includes/IP.php

   1 <?php
   2 /*
   3  * @Author "Ashar Voultoiz" <hashar@altern.org>
   4  * @License GPL v2 or later
   5  */
   6
   7 // Some regex definition to "play" with IP address and IP address blocks
   8
   9 // An IP is made of 4 bytes from x00 to xFF which is d0 to d255
  10 define( 'RE_IP_BYTE', '(25[0-5]|2[0-4][0-9]|1[0-9][0-9]|0?[0-9]?[0-9])');
  11 define( 'RE_IP_ADD' , RE_IP_BYTE . '\.' . RE_IP_BYTE . '\.' . RE_IP_BYTE . '\.' . RE_IP_BYTE );
  12 // An IPv4 block is an IP address and a prefix (d1 to d32)
  13 define( 'RE_IP_PREFIX', '(3[0-2]|[12]?\d)');
  14 define( 'RE_IP_BLOCK', RE_IP_ADD . '\/' . RE_IP_PREFIX);
  15 // For IPv6 canonicalization (NOT for strict validation; these are quite lax!)
  16 define( 'RE_IPV6_WORD', '([0-9A-Fa-f]{1,4})' );
  17 define( 'RE_IPV6_GAP', ':(?:0+:)*(?::(?:0+:)*)?' );
  18 define( 'RE_IPV6_V4_PREFIX', '0*' . RE_IPV6_GAP . '(?:ffff:)?' );
  19 // An IPv6 block is an IP address and a prefix (d1 to d128)
  20 define( 'RE_IPV6_PREFIX', '(12[0-8]|1[01][0-9]|[1-9]?\d)');
  21 // An IPv6 IP is made up of 8 octets. However abbreviations like "::" can be used.
  22 // This is lax! Number of octets/double colons validation not done.
  23 define( 'RE_IPV6_ADD',
  24         '(' .
  25                 ':(:' . RE_IPV6_WORD . '){1,7}' . // IPs that start with ":"
  26         '|' .
  27                 RE_IPV6_WORD . '(:{1,2}' . RE_IPV6_WORD . '|::$){1,7}' . // IPs that don't start with ":"
  28         ')'
  29 );
  30 define( 'RE_IPV6_BLOCK', RE_IPV6_ADD . '\/' . RE_IPV6_PREFIX );
  31 // This might be useful for regexps used elsewhere, matches any IPv6 or IPv6 address or network
  32 define( 'IP_ADDRESS_STRING',
  33         '(?:' .
  34                 RE_IP_ADD . '(\/' . RE_IP_PREFIX . '|)' . // IPv4
  35         '|' .
  36                 RE_IPV6_ADD . '(\/' . RE_IPV6_PREFIX . '|)' . // IPv6
  37         ')'
  38 );
  39
  40 /**
  41  * A collection of public static functions to play with IP address
  42  * and IP blocks.
  43  */
  44 class IP {
  45         /**
  46          * Given a string, determine if it as valid IP
  47          * Unlike isValid(), this looks for networks too
  48          * @param $ip IP address.
  49          * @return string
  50          */
  51         public static function isIPAddress( $ip ) {
  52                 if ( !$ip ) return false;
  53                 if ( is_array( $ip ) ) {
  54                   throw new MWException( "invalid value passed to " . __METHOD__ );
  55                 }
  56                 // IPv6 IPs with two "::" strings are ambiguous and thus invalid
  57                 return preg_match( '/^' . IP_ADDRESS_STRING . '$/', $ip) && ( substr_count($ip, '::') < 2 );
  58         }
  59
  60         public static function isIPv6( $ip ) {
  61                 if ( !$ip ) return false;
  62                 if( is_array( $ip ) ) {
  63                         throw new MWException( "invalid value passed to " . __METHOD__ );
  64                 }
  65                 $doubleColons = substr_count($ip, '::');
  66                 // IPv6 IPs with two "::" strings are ambiguous and thus invalid
  67                 return preg_match( '/^' . RE_IPV6_ADD . '(\/' . RE_IPV6_PREFIX . '|)$/', $ip)
  68                         && ( $doubleColons == 1 || substr_count($ip,':') == 7 );
  69         }
  70
  71         public static function isIPv4( $ip ) {
  72                 if ( !$ip ) return false;
  73                 return preg_match( '/^' . RE_IP_ADD . '(\/' . RE_IP_PREFIX . '|)$/', $ip);
  74         }
  75
  76         /**
  77          * Given an IP address in dotted-quad notation, returns an IPv6 octet.
