includes/libs/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>"
  22  */
  23
  24 use IPSet\IPSet;
  25
  26 // Some regex definition to "play" with IP address and IP address ranges
  27
  28 // An IPv4 address is made of 4 bytes from x00 to xFF which is d0 to d255
  29 define( 'RE_IP_BYTE', '(25[0-5]|2[0-4][0-9]|1[0-9][0-9]|0?[0-9]?[0-9])' );
  30 define( 'RE_IP_ADD', RE_IP_BYTE . '\.' . RE_IP_BYTE . '\.' . RE_IP_BYTE . '\.' . RE_IP_BYTE );
  31 // An IPv4 range is an IP address and a prefix (d1 to d32)
  32 define( 'RE_IP_PREFIX', '(3[0-2]|[12]?\d)' );
  33 define( 'RE_IP_RANGE', RE_IP_ADD . '\/' . RE_IP_PREFIX );
  34
  35 // An IPv6 address is made up of 8 words (each x0000 to xFFFF).
  36 // However, the "::" abbreviation can be used on consecutive x0000 words.
  37 define( 'RE_IPV6_WORD', '([0-9A-Fa-f]{1,4})' );
  38 define( 'RE_IPV6_PREFIX', '(12[0-8]|1[01][0-9]|[1-9]?\d)' );
  39 define( 'RE_IPV6_ADD',
  40         '(?:' . // starts with "::" (including "::")
  41                 ':(?::|(?::' . RE_IPV6_WORD . '){1,7})' .
  42         '|' . // ends with "::" (except "::")
  43                 RE_IPV6_WORD . '(?::' . RE_IPV6_WORD . '){0,6}::' .
  44         '|' . // contains one "::" in the middle (the ^ makes the test fail if none found)
  45                 RE_IPV6_WORD . '(?::((?(-1)|:))?' . RE_IPV6_WORD . '){1,6}(?(-2)|^)' .
  46         '|' . // contains no "::"
  47                 RE_IPV6_WORD . '(?::' . RE_IPV6_WORD . '){7}' .
  48         ')'
  49 );
  50 // An IPv6 range is an IP address and a prefix (d1 to d128)
  51 define( 'RE_IPV6_RANGE', RE_IPV6_ADD . '\/' . RE_IPV6_PREFIX );
  52 // For IPv6 canonicalization (NOT for strict validation; these are quite lax!)
  53 define( 'RE_IPV6_GAP', ':(?:0+:)*(?::(?:0+:)*)?' );
  54 define( 'RE_IPV6_V4_PREFIX', '0*' . RE_IPV6_GAP . '(?:ffff:)?' );
  55
  56 // This might be useful for regexps used elsewhere, matches any IPv4 or IPv6 address or network
  57 define( 'IP_ADDRESS_STRING',
  58         '(?:' .
  59                 RE_IP_ADD . '(?:\/' . RE_IP_PREFIX . ')?' . // IPv4
  60         '|' .
  61                 RE_IPV6_ADD . '(?:\/' . RE_IPV6_PREFIX . ')?' . // IPv6
  62         ')'
  63 );
  64
  65 /**
  66  * A collection of public static functions to play with IP address
  67  * and IP ranges.
  68  */
  69 class IP {
  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 bool
  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 bool
  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 bool
 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 string $ip
 111          * @return bool 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 range (valid address with a valid CIDR prefix).
 120          * SIIT IPv4-translated addresses are rejected.
 121          * @note canonicalize() tries to convert translated addresses to IPv4.
 122          *
 123          * @deprecated since 1.30. Use the equivalent IP::isValidRange().
 124          * @param string $ipRange
 125          * @return bool True if it is valid
 126          */
 127         public static function isValidBlock( $ipRange ) {
 128                 return self::isValidRange( $ipRange );
 129         }
 130
 131         /**
 132          * Validate an IP range (valid address with a valid CIDR prefix).
 133          * SIIT IPv4-translated addresses are rejected.
 134          * @note canonicalize() tries to convert translated addresses to IPv4.
 135          *
 136          * @param string $ipRange
 137          * @return bool True if it is valid
 138          * @since 1.30
 139          */
 140         public static function isValidRange( $ipRange ) {
 141                 return ( preg_match( '/^' . RE_IPV6_RANGE . '$/', $ipRange )
 142                         || preg_match( '/^' . RE_IP_RANGE . '$/', $ipRange ) );
 143         }
 144
 145         /**
 146          * Convert an IP into a verbose, uppercase, normalized form.
