#!/usr/bin/php
<?php
################################################################################
# cc2asn_parse
# ------------
#
# 0: RIR (Regional Internet Registry)
# 1: ISO-3166-1-alpha-2 code
# 2: Type (ASN|IPv4|IPv6)
# 3: Allocation
# 4: Range
# 5: Allocation Date
# 6: Comment
#
# Author: Tor Inge Skaar - Norwegian Honeynet Chapter - www.honeynor.no
#
################################################################################

$conf = parse_ini_file( "/etc/cc2asn.conf" );

$data = array();
$err  = array();

/* Parse each downloaded file and create a data structure */
foreach ( glob( $conf["DATADIR"]."/*.data") as $file ) {
	$fp = @fopen( $file, "r" );
	if ( !$fp ) {
		fprintf( STDERR, "Failed to open %s\n", $file );
		exit( 1 );
	}
	else {
		print("Parsing $file\n");
	}

	while ( !feof( $fp ) ) {
		$buf = fgets( $fp, 4096 );
		$b = explode( "|", $buf );
		foreach( $b as $key => $value ) {
			$b[$key] = strtoupper( trim( $value ));
		}
		if ( $b[2] == "ASN" ) {
			if ( $b[4] > 1 ) {
				/* ASN range defined */
				$dot = strpos( $b[3], "." ); 
				if ( $dot === FALSE ) {
					/* Normal integer representation of both 2-byte and 4-byte ASN */
					for ( $i = $b[3]; $i < ( $b[3] + $b[4] ); $i++ ) {
						$data[$b[1]]["ASN"][] = $i;
					}
				}
				else {
					/* Handle Canonical Textual Form of 4-byte ASN with range defined */
					$x = substr( $b[3], 0, $dot );
					$y = substr( $b[3], $dot + 1 );
					for ( $i = $y; $i < ( $y + $b[4] ); $i++ ) {
						$data[$b[1]]["ASN"][] = $x.".".$i;
					}
				}
			} 
			else {
				/* Single ASN (either 2 og 4 byte) */
				$data[$b[1]]["ASN"][] = $b[3];
			}
		}
		elseif ( $b[2] == "IPV4" ) {
			$suffix = 32 - log( $b[4], 2 );
			if ( strstr( $suffix, "." ) === FALSE ) {
				$data[$b[1]]["IPV4"][] = $b[3]."/".$suffix;	
			}
			else {
				$err[$b[3]]["CC"]     = $b[1];
				$err[$b[3]]["SUFFIX"] = $suffix;
			}
		}
		elseif ( $b[2] == "IPV6" ) {
			$data[$b[1]]["IPV6"][] = $b[3]."/".$b[4];
		}
		else { /* Unknown type */
			$data[$b[1]][$b[2]][] = $b[3];
		}
	}
	fclose($fp);
}

/* Simple fault checking */
if ( sizeof( $err ) > 50000 ) {
	fprintf( STDERR, "More than 50000 faulty prefixes. Somethings funky. Terminating\n" );
	exit ( 1 );
} 


/******** NOTICE ********* 

The above code detects and stores faulty IPv4 prefixes in the err-array. These 
are cases where the amount of allocated IP-addresses do not align with a proper
network mask. One way to correct any faulty prefixes is by correlating the 
entries in the err-array with BGP anouncements. There are several free services
that may be used for this exact purpose, however I chose not to include the code
to do that here, because such services run with limited resources. But please 
do drop us an e-mail, and we'll happily share more code with you. 

********* NOTICE ********/ 


/* Create a flat file structure of the data */
print( "Storing data in file structure at ".$conf["DBDIR"]."\n" );
@mkdir($conf["DBDIR"]);
foreach ( $data as $cc => $d ) {
	if ( sizeof( $d["ASN"] ) > 0 ) {
		$list_array = $d["ASN"];
		natsort( $list_array );
		$list = "";
		foreach ( $list_array as $asn ) {
			$list .= "AS".$asn."\n";	
		}
		file_put_contents( $conf["DBDIR"]."/".$cc."_ASN", $list );
	}	
	
	if ( sizeof( $d["IPV4"] ) > 0 ) {
		$list_array = $d["IPV4"];
		natsort( $list_array );
		$list = "";
		foreach ( $list_array as $prefix ) {
			$list .= $prefix."\n";	
		}
		file_put_contents( $conf["DBDIR"]."/".$cc."_IPV4", $list );
	}

	if ( sizeof( $d["IPV6"] ) > 0 ) {
		$list_array = $d["IPV6"];
		natsort( $list_array );
		$list = "";
		foreach ( $list_array as $prefix ) {
			$list .= $prefix."\n";	
		}
		file_put_contents( $conf["DBDIR"]."/".$cc."_IPV6", $list );
	}
}

?>