includes/IP.php

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