 147          * Both IPv4 and IPv6 addresses are trimmed. Additionally,
 148          * IPv6 addresses in octet notation are expanded to 8 words;
 149          * IPv4 addresses have leading zeros, in each octet, removed.
 150          *
 151          * @param string $ip IP address in quad or octet form (CIDR or not).
 152          * @return string
 153          */
 154         public static function sanitizeIP( $ip ) {
 155                 $ip = trim( $ip );
 156                 if ( $ip === '' ) {
 157                         return null;
 158                 }
 159                 /* If not an IP, just return trimmed value, since sanitizeIP() is called
 160                  * in a number of contexts where usernames are supplied as input.
 161                  */
 162                 if ( !self::isIPAddress( $ip ) ) {
 163                         return $ip;
 164                 }
 165                 if ( self::isIPv4( $ip ) ) {
 166                         // Remove leading 0's from octet representation of IPv4 address
 167                         $ip = preg_replace( '/(?:^|(?<=\.))0+(?=[1-9]|0\.|0$)/', '', $ip );
 168                         return $ip;
 169                 }
 170                 // Remove any whitespaces, convert to upper case
 171                 $ip = strtoupper( $ip );
 172                 // Expand zero abbreviations
 173                 $abbrevPos = strpos( $ip, '::' );
 174                 if ( $abbrevPos !== false ) {
 175                         // We know this is valid IPv6. Find the last index of the
 176                         // address before any CIDR number (e.g. "a:b:c::/24").
 177                         $CIDRStart = strpos( $ip, "/" );
 178                         $addressEnd = ( $CIDRStart !== false )
 179                                 ? $CIDRStart - 1
 180                                 : strlen( $ip ) - 1;
 181                         // If the '::' is at the beginning...
 182                         if ( $abbrevPos == 0 ) {
 183                                 $repeat = '0:';
 184                                 $extra = ( $ip == '::' ) ? '0' : ''; // for the address '::'
 185                                 $pad = 9; // 7+2 (due to '::')
 186                         // If the '::' is at the end...
 187                         } elseif ( $abbrevPos == ( $addressEnd - 1 ) ) {
 188                                 $repeat = ':0';
 189                                 $extra = '';
 190                                 $pad = 9; // 7+2 (due to '::')
 191                         // If the '::' is in the middle...
 192                         } else {
 193                                 $repeat = ':0';
 194                                 $extra = ':';
 195                                 $pad = 8; // 6+2 (due to '::')
 196                         }
 197                         $ip = str_replace( '::',
 198                                 str_repeat( $repeat, $pad - substr_count( $ip, ':' ) ) . $extra,
 199                                 $ip
 200                         );
 201                 }
 202                 // Remove leading zeros from each bloc as needed
 203                 $ip = preg_replace( '/(^|:)0+(' . RE_IPV6_WORD . ')/', '$1$2', $ip );
 204
 205                 return $ip;
 206         }
 207
 208         /**
 209          * Prettify an IP for display to end users.
 210          * This will make it more compact and lower-case.
 211          *
 212          * @param string $ip
 213          * @return string
 214          */
 215         public static function prettifyIP( $ip ) {
 216                 $ip = self::sanitizeIP( $ip ); // normalize (removes '::')
 217                 if ( self::isIPv6( $ip ) ) {
 218                         // Split IP into an address and a CIDR
 219                         if ( strpos( $ip, '/' ) !== false ) {
 220                                 list( $ip, $cidr ) = explode( '/', $ip, 2 );
 221                         } else {
 222                                 list( $ip, $cidr ) = [ $ip, '' ];
 223                         }
 224                         // Get the largest slice of words with multiple zeros
 225                         $offset = 0;
 226                         $longest = $longestPos = false;
 227                         while ( preg_match(
 228                                 '!(?:^|:)0(?::0)+(?:$|:)!', $ip, $m, PREG_OFFSET_CAPTURE, $offset
 229                         ) ) {
 230                                 list( $match, $pos ) = $m[0]; // full match
 231                                 if ( strlen( $match ) > strlen( $longest ) ) {
 232                                         $longest = $match;
 233                                         $longestPos = $pos;
 234                                 }
 235                                 $offset = ( $pos + strlen( $match ) ); // advance
 236                         }
 237                         if ( $longest !== false ) {
 238                                 // Replace this portion of the string with the '::' abbreviation
 239                                 $ip = substr_replace( $ip, '::', $longestPos, strlen( $longest ) );
 240                         }
 241                         // Add any CIDR back on
 242                         if ( $cidr !== '' ) {
 243                                 $ip = "{$ip}/{$cidr}";
 244                         }
 245                         // Convert to lower case to make it more readable
 246                         $ip = strtolower( $ip );
 247                 }
 248
 249                 return $ip;
 250         }
 251
 252         /**
 253          * Given a host/port string, like one might find in the host part of a URL
 254          * per RFC 2732, split the hostname part and the port part and return an
 255          * array with an element for each. If there is no port part, the array will
 256          * have false in place of the port. If the string was invalid in some way,
 257          * false is returned.
