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 blocks
  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 block is an IP address and a prefix (d1 to d32)
  32 define( 'RE_IP_PREFIX', '(3[0-2]|[12]?\d)' );
  33 define( 'RE_IP_BLOCK', 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 block is an IP address and a prefix (d1 to d128)
  51 define( 'RE_IPV6_BLOCK', 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 blocks.
  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 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 string $ipblock
 124          * @return bool 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          * Both IPv4 and IPv6 addresses are trimmed. Additionally,
 134          * IPv6 addresses in octet notation are expanded to 8 words;
 135          * IPv4 addresses have leading zeros, in each octet, removed.
 136          *
 137          * @param string $ip IP address in quad or octet form (CIDR or not).
 138          * @return string
 139          */
 140         public static function sanitizeIP( $ip ) {
 141                 $ip = trim( $ip );
 142                 if ( $ip === '' ) {
 143                         return null;
 144                 }
 145                 /* If not an IP, just return trimmed value, since sanitizeIP() is called
 146                  * in a number of contexts where usernames are supplied as input.
 147                  */
 148                 if ( !self::isIPAddress( $ip ) ) {
 149                         return $ip;
 150                 }
 151                 if ( self::isIPv4( $ip ) ) {
 152                         // Remove leading 0's from octet representation of IPv4 address
 153                         $ip = preg_replace( '/(?:^|(?<=\.))0+(?=[1-9]|0\.|0$)/', '', $ip );
 154                         return $ip;
 155                 }
 156                 // Remove any whitespaces, convert to upper case
 157                 $ip = strtoupper( $ip );
 158                 // Expand zero abbreviations
 159                 $abbrevPos = strpos( $ip, '::' );
 160                 if ( $abbrevPos !== false ) {
 161                         // We know this is valid IPv6. Find the last index of the
 162                         // address before any CIDR number (e.g. "a:b:c::/24").
 163                         $CIDRStart = strpos( $ip, "/" );
 164                         $addressEnd = ( $CIDRStart !== false )
 165                                 ? $CIDRStart - 1
 166                                 : strlen( $ip ) - 1;
 167                         // If the '::' is at the beginning...
 168                         if ( $abbrevPos == 0 ) {
 169                                 $repeat = '0:';
 170                                 $extra = ( $ip == '::' ) ? '0' : ''; // for the address '::'
 171                                 $pad = 9; // 7+2 (due to '::')
 172                         // If the '::' is at the end...
 173                         } elseif ( $abbrevPos == ( $addressEnd - 1 ) ) {
 174                                 $repeat = ':0';
 175                                 $extra = '';
 176                                 $pad = 9; // 7+2 (due to '::')
 177                         // If the '::' is in the middle...
 178                         } else {
 179                                 $repeat = ':0';
 180                                 $extra = ':';
 181                                 $pad = 8; // 6+2 (due to '::')
 182                         }
 183                         $ip = str_replace( '::',
 184                                 str_repeat( $repeat, $pad - substr_count( $ip, ':' ) ) . $extra,
 185                                 $ip
 186                         );
 187                 }
 188                 // Remove leading zeros from each bloc as needed
 189                 $ip = preg_replace( '/(^|:)0+(' . RE_IPV6_WORD . ')/', '$1$2', $ip );
 190
 191                 return $ip;
 192         }
 193
 194         /**
 195          * Prettify an IP for display to end users.
 196          * This will make it more compact and lower-case.
