includes/IP.php

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