 258          *
 259          * This was easy with IPv4 and was generally done in an ad-hoc way, but
 260          * with IPv6 it's somewhat more complicated due to the need to parse the
 261          * square brackets and colons.
 262          *
 263          * A bare IPv6 address is accepted despite the lack of square brackets.
 264          *
 265          * @param string $both The string with the host and port
 266          * @return array|false Array normally, false on certain failures
 267          */
 268         public static function splitHostAndPort( $both ) {
 269                 if ( substr( $both, 0, 1 ) === '[' ) {
 270                         if ( preg_match( '/^\[(' . RE_IPV6_ADD . ')\](?::(?P<port>\d+))?$/', $both, $m ) ) {
 271                                 if ( isset( $m['port'] ) ) {
 272                                         return [ $m[1], intval( $m['port'] ) ];
 273                                 } else {
 274                                         return [ $m[1], false ];
 275                                 }
 276                         } else {
 277                                 // Square bracket found but no IPv6
 278                                 return false;
 279                         }
 280                 }
 281                 $numColons = substr_count( $both, ':' );
 282                 if ( $numColons >= 2 ) {
 283                         // Is it a bare IPv6 address?
 284                         if ( preg_match( '/^' . RE_IPV6_ADD . '$/', $both ) ) {
 285                                 return [ $both, false ];
 286                         } else {
 287                                 // Not valid IPv6, but too many colons for anything else
 288                                 return false;
 289                         }
 290                 }
 291                 if ( $numColons >= 1 ) {
 292                         // Host:port?
 293                         $bits = explode( ':', $both );
 294                         if ( preg_match( '/^\d+/', $bits[1] ) ) {
 295                                 return [ $bits[0], intval( $bits[1] ) ];
 296                         } else {
 297                                 // Not a valid port
 298                                 return false;
 299                         }
 300                 }
 301
 302                 // Plain hostname
 303                 return [ $both, false ];
 304         }
 305
 306         /**
 307          * Given a host name and a port, combine them into host/port string like
 308          * you might find in a URL. If the host contains a colon, wrap it in square
 309          * brackets like in RFC 2732. If the port matches the default port, omit
 310          * the port specification
 311          *
 312          * @param string $host
 313          * @param int $port
 314          * @param bool|int $defaultPort
 315          * @return string
 316          */
 317         public static function combineHostAndPort( $host, $port, $defaultPort = false ) {
 318                 if ( strpos( $host, ':' ) !== false ) {
 319                         $host = "[$host]";
 320                 }
 321                 if ( $defaultPort !== false && $port == $defaultPort ) {
 322                         return $host;
 323                 } else {
 324                         return "$host:$port";
 325                 }
 326         }
 327
 328         /**
 329          * Convert an IPv4 or IPv6 hexadecimal representation back to readable format
 330          *
 331          * @param string $hex Number, with "v6-" prefix if it is IPv6
 332          * @return string Quad-dotted (IPv4) or octet notation (IPv6)
 333          */
 334         public static function formatHex( $hex ) {
 335                 if ( substr( $hex, 0, 3 ) == 'v6-' ) { // IPv6
 336                         return self::hexToOctet( substr( $hex, 3 ) );
 337                 } else { // IPv4
 338                         return self::hexToQuad( $hex );
 339                 }
 340         }
 341
 342         /**
 343          * Converts a hexadecimal number to an IPv6 address in octet notation
 344          *
 345          * @param string $ip_hex Pure hex (no v6- prefix)