 197          *
 198          * @param string $ip
 199          * @return string
 200          */
 201         public static function prettifyIP( $ip ) {
 202                 $ip = self::sanitizeIP( $ip ); // normalize (removes '::')
 203                 if ( self::isIPv6( $ip ) ) {
 204                         // Split IP into an address and a CIDR
 205                         if ( strpos( $ip, '/' ) !== false ) {
 206                                 list( $ip, $cidr ) = explode( '/', $ip, 2 );
 207                         } else {
 208                                 list( $ip, $cidr ) = [ $ip, '' ];
 209                         }
 210                         // Get the largest slice of words with multiple zeros
 211                         $offset = 0;
 212                         $longest = $longestPos = false;
 213                         while ( preg_match(
 214                                 '!(?:^|:)0(?::0)+(?:$|:)!', $ip, $m, PREG_OFFSET_CAPTURE, $offset
 215                         ) ) {
 216                                 list( $match, $pos ) = $m[0]; // full match
 217                                 if ( strlen( $match ) > strlen( $longest ) ) {
 218                                         $longest = $match;
 219                                         $longestPos = $pos;
 220                                 }
 221                                 $offset = ( $pos + strlen( $match ) ); // advance
 222                         }
 223                         if ( $longest !== false ) {
 224                                 // Replace this portion of the string with the '::' abbreviation
 225                                 $ip = substr_replace( $ip, '::', $longestPos, strlen( $longest ) );
 226                         }
 227                         // Add any CIDR back on
 228                         if ( $cidr !== '' ) {
 229                                 $ip = "{$ip}/{$cidr}";
 230                         }
 231                         // Convert to lower case to make it more readable
 232                         $ip = strtolower( $ip );
 233                 }
 234
 235                 return $ip;
 236         }
 237
 238         /**
 239          * Given a host/port string, like one might find in the host part of a URL
 240          * per RFC 2732, split the hostname part and the port part and return an
 241          * array with an element for each. If there is no port part, the array will
 242          * have false in place of the port. If the string was invalid in some way,
 243          * false is returned.
 244          *
 245          * This was easy with IPv4 and was generally done in an ad-hoc way, but
 246          * with IPv6 it's somewhat more complicated due to the need to parse the
 247          * square brackets and colons.
 248          *
 249          * A bare IPv6 address is accepted despite the lack of square brackets.
 250          *
 251          * @param string $both The string with the host and port
 252          * @return array|false Array normally, false on certain failures
 253          */
 254         public static function splitHostAndPort( $both ) {
 255                 if ( substr( $both, 0, 1 ) === '[' ) {
 256                         if ( preg_match( '/^\[(' . RE_IPV6_ADD . ')\](?::(?P<port>\d+))?$/', $both, $m ) ) {
 257                                 if ( isset( $m['port'] ) ) {
 258                                         return [ $m[1], intval( $m['port'] ) ];
 259                                 } else {
 260                                         return [ $m[1], false ];
 261                                 }
 262                         } else {
 263                                 // Square bracket found but no IPv6
 264                                 return false;
 265                         }
 266                 }
 267                 $numColons = substr_count( $both, ':' );
 268                 if ( $numColons >= 2 ) {
 269                         // Is it a bare IPv6 address?
 270                         if ( preg_match( '/^' . RE_IPV6_ADD . '$/', $both ) ) {
 271                                 return [ $both, false ];
 272                         } else {
 273                                 // Not valid IPv6, but too many colons for anything else
 274                                 return false;
 275                         }
 276                 }
 277                 if ( $numColons >= 1 ) {
 278                         // Host:port?
 279                         $bits = explode( ':', $both );
 280                         if ( preg_match( '/^\d+/', $bits[1] ) ) {
 281                                 return [ $bits[0], intval( $bits[1] ) ];
 282                         } else {
 283                                 // Not a valid port
 284                                 return false;
 285                         }
 286                 }
 287
 288                 // Plain hostname
 289                 return [ $both, false ];
 290         }
 291
 292         /**
 293          * Given a host name and a port, combine them into host/port string like
 294          * you might find in a URL. If the host contains a colon, wrap it in square
 295          * brackets like in RFC 2732. If the port matches the default port, omit
 296          * the port specification
 297          *
 298          * @param string $host
 299          * @param int $port
 300          * @param bool|int $defaultPort
 301          * @return string
 302          */
