquidServersNoPurge;
 728
 729                 wfProfileIn( __METHOD__ );
 730                 // Quick check of known singular proxy servers
 731                 $trusted = in_array( $ip, $wgSquidServers );
 732
 733                 // Check against addresses and CIDR nets in the NoPurge list
 734                 if ( !$trusted ) {
 735                         if ( !self::$ipSet ) {
 736                                 self::$ipSet = new IPSet( $wgSquidServersNoPurge );
 737                         }
 738                         $trusted = self::$ipSet->match( $ip );
 739                 }
 740                 wfProfileOut( __METHOD__ );
 741
 742                 return $trusted;
 743         }
 744
 745         /**
 746          * Clears precomputed data used for proxy support.
 747          * Use this only for unit tests.
 748          */
 749         public static function clearCaches() {
 750                 self::$ipSet = null;
 751         }
 752 }