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                 $ip = str_replace('::', str_repeat(':0000', 8 - substr_count($ip, ':')) . ':', $ip);
 107         }
 108         // Add leading zereos to each bloc as needed
 109         $ip = explode( ':', $ip );
 110         foreach ( $ip as $v ) {
 111                 $v = str_pad( $v, 4, 0, STR_PAD_LEFT );
 112         }
 113         $ip = implode( ':', $ip );
 114         return $ip;
 115         }
 116
 117         /**
 118          * Given an unsigned integer, returns an IPv6 address in octet notation
 119          * @param $ip integer IP address.
 120          * @return string
 121          */
 122         public static function toOctet( $ip_int ) {
 123                 // Convert integer to binary
 124                 $ip_int = wfBaseConvert($ip_int, 10, 2, 128);
 125                 // Seperate into 8 octets
 126                 $ip_oct = wfBaseConvert( substr( $ip_int, 0, 16 ), 2, 16, 1, false );
 127                 for ($n=1; $n < 8; $n++) {
 128                         // Convert to hex, and add ":" marks, with leading zeroes
 129                         $ip_oct .= ':' . wfBaseConvert( substr($ip_int, 16*$n, 16), 2, 16, 4, false );
 130                 }
 131         return $ip_oct;
 132         }
 133
 134         /**
 135          * Convert a network specification in IPv6 CIDR notation to an integer network and a number of bits
 136          * @return array(string, int)
 137          */
 138         public static function parseCIDR6( $range ) {
 139                 # Expand any IPv6 IP
 140                 $parts = explode( '/', IP::sanitizeIP( $range ), 2 );
 141                 if ( count( $parts ) != 2 ) {
 142                         return array( false, false );
 143                 }
 144                 $network = self::toUnsigned6( $parts[0] );
 145                 if ( $network !== false && is_numeric( $parts[1] ) && $parts[1] >= 0 && $parts[1] <= 128 ) {
 146                         $bits = $parts[1];
 147                         if ( $bits == 0 ) {
 148                                 $network = 0;
 149                         } else {
 150                         # Native 32 bit functions WONT work here!!!
 151                         # Convert to a padded binary number
 152                                 $network = wfBaseConvert( $network, 10, 2, 128 );
 153                         # Truncate the last (128-$bits) bits and replace them with zeros
 154                                 $network = str_pad( substr( $network, 0, $bits ), 128, 0, STR_PAD_RIGHT );
 155                         # Convert back to an integer
 156                                 $network = wfBaseConvert( $network, 2, 10 );
 157                         }
 158                 } else {
 159                         $network = false;
 160                         $bits = false;
 161                 }
 162
 163                 return array( $network, $bits );
 164         }
 165
 166         /**
 167          * Given a string range in a number of formats, return the start and end of
 168          * the range in hexadecimal. For IPv6.
 169          *
 170          * Formats are:
 171          *     2001:0db8:85a3::7344/96                                   CIDR
 172          *     2001:0db8:85a3::7344 - 2001:0db8:85a3::7344   Explicit range
 173          *     2001:0db8:85a3::7344/96                                   Single IP
 174          * @return array(string, int)
 175          */
 176         public static function parseRange6( $range ) {
 177                 # Expand any IPv6 IP
 178                 $range = IP::sanitizeIP( $range );
 179                 if ( strpos( $range, '/' ) !== false ) {
 180                         # CIDR
 181                         list( $network, $bits ) = self::parseCIDR6( $range );
 182                         if ( $network === false ) {
 183                                 $start = $end = false;
 184                         } else {
 185                                 $start = wfBaseConvert( $network, 10, 16, 32, false );
 186                                 # Turn network to binary (again)
 187                                 $end = wfBaseConvert( $network, 10, 2, 128 );
 188                                 # Truncate the last (128-$bits) bits and replace them with ones
 189                                 $end = str_pad( substr( $end, 0, $bits ), 128, 1, STR_PAD_RIGHT );
 190                                 # Convert to hex
 191                                 $end = wfBaseConvert( $end, 2, 16, 32, false );
 192                         }
 193                 } elseif ( strpos( $range, '-' ) !== false ) {
 194                         # Explicit range
 195                         list( $start, $end ) = array_map( 'trim', explode( '-', $range, 2 ) );
 196                         $start = self::toUnsigned6( $start ); $end = self::toUnsigned6( $end );
 197                         if ( $start > $end ) {
 198                                 $start = $end = false;
 199                         } else {
 200                                 $start = wfBaseConvert( $start, 10, 16, 32, false );
 201                                 $end = wfBaseConvert( $end, 10, 16, 32, false );
 202                         }
 203                 } else {
 204                         # Single IP
 205                         $start = $end = self::toHex( $range );
 206                 }
 207                 if ( $start === false || $end === false ) {
 208                         return array( false, false );
 209                 } else {
 210                         return array( $start, $end );
 211                 }
 212     }
 213
 214         /**
 215          * Validate an IP address.