 303         public static function combineHostAndPort( $host, $port, $defaultPort = false ) {
 304                 if ( strpos( $host, ':' ) !== false ) {
 305                         $host = "[$host]";
 306                 }
 307                 if ( $defaultPort !== false && $port == $defaultPort ) {
 308                         return $host;
 309                 } else {
 310                         return "$host:$port";
 311                 }
 312         }
 313
 314         /**
 315          * Convert an IPv4 or IPv6 hexadecimal representation back to readable format
 316          *
 317          * @param string $hex Number, with "v6-" prefix if it is IPv6
 318          * @return string Quad-dotted (IPv4) or octet notation (IPv6)
 319          */
 320         public static function formatHex( $hex ) {
 321                 if ( substr( $hex, 0, 3 ) == 'v6-' ) { // IPv6
 322                         return self::hexToOctet( substr( $hex, 3 ) );
 323                 } else { // IPv4
 324                         return self::hexToQuad( $hex );
 325                 }
 326         }
 327
 328         /**
 329          * Converts a hexadecimal number to an IPv6 address in octet notation
 330          *
 331          * @param string $ip_hex Pure hex (no v6- prefix)
 332          * @return string (of format a:b:c:d:e:f:g:h)
 333          */
 334         public static function hexToOctet( $ip_hex ) {
 335                 // Pad hex to 32 chars (128 bits)
 336                 $ip_hex = str_pad( strtoupper( $ip_hex ), 32, '0', STR_PAD_LEFT );
 337                 // Separate into 8 words
 338                 $ip_oct = substr( $ip_hex, 0, 4 );
 339                 for ( $n = 1; $n < 8; $n++ ) {
 340                         $ip_oct .= ':' . substr( $ip_hex, 4 * $n, 4 );
 341                 }
 342                 // NO leading zeroes
 343                 $ip_oct = preg_replace( '/(^|:)0+(' . RE_IPV6_WORD . ')/', '$1$2', $ip_oct );
 344
 345                 return $ip_oct;
 346         }
 347
 348         /**
 349          * Converts a hexadecimal number to an IPv4 address in quad-dotted notation
 350          *
 351          * @param string $ip_hex Pure hex
 352          * @return string (of format a.b.c.d)
 353          */
 354         public static function hexToQuad( $ip_hex ) {
 355                 // Pad hex to 8 chars (32 bits)
 356                 $ip_hex = str_pad( strtoupper( $ip_hex ), 8, '0', STR_PAD_LEFT );
 357                 // Separate into four quads
 358                 $s = '';
 359                 for ( $i = 0; $i < 4; $i++ ) {
 360                         if ( $s !== '' ) {
 361                                 $s .= '.';
 362                         }
 363                         $s .= base_convert( substr( $ip_hex, $i * 2, 2 ), 16, 10 );
 364                 }
 365
 366                 return $s;
 367         }
 368
 369         /**
 370          * Determine if an IP address really is an IP address, and if it is public,
 371          * i.e. not RFC 1918 or similar
 372          *
 373          * @param string $ip
 374          * @return bool
 375          */
 376         public static function isPublic( $ip ) {
 377                 static $privateSet = null;
 378                 if ( !$privateSet ) {
 379                         $privateSet = new IPSet( [
 380                                 '10.0.0.0/8', # RFC 1918 (private)
 381                                 '172.16.0.0/12', # RFC 1918 (private)
 382                                 '192.168.0.0/16', # RFC 1918 (private)
 383                                 '0.0.0.0/8', # this network
 384                                 '127.0.0.0/8', # loopback
 385                                 'fc00::/7', # RFC 4193 (local)
 386                                 '0:0:0:0:0:0:0:1', # loopback
 387                                 '169.254.0.0/16', # link-local
 388                                 'fe80::/10', # link-local
 389                         ] );
 390                 }
 391                 return !$privateSet->match( $ip );
 392         }
 393
 394         /**
 395          * Return a zero-padded upper case hexadecimal representation of an IP address.
 396          *
 397          * Hexadecimal addresses are used because they can easily be extended to
 398          * IPv6 support. To separate the ranges, the return value from this
 399          * function for an IPv6 address will be prefixed with "v6-", a non-
 400          * hexadecimal string which sorts after the IPv4 addresses.
 401          *
 402          * @param string $ip Quad dotted/octet IP address.
 403          * @return string|bool False on failure
 404          */
 405         public static function toHex( $ip ) {
 406                 if ( self::isIPv6( $ip ) ) {
 407                         $n = 'v6-' . self::IPv6ToRawHex( $ip );
 408                 } elseif ( self::isIPv4( $ip ) ) {
 409                         // T62035/T97897: An IP with leading 0's fails in ip2long sometimes (e.g. *.08),
 410                         // also double/triple 0 needs to be changed to just a single 0 for ip2long.
 411                         $ip = self::sanitizeIP( $ip );
 412                         $n = ip2long( $ip );
 413                         if ( $n < 0 ) {
 414                                 $n += pow( 2, 32 );
 415                                 # On 32-bit platforms (and on Windows), 2^32 does not fit into an int,
 416                                 # so $n becomes a float. We convert it to string instead.