 346          * @return string (of format a:b:c:d:e:f:g:h)
 347          */
 348         public static function hexToOctet( $ip_hex ) {
 349                 // Pad hex to 32 chars (128 bits)
 350                 $ip_hex = str_pad( strtoupper( $ip_hex ), 32, '0', STR_PAD_LEFT );
 351                 // Separate into 8 words
 352                 $ip_oct = substr( $ip_hex, 0, 4 );
 353                 for ( $n = 1; $n < 8; $n++ ) {
 354                         $ip_oct .= ':' . substr( $ip_hex, 4 * $n, 4 );
 355                 }
 356                 // NO leading zeroes
 357                 $ip_oct = preg_replace( '/(^|:)0+(' . RE_IPV6_WORD . ')/', '$1$2', $ip_oct );
 358
 359                 return $ip_oct;
 360         }
 361
 362         /**
 363          * Converts a hexadecimal number to an IPv4 address in quad-dotted notation
 364          *
 365          * @param string $ip_hex Pure hex
 366          * @return string (of format a.b.c.d)
 367          */
 368         public static function hexToQuad( $ip_hex ) {
 369                 // Pad hex to 8 chars (32 bits)
 370                 $ip_hex = str_pad( strtoupper( $ip_hex ), 8, '0', STR_PAD_LEFT );
 371                 // Separate into four quads
 372                 $s = '';
 373                 for ( $i = 0; $i < 4; $i++ ) {
 374                         if ( $s !== '' ) {
 375                                 $s .= '.';
 376                         }
 377                         $s .= base_convert( substr( $ip_hex, $i * 2, 2 ), 16, 10 );
 378                 }
 379
 380                 return $s;
 381         }
 382
 383         /**
 384          * Determine if an IP address really is an IP address, and if it is public,
 385          * i.e. not RFC 1918 or similar
 386          *
 387          * @param string $ip
 388          * @return bool
 389          */
 390         public static function isPublic( $ip ) {
 391                 static $privateSet = null;
 392                 if ( !$privateSet ) {
 393                         $privateSet = new IPSet( [
 394                                 '10.0.0.0/8', # RFC 1918 (private)
 395                                 '172.16.0.0/12', # RFC 1918 (private)
 396                                 '192.168.0.0/16', # RFC 1918 (private)
 397                                 '0.0.0.0/8', # this network
 398                                 '127.0.0.0/8', # loopback
 399                                 'fc00::/7', # RFC 4193 (local)
 400                                 '0:0:0:0:0:0:0:1', # loopback
 401                                 '169.254.0.0/16', # link-local
 402                                 'fe80::/10', # link-local
 403                         ] );
 404                 }
 405                 return !$privateSet->match( $ip );
 406         }
 407
 408         /**
 409          * Return a zero-padded upper case hexadecimal representation of an IP address.
 410          *
 411          * Hexadecimal addresses are used because they can easily be extended to
 412          * IPv6 support. To separate the ranges, the return value from this
 413          * function for an IPv6 address will be prefixed with "v6-", a non-
 414          * hexadecimal string which sorts after the IPv4 addresses.
 415          *
 416          * @param string $ip Quad dotted/octet IP address.
 417          * @return string|bool False on failure
 418          */
 419         public static function toHex( $ip ) {
 420                 if ( self::isIPv6( $ip ) ) {
 421                         $n = 'v6-' . self::IPv6ToRawHex( $ip );
 422                 } elseif ( self::isIPv4( $ip ) ) {
 423                         // T62035/T97897: An IP with leading 0's fails in ip2long sometimes (e.g. *.08),
 424                         // also double/triple 0 needs to be changed to just a single 0 for ip2long.
 425                         $ip = self::sanitizeIP( $ip );
 426                         $n = ip2long( $ip );
 427                         if ( $n < 0 ) {
 428                                 $n += pow( 2, 32 );
 429                                 # On 32-bit platforms (and on Windows), 2^32 does not fit into an int,
 430                                 # so $n becomes a float. We convert it to string instead.