 216          * @return boolean True if it is valid.
 217          */
 218         public static function isValid( $ip ) {
 219                 return ( preg_match( '/^' . RE_IP_ADD . '$/', $ip) || preg_match( '/^' . RE_IPV6_ADD . '$/', $ip) );
 220         }
 221
 222         /**
 223          * Validate an IP Block.
 224          * @return boolean True if it is valid.
 225          */
 226         public static function isValidBlock( $ipblock ) {
 227                 return ( count(self::toArray($ipblock)) == 1 + 5 );
 228         }
 229
 230         /**
 231          * Determine if an IP address really is an IP address, and if it is public,
 232          * i.e. not RFC 1918 or similar
 233          * Comes from ProxyTools.php
 234          */
 235         public static function isPublic( $ip ) {
 236                 $n = self::toUnsigned( $ip );
 237                 if ( !$n ) {
 238                         return false;
 239                 }
 240
 241                 // ip2long accepts incomplete addresses, as well as some addresses
 242                 // followed by garbage characters. Check that it's really valid.
 243                 if( $ip != long2ip( $n ) ) {
 244                         return false;
 245                 }
 246
 247                 static $privateRanges = false;
 248                 if ( !$privateRanges ) {
 249                         $privateRanges = array(
 250                                 array( '10.0.0.0',    '10.255.255.255' ),   # RFC 1918 (private)
 251                                 array( '172.16.0.0',  '172.31.255.255' ),   #     "
 252                                 array( '192.168.0.0', '192.168.255.255' ),  #     "
 253                                 array( '0.0.0.0',     '0.255.255.255' ),    # this network
 254                                 array( '127.0.0.0',   '127.255.255.255' ),  # loopback
 255                         );
 256                 }
 257
 258                 foreach ( $privateRanges as $r ) {
 259                         $start = self::toUnsigned( $r[0] );
 260                         $end = self::toUnsigned( $r[1] );
 261                         if ( $n >= $start && $n <= $end ) {
 262                                 return false;
 263                         }
 264                 }
 265                 return true;
 266         }
 267
 268         /**
 269          * Split out an IP block as an array of 4 bytes and a mask,
 270          * return false if it can't be determined
 271          *
 272          * @parameter $ip string A quad dotted/octet IP address
 273          * @return array
 274          */
 275         public static function toArray( $ipblock ) {
 276                 $matches = array();
 277                 if( preg_match( '/^' . RE_IP_ADD . '(?:\/(?:'.RE_IP_PREFIX.'))?' . '$/', $ipblock, $matches ) ) {
 278                         return $matches;
 279                 } else if ( preg_match( '/^' . RE_IPV6_ADD . '(?:\/(?:'.RE_IPV6_PREFIX.'))?' . '$/', $ipblock, $matches ) ) {
 280                         return $matches;
 281                 } else {
 282                         return false;
 283                 }
 284         }
 285
 286         /**
 287          * Return a zero-padded hexadecimal representation of an IP address.
 288          *
 289          * Hexadecimal addresses are used because they can easily be extended to
 290          * IPv6 support. To separate the ranges, the return value from this
 291          * function for an IPv6 address will be prefixed with "v6-", a non-
 292          * hexadecimal string which sorts after the IPv4 addresses.
 293          *
 294          * @param $ip Quad dotted/octet IP address.
 295          * @return hexidecimal
 296          */
 297         public static function toHex( $ip ) {
 298                 $n = self::toUnsigned( $ip );
 299                 if ( $n !== false ) {
 300                         $n = ( self::isIPv6($ip) ) ? wfBaseConvert( $n, 10, 16, 32, false ) : wfBaseConvert( $n, 10, 16, 8, false );
 301                 }
 302                 return $n;
 303         }
 304
 305         /**
 306          * Given an IP address in dotted-quad notation, returns an unsigned integer.
 307          * Like ip2long() except that it actually works and has a consistent error return value.
 308          * Comes from ProxyTools.php
 309          * @param $ip Quad dotted IP address.
