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