  78          * See http://www.answers.com/topic/ipv4-compatible-address
  79          * IPs with the first 92 bits as zeros are reserved from IPv6
  80          * @param $ip quad-dotted IP address.
  81          * @return string
  82          */
  83         public static function IPv4toIPv6( $ip ) {
  84                 if ( !$ip ) return null;
  85                 // Convert only if needed
  86                 if ( self::isIPv6( $ip ) ) return $ip;
  87                 // IPv4 CIDRs
  88                 if ( strpos( $ip, '/' ) !== false ) {
  89                         $parts = explode( '/', $ip, 2 );
  90                         if ( count( $parts ) != 2 ) {
  91                                 return false;
  92                         }
  93                         $network = self::toUnsigned( $parts[0] );
  94                         if ( $network !== false && is_numeric( $parts[1] ) && $parts[1] >= 0 && $parts[1] <= 32 ) {
  95                                 $bits = $parts[1] + 96;
  96                                 return self::toOctet( $network ) . "/$bits";
  97                         } else {
  98                                 return false;
  99                         }
 100                 }
 101                 return self::toOctet( self::toUnsigned( $ip ) );
 102         }
 103
 104         /**
 105          * Given an IPv6 address in octet notation, returns an unsigned integer.
 106          * @param $ip octet ipv6 IP address.
 107          * @return string
 108          */
 109         public static function toUnsigned6( $ip ) {
 110                 if ( !$ip ) return null;
 111         $ip = explode(':', self::sanitizeIP( $ip ) );
 112         $r_ip = '';
 113         foreach ($ip as $v) {
 114                 $r_ip .= str_pad( $v, 4, 0, STR_PAD_LEFT );
 115         }
 116         $r_ip = wfBaseConvert( $r_ip, 16, 10 );
 117         return $r_ip;
 118         }
 119
 120         /**
 121          * Given an IPv6 address in octet notation, returns the expanded octet.
 122          * IPv4 IPs will be trimmed, thats it...
 123          * @param $ip octet ipv6 IP address.
 124          * @return string
 125          */
 126         public static function sanitizeIP( $ip ) {
 127                 $ip = trim( $ip );
 128                 if ( $ip === '' ) return null;
 129                 // Trim and return IPv4 addresses
 130                 if ( self::isIPv4($ip) ) return $ip;
 131                 // Only IPv6 addresses can be expanded
 132                 if ( !self::isIPv6($ip) ) return $ip;
 133                 // Remove any whitespaces, convert to upper case
 134                 $ip = strtoupper( $ip );
 135                 // Expand zero abbreviations
 136                 if ( strpos( $ip, '::' ) !== false ) {
 137                 $ip = str_replace('::', str_repeat(':0', 8 - substr_count($ip, ':')) . ':', $ip);
 138         }
 139         // For IPs that start with "::", correct the final IP so that it starts with '0' and not ':'
 140         if ( $ip[0] == ':' ) $ip = "0$ip";
 141         // Remove leading zereos from each bloc as needed
 142         $ip = preg_replace( '/(^|:)0+' . RE_IPV6_WORD . '/', '$1$2', $ip );
 143         return $ip;
 144         }
 145
 146         /**
 147          * Given an unsigned integer, returns an IPv6 address in octet notation
 148          * @param $ip integer IP address.