 431                                 if ( is_float( $n ) ) {
 432                                         $n = (string)$n;
 433                                 }
 434                         }
 435                         if ( $n !== false ) {
 436                                 # Floating points can handle the conversion; faster than Wikimedia\base_convert()
 437                                 $n = strtoupper( str_pad( base_convert( $n, 10, 16 ), 8, '0', STR_PAD_LEFT ) );
 438                         }
 439                 } else {
 440                         $n = false;
 441                 }
 442
 443                 return $n;
 444         }
 445
 446         /**
 447          * Given an IPv6 address in octet notation, returns a pure hex string.
 448          *
 449          * @param string $ip Octet ipv6 IP address.
 450          * @return string|bool Pure hex (uppercase); false on failure
 451          */
 452         private static function IPv6ToRawHex( $ip ) {
 453                 $ip = self::sanitizeIP( $ip );
 454                 if ( !$ip ) {
 455                         return false;
 456                 }
 457                 $r_ip = '';
 458                 foreach ( explode( ':', $ip ) as $v ) {
 459                         $r_ip .= str_pad( $v, 4, 0, STR_PAD_LEFT );
 460                 }
 461
 462                 return $r_ip;
 463         }
 464
 465         /**
 466          * Convert a network specification in CIDR notation
 467          * to an integer network and a number of bits
 468          *
 469          * @param string $range IP with CIDR prefix
 470          * @return array(int or string, int)
 471          */
 472         public static function parseCIDR( $range ) {
 473                 if ( self::isIPv6( $range ) ) {
 474                         return self::parseCIDR6( $range );
 475                 }
 476                 $parts = explode( '/', $range, 2 );
 477                 if ( count( $parts ) != 2 ) {
 478                         return [ false, false ];
 479                 }
 480                 list( $network, $bits ) = $parts;
 481                 $network = ip2long( $network );
 482                 if ( $network !== false && is_numeric( $bits ) && $bits >= 0 && $bits <= 32 ) {
 483                         if ( $bits == 0 ) {
 484                                 $network = 0;
 485                         } else {
 486                                 $network &= ~( ( 1 << ( 32 - $bits ) ) - 1 );
 487                         }
 488                         # Convert to unsigned
 489                         if ( $network < 0 ) {
 490                                 $network += pow( 2, 32 );
 491                         }
 492                 } else {
 493                         $network = false;
 494                         $bits = false;
 495                 }
 496
 497                 return [ $network, $bits ];
 498         }
 499
 500         /**
 501          * Given a string range in a number of formats,
 502          * return the start and end of the range in hexadecimal.
 503          *
 504          * Formats are:
 505          *     1.2.3.4/24          CIDR
 506          *     1.2.3.4 - 1.2.3.5   Explicit range
 507          *     1.2.3.4             Single IP
 508          *
 509          *     2001:0db8:85a3::7344/96                       CIDR
 510          *     2001:0db8:85a3::7344 - 2001:0db8:85a3::7344   Explicit range
 511          *     2001:0db8:85a3::7344                          Single IP
 512          * @param string $range IP range
 513          * @return array [ string, string ]
 514          */
 515         public static function parseRange( $range ) {
 516                 // CIDR notation
 517                 if ( strpos( $range, '/' ) !== false ) {
 518                         if ( self::isIPv6( $range ) ) {
 519                                 return self::parseRange6( $range );
 520                         }
 521                         list( $network, $bits ) = self::parseCIDR( $range );
 522                         if ( $network === false ) {
 523                                 $start = $end = false;
 524                         } else {
 525                                 $start = sprintf( '%08X', $network );
 526                                 $end = sprintf( '%08X', $network + pow( 2, ( 32 - $bits ) ) - 1 );
 527                         }
 528                 // Explicit range
 529                 } elseif ( strpos( $range, '-' ) !== false ) {
 530                         list( $start, $end ) = array_map( 'trim', explode( '-', $range, 2 ) );
 531                         if ( self::isIPv6( $start ) && self::isIPv6( $end ) ) {
 532                                 return self::parseRange6( $range );
 533                         }
 534                         if ( self::isIPv4( $start ) && self::isIPv4( $end ) ) {
 535                                 $start = self::toHex( $start );
 536                                 $end = self::toHex( $end );
 537                                 if ( $start > $end ) {
 538                                         $start = $end = false;
 539                                 }
 540                         } else {
 541                                 $start = $end = false;
 542                         }
 543                 } else {
 544                         # Single IP
 545                         $start = $end = self::toHex( $range );
 546                 }
 547                 if ( $start === false || $end === false ) {
 548                         return [ false, false ];
 549                 } else {
 550                         return [ $start, $end ];
 551                 }
 552         }
 553
 554         /**
 555          * Convert a network specification in IPv6 CIDR notation to an
 556          * integer network and a number of bits
 557          *
 558          * @param string $range
 559          *
 560          * @return array(string, int)
 561          */
 562         private static function parseCIDR6( $range ) {
 563                 # Explode into <expanded IP,range>
 564                 $parts = explode( '/', self::sanitizeIP( $range ), 2 );
 565                 if ( count( $parts ) != 2 ) {
 566                         return [ false, false ];
 567                 }
 568                 list( $network, $bits ) = $parts;
 569                 $network = self::IPv6ToRawHex( $network );
 570                 if ( $network !== false && is_numeric( $bits ) && $bits >= 0 && $bits <= 128 ) {
 571                         if ( $bits == 0 ) {
 572                                 $network = "0";
 573                         } else {
 574                                 # Native 32 bit functions WONT work here!!!