 417                                 if ( is_float( $n ) ) {
 418                                         $n = (string)$n;
 419                                 }
 420                         }
 421                         if ( $n !== false ) {
 422                                 # Floating points can handle the conversion; faster than Wikimedia\base_convert()
 423                                 $n = strtoupper( str_pad( base_convert( $n, 10, 16 ), 8, '0', STR_PAD_LEFT ) );
 424                         }
 425                 } else {
 426                         $n = false;
 427                 }
 428
 429                 return $n;
 430         }
 431
 432         /**
 433          * Given an IPv6 address in octet notation, returns a pure hex string.
 434          *
 435          * @param string $ip Octet ipv6 IP address.
 436          * @return string|bool Pure hex (uppercase); false on failure
 437          */
 438         private static function IPv6ToRawHex( $ip ) {
 439                 $ip = self::sanitizeIP( $ip );
 440                 if ( !$ip ) {
 441                         return false;
 442                 }
 443                 $r_ip = '';
 444                 foreach ( explode( ':', $ip ) as $v ) {
 445                         $r_ip .= str_pad( $v, 4, 0, STR_PAD_LEFT );
 446                 }
 447
 448                 return $r_ip;
 449         }
 450
 451         /**
 452          * Convert a network specification in CIDR notation
 453          * to an integer network and a number of bits
 454          *
 455          * @param string $range IP with CIDR prefix
 456          * @return array(int or string, int)
 457          */
 458         public static function parseCIDR( $range ) {
 459                 if ( self::isIPv6( $range ) ) {
 460                         return self::parseCIDR6( $range );
 461                 }
 462                 $parts = explode( '/', $range, 2 );
 463                 if ( count( $parts ) != 2 ) {
 464                         return [ false, false ];
 465                 }
 466                 list( $network, $bits ) = $parts;
 467                 $network = ip2long( $network );
 468                 if ( $network !== false && is_numeric( $bits ) && $bits >= 0 && $bits <= 32 ) {
 469                         if ( $bits == 0 ) {
 470                                 $network = 0;
 471                         } else {
 472                                 $network &= ~( ( 1 << ( 32 - $bits ) ) - 1 );
 473                         }
 474                         # Convert to unsigned
 475                         if ( $network < 0 ) {
 476                                 $network += pow( 2, 32 );
 477                         }
 478                 } else {
 479                         $network = false;
 480                         $bits = false;
 481                 }
 482
 483                 return [ $network, $bits ];
 484         }
 485
 486         /**
 487          * Given a string range in a number of formats,
 488          * return the start and end of the range in hexadecimal.
 489          *
 490          * Formats are:
 491          *     1.2.3.4/24          CIDR
 492          *     1.2.3.4 - 1.2.3.5   Explicit range
 493          *     1.2.3.4             Single IP
 494          *
 495          *     2001:0db8:85a3::7344/96                       CIDR
 496          *     2001:0db8:85a3::7344 - 2001:0db8:85a3::7344   Explicit range
 497          *     2001:0db8:85a3::7344                          Single IP
 498          * @param string $range IP range
 499          * @return array(string, string)
 500          */
 501         public static function parseRange( $range ) {
 502                 // CIDR notation
 503                 if ( strpos( $range, '/' ) !== false ) {
 504                         if ( self::isIPv6( $range ) ) {
 505                                 return self::parseRange6( $range );
 506                         }
 507                         list( $network, $bits ) = self::parseCIDR( $range );
 508                         if ( $network === false ) {
 509                                 $start = $end = false;
 510                         } else {
 511                                 $start = sprintf( '%08X', $network );
 512                                 $end = sprintf( '%08X', $network + pow( 2, ( 32 - $bits ) ) - 1 );
 513                         }
 514                 // Explicit range
 515                 } elseif ( strpos( $range, '-' ) !== false ) {
 516                         list( $start, $end ) = array_map( 'trim', explode( '-', $range, 2 ) );
 517                         if ( self::isIPv6( $start ) && self::isIPv6( $end ) ) {
 518                                 return self::parseRange6( $range );
 519                         }
 520                         if ( self::isIPv4( $start ) && self::isIPv4( $end ) ) {
 521                                 $start = self::toHex( $start );
 522                                 $end = self::toHex( $end );
 523                                 if ( $start > $end ) {
 524                                         $start = $end = false;
 525                                 }
 526                         } else {
 527                                 $start = $end = false;
 528                         }
 529                 } else {
 530                         # Single IP
 531                         $start = $end = self::toHex( $range );
 532                 }
 533                 if ( $start === false || $end === false ) {
 534                         return [ false, false ];
 535                 } else {
 536                         return [ $start, $end ];
 537                 }
 538         }
 539
 540         /**
 541          * Convert a network specification in IPv6 CIDR notation to an
 542          * integer network and a number of bits
 543          *
 544          * @param string $range
 545          *
 546          * @return array(string, int)
 547          */
 548         private static function parseCIDR6( $range ) {
 549                 # Explode into <expanded IP,range>
 550                 $parts = explode( '/', self::sanitizeIP( $range ), 2 );
 551                 if ( count( $parts ) != 2 ) {
 552                         return [ false, false ];
 553                 }
 554                 list( $network, $bits ) = $parts;
 555                 $network = self::IPv6ToRawHex( $network );
 556                 if ( $network !== false && is_numeric( $bits ) && $bits >= 0 && $bits <= 128 ) {
 557                         if ( $bits == 0 ) {
 558                                 $network = "0";
 559                         } else {
 560                                 # Native 32 bit functions WONT work here!!!