 149          * @return string
 150          */
 151         public static function toOctet( $ip_int ) {
 152                 // Convert to padded uppercase hex
 153                 $ip_hex = wfBaseConvert($ip_int, 10, 16, 32, false);
 154                 // Separate into 8 octets
 155                 $ip_oct = substr( $ip_hex, 0, 4 );
 156                 for ($n=1; $n < 8; $n++) {
 157                         $ip_oct .= ':' . substr($ip_hex, 4*$n, 4);
 158                 }
 159                 // NO leading zeroes
 160                 $ip_oct = preg_replace( '/(^|:)0+' . RE_IPV6_WORD . '/', '$1$2', $ip_oct );
 161         return $ip_oct;
 162         }
 163
 164         /**
 165          * Convert an IPv4 or IPv6 hexadecimal representation back to readable format
 166          */
 167         public static function formatHex( $hex ) {
 168                 if ( substr( $hex, 0, 3 ) == 'v6-' ) {
 169                         return self::hexToOctet( $hex );
 170                 } else {
 171                         return self::hexToQuad( $hex );
 172                 }
 173         }
 174
 175         /**
 176          * Given a hexadecimal number, returns to an IPv6 address in octet notation
 177          * @param $ip string hex IP
 178          * @return string
 179          */
 180         public static function hextoOctet( $ip_hex ) {
 181                 // Convert to padded uppercase hex
 182                 $ip_hex = str_pad( strtoupper($ip_hex), 32, '0');
 183                 // Separate into 8 octets
 184                 $ip_oct = substr( $ip_hex, 0, 4 );
 185                 for ($n=1; $n < 8; $n++) {
 186                         $ip_oct .= ':' . substr($ip_hex, 4*$n, 4);
 187                 }
 188                 // NO leading zeroes
 189                 $ip_oct = preg_replace( '/(^|:)0+' . RE_IPV6_WORD . '/', '$1$2', $ip_oct );
 190         return $ip_oct;
 191         }
 192
 193         /**
 194          * Converts a hexadecimal number to an IPv4 address in octet notation
 195          * @param $ip string Hex IP
 196          * @return string
 197          */
 198         public static function hexToQuad( $ip ) {
 199                 // Converts a hexadecimal IP to nnn.nnn.nnn.nnn format
 200                 $s = '';
 201                 for ( $i = 0; $i < 4; $i++ ) {
 202                         if ( $s !== '' ) {
 203                                 $s .= '.';
 204                         }
 205                         $s .= base_convert( substr( $ip, $i * 2, 2 ), 16, 10 );
 206                 }
 207                 return $s;
 208         }
 209
 210         /**
 211          * Convert a network specification in IPv6 CIDR notation to an integer network and a number of bits
 212          * @return array(string, int)
 213          */
 214         public static function parseCIDR6( $range ) {
 215                 # Expand any IPv6 IP
 216                 $parts = explode( '/', IP::sanitizeIP( $range ), 2 );
 217                 if ( count( $parts ) != 2 ) {
 218                         return array( false, false );
 219                 }
 220                 $network = self::toUnsigned6( $parts[0] );
 221                 if ( $network !== false && is_numeric( $parts[1] ) && $parts[1] >= 0 && $parts[1] <= 128 ) {
 222                         $bits = $parts[1];
 223                         if ( $bits == 0 ) {
 224                                 $network = 0;
 225                         } else {
 226                         # Native 32 bit functions WONT work here!!!
 227                         # Convert to a padded binary number
 228                                 $network = wfBaseConvert( $network, 10, 2, 128 );
 229                         # Truncate the last (128-$bits) bits and replace them with zeros
 230                                 $network = str_pad( substr( $network, 0, $bits ), 128, 0, STR_PAD_RIGHT );
 231                         # Convert back to an integer
 232                                 $network = wfBaseConvert( $network, 2, 10 );
 233                         }
 234                 } else {
 235                         $network = false;
 236                         $bits = false;
 237                 }
 238                 return array( $network, $bits );
 239         }
 240
 241         /**
 242          * Given a string range in a number of formats, return the start and end of
 243          * the range in hexadecimal. For IPv6.