 575                                 # Convert to a padded binary number
 576                                 $network = Wikimedia\base_convert( $network, 16, 2, 128 );
 577                                 # Truncate the last (128-$bits) bits and replace them with zeros
 578                                 $network = str_pad( substr( $network, 0, $bits ), 128, 0, STR_PAD_RIGHT );
 579                                 # Convert back to an integer
 580                                 $network = Wikimedia\base_convert( $network, 2, 10 );
 581                         }
 582                 } else {
 583                         $network = false;
 584                         $bits = false;
 585                 }
 586
 587                 return [ $network, (int)$bits ];
 588         }
 589
 590         /**
 591          * Given a string range in a number of formats, return the
 592          * start and end of the range in hexadecimal. For IPv6.
 593          *
 594          * Formats are:
 595          *     2001:0db8:85a3::7344/96                       CIDR
 596          *     2001:0db8:85a3::7344 - 2001:0db8:85a3::7344   Explicit range
 597          *     2001:0db8:85a3::7344/96                       Single IP
 598          *
 599          * @param string $range
 600          *
 601          * @return array(string, string)
 602          */
 603         private static function parseRange6( $range ) {
 604                 # Expand any IPv6 IP
 605                 $range = self::sanitizeIP( $range );
 606                 // CIDR notation...
 607                 if ( strpos( $range, '/' ) !== false ) {
 608                         list( $network, $bits ) = self::parseCIDR6( $range );
 609                         if ( $network === false ) {
 610                                 $start = $end = false;
 611                         } else {
 612                                 $start = Wikimedia\base_convert( $network, 10, 16, 32, false );
 613                                 # Turn network to binary (again)
 614                                 $end = Wikimedia\base_convert( $network, 10, 2, 128 );
 615                                 # Truncate the last (128-$bits) bits and replace them with ones
 616                                 $end = str_pad( substr( $end, 0, $bits ), 128, 1, STR_PAD_RIGHT );
 617                                 # Convert to hex
 618                                 $end = Wikimedia\base_convert( $end, 2, 16, 32, false );
 619                                 # see toHex() comment
 620                                 $start = "v6-$start";
 621                                 $end = "v6-$end";
 622                         }
 623                 // Explicit range notation...
 624                 } elseif ( strpos( $range, '-' ) !== false ) {
 625                         list( $start, $end ) = array_map( 'trim', explode( '-', $range, 2 ) );
 626                         $start = self::toHex( $start );
 627                         $end = self::toHex( $end );
 628                         if ( $start > $end ) {
 629                                 $start = $end = false;
 630                         }
 631                 } else {
 632                         # Single IP
 633                         $start = $end = self::toHex( $range );
 634                 }
 635                 if ( $start === false || $end === false ) {
 636                         return [ false, false ];
 637                 } else {
 638                         return [ $start, $end ];
 639                 }
 640         }
 641
 642         /**
 643          * Determine if a given IPv4/IPv6 address is in a given CIDR network
 644          *
 645          * @param string $addr The address to check against the given range.
 646          * @param string $range The range to check the given address against.
 647          * @return bool Whether or not the given address is in the given range.
 648          *
 649          * @note This can return unexpected results for invalid arguments!
 650          *       Make sure you pass a valid IP address and IP range.