 561                                 # Convert to a padded binary number
 562                                 $network = Wikimedia\base_convert( $network, 16, 2, 128 );
 563                                 # Truncate the last (128-$bits) bits and replace them with zeros
 564                                 $network = str_pad( substr( $network, 0, $bits ), 128, 0, STR_PAD_RIGHT );
 565                                 # Convert back to an integer
 566                                 $network = Wikimedia\base_convert( $network, 2, 10 );
 567                         }
 568                 } else {
 569                         $network = false;
 570                         $bits = false;
 571                 }
 572
 573                 return [ $network, (int)$bits ];
 574         }
 575
 576         /**
 577          * Given a string range in a number of formats, return the
 578          * start and end of the range in hexadecimal. For IPv6.
 579          *
 580          * Formats are:
 581          *     2001:0db8:85a3::7344/96                       CIDR
 582          *     2001:0db8:85a3::7344 - 2001:0db8:85a3::7344   Explicit range
 583          *     2001:0db8:85a3::7344/96                       Single IP
 584          *
 585          * @param string $range
 586          *
 587          * @return array(string, string)
 588          */
 589         private static function parseRange6( $range ) {
 590                 # Expand any IPv6 IP
 591                 $range = self::sanitizeIP( $range );
 592                 // CIDR notation...
 593                 if ( strpos( $range, '/' ) !== false ) {
 594                         list( $network, $bits ) = self::parseCIDR6( $range );
 595                         if ( $network === false ) {
 596                                 $start = $end = false;
 597                         } else {
 598                                 $start = Wikimedia\base_convert( $network, 10, 16, 32, false );
 599                                 # Turn network to binary (again)
 600                                 $end = Wikimedia\base_convert( $network, 10, 2, 128 );
 601                                 # Truncate the last (128-$bits) bits and replace them with ones
 602                                 $end = str_pad( substr( $end, 0, $bits ), 128, 1, STR_PAD_RIGHT );
 603                                 # Convert to hex
 604                                 $end = Wikimedia\base_convert( $end, 2, 16, 32, false );
 605                                 # see toHex() comment
 606                                 $start = "v6-$start";
 607                                 $end = "v6-$end";
 608                         }
 609                 // Explicit range notation...
 610                 } elseif ( strpos( $range, '-' ) !== false ) {
 611                         list( $start, $end ) = array_map( 'trim', explode( '-', $range, 2 ) );
 612                         $start = self::toHex( $start );
 613                         $end = self::toHex( $end );
 614                         if ( $start > $end ) {
 615                                 $start = $end = false;
 616                         }
 617                 } else {
 618                         # Single IP
 619                         $start = $end = self::toHex( $range );
 620                 }
 621                 if ( $start === false || $end === false ) {
 622                         return [ false, false ];
 623                 } else {
 624                         return [ $start, $end ];
 625                 }
 626         }
 627
 628         /**
 629          * Determine if a given IPv4/IPv6 address is in a given CIDR network
 630          *
 631          * @param string $addr The address to check against the given range.
 632          * @param string $range The range to check the given address against.
 633          * @return bool Whether or not the given address is in the given range.
 634          *
 635          * @note This can return unexpected results for invalid arguments!