 244          *
 245          * Formats are:
 246          *     2001:0db8:85a3::7344/96                                   CIDR
 247          *     2001:0db8:85a3::7344 - 2001:0db8:85a3::7344   Explicit range
 248          *     2001:0db8:85a3::7344/96                                   Single IP
 249          * @return array(string, int)
 250          */
 251         public static function parseRange6( $range ) {
 252                 # Expand any IPv6 IP
 253                 $range = IP::sanitizeIP( $range );
 254                 if ( strpos( $range, '/' ) !== false ) {
 255                         # CIDR
 256                         list( $network, $bits ) = self::parseCIDR6( $range );
 257                         if ( $network === false ) {
 258                                 $start = $end = false;
 259                         } else {
 260                                 $start = wfBaseConvert( $network, 10, 16, 32, false );
 261                                 # Turn network to binary (again)
 262                                 $end = wfBaseConvert( $network, 10, 2, 128 );
 263                                 # Truncate the last (128-$bits) bits and replace them with ones
 264                                 $end = str_pad( substr( $end, 0, $bits ), 128, 1, STR_PAD_RIGHT );
 265                                 # Convert to hex
 266                                 $end = wfBaseConvert( $end, 2, 16, 32, false );
 267                                 # see toHex() comment
 268                                 $start = "v6-$start"; $end = "v6-$end";
 269                         }
 270                 } elseif ( strpos( $range, '-' ) !== false ) {
 271                         # Explicit range
 272                         list( $start, $end ) = array_map( 'trim', explode( '-', $range, 2 ) );
 273                         $start = self::toUnsigned6( $start ); $end = self::toUnsigned6( $end );
 274                         if ( $start > $end ) {
 275                                 $start = $end = false;
 276                         } else {
 277                                 $start = wfBaseConvert( $start, 10, 16, 32, false );
 278                                 $end = wfBaseConvert( $end, 10, 16, 32, false );
 279                         }
 280                         # see toHex() comment
 281                         $start = "v6-$start"; $end = "v6-$end";
 282                 } else {
 283                         # Single IP
 284                         $start = $end = self::toHex( $range );
 285                 }
 286                 if ( $start === false || $end === false ) {
 287                         return array( false, false );
 288                 } else {
 289                         return array( $start, $end );
 290                 }
 291     }
 292
 293         /**
 294          * Validate an IP address.
 295          * @return boolean True if it is valid.
 296          */
 297         public static function isValid( $ip ) {
 298                 return ( preg_match( '/^' . RE_IP_ADD . '$/', $ip) || preg_match( '/^' . RE_IPV6_ADD . '$/', $ip) );
 299         }
 300
 301         /**
 302          * Validate an IP Block.
 303          * @return boolean True if it is valid.
 304          */
 305         public static function isValidBlock( $ipblock ) {
 306                 return ( count(self::toArray($ipblock)) == 1 + 5 );
 307         }
 308
 309         /**
 310          * Determine if an IP address really is an IP address, and if it is public,
 311          * i.e. not RFC 1918 or similar
 312          * Comes from ProxyTools.php
 313          */
 314         public static function isPublic( $ip ) {
 315                 $n = self::toUnsigned( $ip );
 316                 if ( !$n ) {
 317                         return false;
 318                 }
 319
 320                 // ip2long accepts incomplete addresses, as well as some addresses
 321                 // followed by garbage characters. Check that it's really valid.
 322                 if( $ip != long2ip( $n ) ) {
 323                         return false;
 324                 }
 325
 326                 static $privateRanges = false;
 327                 if ( !$privateRanges ) {
 328                         $privateRanges = array(
 329                                 array( '10.0.0.0',    '10.255.255.255' ),   # RFC 1918 (private)
 330                                 array( '172.16.0.0',  '172.31.255.255' ),   #     "
 331                                 array( '192.168.0.0', '192.168.255.255' ),  #     "
 332                                 array( '0.0.0.0',     '0.255.255.255' ),    # this network
 333                                 array( '127.0.0.0',   '127.255.255.255' ),  # loopback
 334                         );
 335                 }
 336
 337                 foreach ( $privateRanges as $r ) {
 338                         $start = self::toUnsigned( $r[0] );
 339                         $end = self::toUnsigned( $r[1] );
 340                         if ( $n >= $start && $n <= $end ) {
 341                                 return false;
 342                         }
 343                 }
 344                 return true;
 345         }
 346
 347         /**
 348          * Split out an IP block as an array of 4 bytes and a mask,
 349          * return false if it can't be determined
 350          *
 351          * @param $ip string A quad dotted/octet IP address
 352          * @return array
 353          */
 354         public static function toArray( $ipblock ) {
 355                 $matches = array();
 356                 if( preg_match( '/^' . RE_IP_ADD . '(?:\/(?:'.RE_IP_PREFIX.'))?' . '$/', $ipblock, $matches ) ) {
 357                         return $matches;
 358                 } else if ( preg_match( '/^' . RE_IPV6_ADD . '(?:\/(?:'.RE_IPV6_PREFIX.'))?' . '$/', $ipblock, $matches ) ) {
 359                         return $matches;
 360                 } else {
 361                         return false;
 362                 }
 363         }
 364
 365         /**
 366          * Return a zero-padded hexadecimal representation of an IP address.