 310          * @return integer
 311          */
 312         public static function toUnsigned( $ip ) {
 313                 // Use IPv6 functions if needed
 314                 if ( self::isIPv6( $ip ) ) {
 315                         return self::toUnsigned6( $ip );
 316                 }
 317                 if ( $ip == '255.255.255.255' ) {
 318                         $n = -1;
 319                 } else {
 320                         $n = ip2long( $ip );
 321                         if ( $n == -1 || $n === false ) { # Return value on error depends on PHP version
 322                                 $n = false;
 323                         }
 324                 }
 325                 if ( $n < 0 ) {
 326                         $n += pow( 2, 32 );
 327                 }
 328                 return $n;
 329         }
 330
 331         /**
 332          * Convert a dotted-quad IP to a signed integer
 333          * Returns false on failure
 334          */
 335         public static function toSigned( $ip ) {
 336                 if ( $ip == '255.255.255.255' ) {
 337                         $n = -1;
 338                 } else {
 339                         $n = ip2long( $ip );
 340                         if ( $n == -1 ) {
 341                                 $n = false;
 342                         }
 343                 }
 344                 return $n;
 345         }
 346
 347         /**
 348          * Convert a network specification in CIDR notation to an integer network and a number of bits
 349          * @return array(string, int)
 350          */
 351         public static function parseCIDR( $range ) {
 352                 $parts = explode( '/', $range, 2 );
 353                 if ( count( $parts ) != 2 ) {
 354                         return array( false, false );
 355                 }
 356                 $network = self::toSigned( $parts[0] );
 357                 if ( $network !== false && is_numeric( $parts[1] ) && $parts[1] >= 0 && $parts[1] <= 32 ) {
 358                         $bits = $parts[1];
 359                         if ( $bits == 0 ) {
 360                                 $network = 0;
 361                         } else {
 362                                 $network &= ~((1 << (32 - $bits)) - 1);
 363                         }
 364                         # Convert to unsigned
 365                         if ( $network < 0 ) {
 366                                 $network += pow( 2, 32 );
 367                         }
 368                 } else {
 369                         $network = false;
 370                         $bits = false;
 371                 }
 372                 return array( $network, $bits );
 373         }
 374
 375         /**
 376          * Given a string range in a number of formats, return the start and end of
 377          * the range in hexadecimal. For IPv4.
 378          *
 379          * Formats are:
 380          *     1.2.3.4/24          CIDR
 381          *     1.2.3.4 - 1.2.3.5   Explicit range
 382          *     1.2.3.4             Single IP
 383          * @return array(string, int)
 384          */
 385         public static function parseRange( $range ) {
 386                 // Use IPv6 functions if needed
 387                 if ( self::isIPv6( $range ) ) {
 388                         return self::parseRange6( $range );
 389                 }
 390                 if ( strpos( $range, '/' ) !== false ) {
 391                         # CIDR
 392                         list( $network, $bits ) = self::parseCIDR( $range );
 393                         if ( $network === false ) {
 394                                 $start = $end = false;
 395                         } else {
 396                                 $start = sprintf( '%08X', $network );
 397                                 $end = sprintf( '%08X', $network + pow( 2, (32 - $bits) ) - 1 );
 398                         }
 399                 } elseif ( strpos( $range, '-' ) !== false ) {
 400                         # Explicit range
 401                         list( $start, $end ) = array_map( 'trim', explode( '-', $range, 2 ) );
 402                         $start = self::toUnsigned( $start ); $end = self::toUnsigned( $end );
 403                         if ( $start > $end ) {
 404                                 $start = $end = false;
 405                         } else {
 406                                 $start = sprintf( '%08X', $start );
 407                                 $end = sprintf( '%08X', $end );
 408                         }
 409                 } else {
 410                         # Single IP
 411                         $start = $end = self::toHex( $range );
 412                 }
 413                 if ( $start === false || $end === false ) {
 414                         return array( false, false );
 415                 } else {
 416                         return array( $start, $end );
 417                 }
 418     }
 419
 420     /**
 421      * Determine if a given IPv4 address is in a given CIDR network
 422      * @param $addr The address to check against the given range.
 423      * @param $range The range to check the given address against.
 424      * @return bool Whether or not the given address is in the given range.
 425      */
 426     public static function isInRange( $addr, $range ) {
 427     // Convert to IPv6 if needed
 428         $unsignedIP = self::toUnsigned6( self::IPv4toIPv6($addr) );
 429         list( $start, $end ) = self::parseRange6( self::IPv4toIPv6($range) );
 430         return (($unsignedIP >= $start) && ($unsignedIP <= $end));
 431     }
 432
 433     /**
 434      * Convert some unusual representations of IPv4 addresses to their
 435      * canonical dotted quad representation.
 436      *
 437      * This currently only checks a few IPV4-to-IPv6 related cases.  More
 438      * unusual representations may be added later.
 439      *
 440      * @param $addr something that might be an IP address
 441      * @return valid dotted quad IPv4 address or null
 442      */
 443     public static function canonicalize( $addr ) {
 444         if ( self::isValid( $addr ) )
 445             return $addr;
 446
 447         // IPv6 loopback address
 448         $m = array();
 449         if ( preg_match( '/^0*' . RE_IPV6_GAP . '1$/', $addr, $m ) )
 450             return '127.0.0.1';
 451
 452         // IPv4-mapped and IPv4-compatible IPv6 addresses
 453         if ( preg_match( '/^' . RE_IPV6_V4_PREFIX . '(' . RE_IP_ADD . ')$/i', $addr, $m ) )
 454             return $m[1];
 455         if ( preg_match( '/^' . RE_IPV6_V4_PREFIX . RE_IPV6_WORD . ':' . RE_IPV6_WORD . '$/i', $addr, $m ) )
 456             return long2ip( ( hexdec( $m[1] ) << 16 ) + hexdec( $m[2] ) );
 457
 458         return null;  // give up
 459     }
 460 }
 461
 462 ?>