 636          *       Make sure you pass a valid IP address and IP range.
 637          */
 638         public static function isInRange( $addr, $range ) {
 639                 $hexIP = self::toHex( $addr );
 640                 list( $start, $end ) = self::parseRange( $range );
 641
 642                 return ( strcmp( $hexIP, $start ) >= 0 &&
 643                         strcmp( $hexIP, $end ) <= 0 );
 644         }
 645
 646         /**
 647          * Determines if an IP address is a list of CIDR a.b.c.d/n ranges.
 648          *
 649          * @since 1.25
 650          *
 651          * @param string $ip the IP to check
 652          * @param array $ranges the IP ranges, each element a range
 653          *
 654          * @return bool true if the specified adress belongs to the specified range; otherwise, false.
 655          */
 656         public static function isInRanges( $ip, $ranges ) {
 657                 foreach ( $ranges as $range ) {
 658                         if ( self::isInRange( $ip, $range ) ) {
 659                                 return true;
 660                         }
 661                 }
 662                 return false;
 663         }
 664
 665         /**
 666          * Convert some unusual representations of IPv4 addresses to their
 667          * canonical dotted quad representation.
 668          *
 669          * This currently only checks a few IPV4-to-IPv6 related cases.  More
 670          * unusual representations may be added later.
 671          *
 672          * @param string $addr Something that might be an IP address
 673          * @return string|null Valid dotted quad IPv4 address or null
 674          */
 675         public static function canonicalize( $addr ) {
 676                 // remove zone info (T37738)
 677                 $addr = preg_replace( '/\%.*/', '', $addr );
 678
 679                 if ( self::isValid( $addr ) ) {
 680                         return $addr;
 681                 }
 682                 // Turn mapped addresses from ::ce:ffff:1.2.3.4 to 1.2.3.4
 683                 if ( strpos( $addr, ':' ) !== false && strpos( $addr, '.' ) !== false ) {
 684                         $addr = substr( $addr, strrpos( $addr, ':' ) + 1 );
 685                         if ( self::isIPv4( $addr ) ) {
 686                                 return $addr;
 687                         }
 688                 }
 689                 // IPv6 loopback address
 690                 $m = [];
 691                 if ( preg_match( '/^0*' . RE_IPV6_GAP . '1$/', $addr, $m ) ) {
 692                         return '127.0.0.1';
 693                 }
 694                 // IPv4-mapped and IPv4-compatible IPv6 addresses
 695                 if ( preg_match( '/^' . RE_IPV6_V4_PREFIX . '(' . RE_IP_ADD . ')$/i', $addr, $m ) ) {
 696                         return $m[1];
 697                 }
 698                 if ( preg_match( '/^' . RE_IPV6_V4_PREFIX . RE_IPV6_WORD .
 699                         ':' . RE_IPV6_WORD . '$/i', $addr, $m )
 700                 ) {
 701                         return long2ip( ( hexdec( $m[1] ) << 16 ) + hexdec( $m[2] ) );
 702                 }
 703
 704                 return null; // give up
 705         }
 706
 707         /**
 708          * Gets rid of unneeded numbers in quad-dotted/octet IP strings
 709          * For example, 127.111.113.151/24 -> 127.111.113.0/24
 710          * @param string $range IP address to normalize
 711          * @return string
 712          */
 713         public static function sanitizeRange( $range ) {
 714                 list( /*...*/, $bits ) = self::parseCIDR( $range );
 715                 list( $start, /*...*/ ) = self::parseRange( $range );
 716                 $start = self::formatHex( $start );
 717                 if ( $bits === false ) {
 718                         return $start; // wasn't actually a range
 719                 }
 720
 721                 return "$start/$bits";
 722         }
 723
 724         /**
 725          * Returns the subnet of a given IP
 726          *
 727          * @param string $ip
 728          * @return string|false
 729          */
 730         public static function getSubnet( $ip ) {
 731                 $matches = [];
 732                 $subnet = false;
 733                 if ( self::isIPv6( $ip ) ) {
 734                         $parts = self::parseRange( "$ip/64" );
 735                         $subnet = $parts[0];
 736                 } elseif ( preg_match( '/^(\d+\.\d+\.\d+)\.\d+$/', $ip, $matches ) ) {
 737                         // IPv4
 738                         $subnet = $matches[1];
 739                 }
 740                 return $subnet;
 741         }
 742 }