 367          *
 368          * Hexadecimal addresses are used because they can easily be extended to
 369          * IPv6 support. To separate the ranges, the return value from this
 370          * function for an IPv6 address will be prefixed with "v6-", a non-
 371          * hexadecimal string which sorts after the IPv4 addresses.
 372          *
 373          * @param $ip Quad dotted/octet IP address.
 374          * @return hexidecimal
 375          */
 376         public static function toHex( $ip ) {
 377                 $n = self::toUnsigned( $ip );
 378                 if ( $n !== false ) {
 379                         $n = self::isIPv6($ip) ? "v6-" . wfBaseConvert( $n, 10, 16, 32, false ) : wfBaseConvert( $n, 10, 16, 8, false );
 380                 }
 381                 return $n;
 382         }
 383
 384         /**
 385          * Given an IP address in dotted-quad/octet notation, returns an unsigned integer.
 386          * Like ip2long() except that it actually works and has a consistent error return value.
 387          * Comes from ProxyTools.php
 388          * @param $ip Quad dotted IP address.
 389          * @return integer
 390          */
 391         public static function toUnsigned( $ip ) {
 392                 // Use IPv6 functions if needed
 393                 if ( self::isIPv6( $ip ) ) {
 394                         return self::toUnsigned6( $ip );
 395                 }
 396                 if ( $ip == '255.255.255.255' ) {
 397                         $n = -1;
 398                 } else {
 399                         $n = ip2long( $ip );
 400                         if ( $n == -1 || $n === false ) { # Return value on error depends on PHP version
 401                                 $n = false;
 402                         }
 403                 }
 404                 if ( $n < 0 ) {
 405                         $n += pow( 2, 32 );
 406                 }
 407                 return $n;
 408         }
 409
 410         /**
 411          * Convert a dotted-quad IP to a signed integer
 412          * Returns false on failure
 413          */
 414         public static function toSigned( $ip ) {
 415                 if ( $ip == '255.255.255.255' ) {
 416                         $n = -1;
 417                 } else {
 418                         $n = ip2long( $ip );
 419                         if ( $n == -1 ) {
 420                                 $n = false;
 421                         }
 422                 }
 423                 return $n;
 424         }
 425
 426         /**
 427          * Convert a network specification in CIDR notation to an integer network and a number of bits
 428          * @return array(string, int)
 429          */
 430         public static function parseCIDR( $range ) {
 431                 $parts = explode( '/', $range, 2 );
 432                 if ( count( $parts ) != 2 ) {
 433                         return array( false, false );
 434                 }
 435                 $network = self::toSigned( $parts[0] );
 436                 if ( $network !== false && is_numeric( $parts[1] ) && $parts[1] >= 0 && $parts[1] <= 32 ) {
 437                         $bits = $parts[1];
 438                         if ( $bits == 0 ) {
 439                                 $network = 0;
 440                         } else {
 441                                 $network &= ~((1 << (32 - $bits)) - 1);
 442                         }
 443                         # Convert to unsigned
 444                         if ( $network < 0 ) {
 445                                 $network += pow( 2, 32 );
 446                         }
 447                 } else {
 448                         $network = false;
 449                         $bits = false;
 450                 }
 451                 return array( $network, $bits );
 452         }
 453
 454         /**
 455          * Given a string range in a number of formats, return the start and end of
 456          * the range in hexadecimal.