 651          */
 652         public static function isInRange( $addr, $range ) {
 653                 $hexIP = self::toHex( $addr );
 654                 list( $start, $end ) = self::parseRange( $range );
 655
 656                 return ( strcmp( $hexIP, $start ) >= 0 &&
 657                         strcmp( $hexIP, $end ) <= 0 );
 658         }
 659
 660         /**
 661          * Determines if an IP address is a list of CIDR a.b.c.d/n ranges.
 662          *
 663          * @since 1.25
 664          *
 665          * @param string $ip the IP to check
 666          * @param array $ranges the IP ranges, each element a range
 667          *
 668          * @return bool true if the specified adress belongs to the specified range; otherwise, false.
 669          */
 670         public static function isInRanges( $ip, $ranges ) {
 671                 foreach ( $ranges as $range ) {
 672                         if ( self::isInRange( $ip, $range ) ) {
 673                                 return true;
 674                         }
 675                 }
 676                 return false;
 677         }
 678
 679         /**
 680          * Convert some unusual representations of IPv4 addresses to their
 681          * canonical dotted quad representation.
 682          *
 683          * This currently only checks a few IPV4-to-IPv6 related cases.  More
 684          * unusual representations may be added later.
 685          *
 686          * @param string $addr Something that might be an IP address
 687          * @return string|null Valid dotted quad IPv4 address or null
 688          */
 689         public static function canonicalize( $addr ) {
 690                 // remove zone info (T37738)
 691                 $addr = preg_replace( '/\%.*/', '', $addr );
 692
 693                 if ( self::isValid( $addr ) ) {
 694                         return $addr;
 695                 }
 696                 // Turn mapped addresses from ::ce:ffff:1.2.3.4 to 1.2.3.4
 697                 if ( strpos( $addr, ':' ) !== false && strpos( $addr, '.' ) !== false ) {
 698                         $addr = substr( $addr, strrpos( $addr, ':' ) + 1 );
 699                         if ( self::isIPv4( $addr ) ) {
 700                                 return $addr;
 701                         }
 702                 }
 703                 // IPv6 loopback address
 704                 $m = [];
 705                 if ( preg_match( '/^0*' . RE_IPV6_GAP . '1$/', $addr, $m ) ) {
 706                         return '127.0.0.1';
 707                 }
 708                 // IPv4-mapped and IPv4-compatible IPv6 addresses
 709                 if ( preg_match( '/^' . RE_IPV6_V4_PREFIX . '(' . RE_IP_ADD . ')$/i', $addr, $m ) ) {
 710                         return $m[1];
 711                 }
 712                 if ( preg_match( '/^' . RE_IPV6_V4_PREFIX . RE_IPV6_WORD .
 713                         ':' . RE_IPV6_WORD . '$/i', $addr, $m )
 714                 ) {
 715                         return long2ip( ( hexdec( $m[1] ) << 16 ) + hexdec( $m[2] ) );
 716                 }
 717
 718                 return null; // give up
 719         }
 720
 721         /**
 722          * Gets rid of unneeded numbers in quad-dotted/octet IP strings
 723          * For example, 127.111.113.151/24 -> 127.111.113.0/24
 724          * @param string $range IP address to normalize
 725          * @return string
 726          */
 727         public static function sanitizeRange( $range ) {
 728                 list( /*...*/, $bits ) = self::parseCIDR( $range );
 729                 list( $start, /*...*/ ) = self::parseRange( $range );
 730                 $start = self::formatHex( $start );
 731                 if ( $bits === false ) {
 732                         return $start; // wasn't actually a range
 733                 }
 734
 735                 return "$start/$bits";
 736         }
 737
 738         /**
 739          * Returns the subnet of a given IP
 740          *
 741          * @param string $ip
 742          * @return string|false
 743          */
 744         public static function getSubnet( $ip ) {
 745                 $matches = [];
 746                 $subnet = false;
 747                 if ( self::isIPv6( $ip ) ) {
 748                         $parts = self::parseRange( "$ip/64" );
 749                         $subnet = $parts[0];
 750                 } elseif ( preg_match( '/^(\d+\.\d+\.\d+)\.\d+$/', $ip, $matches ) ) {
 751                         // IPv4
 752                         $subnet = $matches[1];
 753                 }
 754                 return $subnet;
 755         }
 756 }