 457          *
 458          * Formats are:
 459          *     1.2.3.4/24          CIDR
 460          *     1.2.3.4 - 1.2.3.5   Explicit range
 461          *     1.2.3.4             Single IP
 462          *
 463          *     2001:0db8:85a3::7344/96                                   CIDR
 464          *     2001:0db8:85a3::7344 - 2001:0db8:85a3::7344   Explicit range
 465          *     2001:0db8:85a3::7344                                      Single IP
 466          * @return array(string, int)
 467          */
 468         public static function parseRange( $range ) {
 469                 // Use IPv6 functions if needed
 470                 if ( self::isIPv6( $range ) ) {
 471                         return self::parseRange6( $range );
 472                 }
 473                 if ( strpos( $range, '/' ) !== false ) {
 474                         # CIDR
 475                         list( $network, $bits ) = self::parseCIDR( $range );
 476                         if ( $network === false ) {
 477                                 $start = $end = false;
 478                         } else {
 479                                 $start = sprintf( '%08X', $network );
 480                                 $end = sprintf( '%08X', $network + pow( 2, (32 - $bits) ) - 1 );
 481                         }
 482                 } elseif ( strpos( $range, '-' ) !== false ) {
 483                         # Explicit range
 484                         list( $start, $end ) = array_map( 'trim', explode( '-', $range, 2 ) );
 485                         if( self::isIPAddress( $start ) && self::isIPAddress( $end ) ) {
 486                                 $start = self::toUnsigned( $start ); $end = self::toUnsigned( $end );
 487                                 if ( $start > $end ) {
 488                                         $start = $end = false;
 489                                 } else {
 490                                         $start = sprintf( '%08X', $start );
 491                                         $end = sprintf( '%08X', $end );
 492                                 }
 493                         } else {
 494                                 $start = $end = false;
 495                         }
 496                 } else {
 497                         # Single IP
 498                         $start = $end = self::toHex( $range );
 499                 }
 500                 if ( $start === false || $end === false ) {
 501                         return array( false, false );
 502                 } else {
 503                         return array( $start, $end );
 504                 }
 505     }
 506
 507     /**
 508      * Determine if a given IPv4/IPv6 address is in a given CIDR network
 509      * @param $addr The address to check against the given range.
 510      * @param $range The range to check the given address against.
 511      * @return bool Whether or not the given address is in the given range.
 512      */
 513     public static function isInRange( $addr, $range ) {
 514     // Convert to IPv6 if needed
 515         $hexIP = self::toHex( $addr );
 516         list( $start, $end ) = self::parseRange( $range );
 517         return (strcmp($hexIP, $start) >= 0 &&
 518                 strcmp($hexIP, $end) <= 0);
 519     }
 520
 521     /**
 522      * Convert some unusual representations of IPv4 addresses to their
 523      * canonical dotted quad representation.
 524      *
 525      * This currently only checks a few IPV4-to-IPv6 related cases.  More
 526      * unusual representations may be added later.
 527      *
 528      * @param $addr something that might be an IP address
 529      * @return valid dotted quad IPv4 address or null
 530      */
 531     public static function canonicalize( $addr ) {
 532                 if ( self::isValid( $addr ) )
 533                         return $addr;
 534
 535                 // Annoying IPv6 representations like ::ffff:1.2.3.4
 536                 if ( strpos($addr,':') !==false && strpos($addr,'.') !==false ) {
 537                         $addr = str_replace( '.', ':', $addr );
 538                         if( IP::isIPv6( $addr ) )
 539                                 return $addr;
 540                 }
 541
 542                 // IPv6 loopback address
 543                 $m = array();
 544                 if ( preg_match( '/^0*' . RE_IPV6_GAP . '1$/', $addr, $m ) )
 545                         return '127.0.0.1';
 546
 547                 // IPv4-mapped and IPv4-compatible IPv6 addresses
 548                 if ( preg_match( '/^' . RE_IPV6_V4_PREFIX . '(' . RE_IP_ADD . ')$/i', $addr, $m ) )
 549                     return $m[1];
 550                 if ( preg_match( '/^' . RE_IPV6_V4_PREFIX . RE_IPV6_WORD . ':' . RE_IPV6_WORD . '$/i', $addr, $m ) )
 551                     return long2ip( ( hexdec( $m[1] ) << 16 ) + hexdec( $m[2] ) );
 552
 553                 return null;  // give up
 554     }
 555 }