Example to update GeoIP (node.js geoip-lite) from commercial DB

update-geoip.js

// fetches and converts maxmind lite databases

'use strict';

var cp = require('child_process');
var fs = require('fs');
var http = require('http');
var path = require('path');
var url = require('url');
var zlib = require('zlib');

fs.existsSync = fs.existsSync || path.existsSync;

var async = require('async');
var colors = require('colors');
var LineInputStream = require('line-input-stream');
var rimraf = require('rimraf').sync;
var unzip = require('unzip');
var utils = {};


var dataPath = path.join(__dirname, '..', 'node_modules', 'geoip-lite', 'data');
var tmpPath = path.join(__dirname, '..', 'tmp');

var databases = [{
	type: 'country',
	url: 'http://download.maxmind.com/app/geoip_download?edition_id=108&suffix=zip&license_key=XXXXXX#/GeoIP.zip',
	src: 'GeoIP-108.csv',
	dest: 'geoip-country.dat'
},{
	type: 'country',
	url: 'http://geolite.maxmind.com/download/geoip/database/GeoIPv6.csv.gz',
	src: 'GeoIPv6.csv',
	dest: 'geoip-country6.dat'
}/*,{
	type: 'city-extended',
	url: 'http://geolite.maxmind.com/download/geoip/database/GeoLiteCity_CSV/GeoLiteCity_20121204.zip',
	src: [
		'GeoLiteCity-Blocks.csv',
		'GeoLiteCity-Location.csv'
	],
	dest: [
		'geoip-city.dat',
		'geoip-city-names.dat'
	]
},{
	type: 'city',
	url: 'http://geolite.maxmind.com/download/geoip/database/GeoLiteCityv6-beta/GeoLiteCityv6.csv.gz',
	src: 'GeoLiteCityv6.csv',
	dest: 'geoip-city6.dat'
}*/];

utils.aton4 = function(a) {
  a = a.split(/\./);
  return ((parseInt(a[0], 10)<<24)>>>0) + ((parseInt(a[1], 10)<<16)>>>0) + ((parseInt(a[2], 10)<<8)>>>0) + (parseInt(a[3], 10)>>>0);
};

utils.aton6 = function(a) {
  a = a.replace(/"/g, '').split(/:/);

  var l = a.length - 1;
  var i;

  if (a[l] === '') {
    a[l] = 0;
  }

  if (l < 7) {
    a.length = 8;

    for (i = l; i >= 0 && a[i] !== ''; i--) {
      a[7-l+i] = a[i];
    }
  }

  for (i = 0; i < 8; i++) {
    if (!a[i]) {
      a[i]=0;
    } else {
      a[i] = parseInt(a[i], 16);
    }
  }

  var r = [];
  for (i = 0; i<4; i++) {
    r.push(((a[2*i]<<16) + a[2*i+1])>>>0);
  }

  return r;
};


utils.cmp = function(a, b) {
  if (typeof a === 'number' && typeof b === 'number') {
    return (a < b ? -1 : (a > b ? 1 : 0));
  }

  if (a instanceof Array && b instanceof Array) {
    return this.cmp6(a, b);
  }

  return null;
};

utils.cmp6 = function(a, b) {
  for (var ii = 0; ii < 2; ii++) {
    if (a[ii] < b[ii]) {
      return -1;
    }

    if (a[ii] > b[ii]) {
      return 1;
    }
  }

  return 0;
};

utils.isPrivateIP = function(addr) {
  addr = addr.toString();

  return addr.match(/^10\.([0-9]{1,3})\.([0-9]{1,3})\.([0-9]{1,3})/) != null ||
    addr.match(/^192\.168\.([0-9]{1,3})\.([0-9]{1,3})/) != null ||
    addr.match(/^172\.16\.([0-9]{1,3})\.([0-9]{1,3})/) != null ||
    addr.match(/^127\.([0-9]{1,3})\.([0-9]{1,3})\.([0-9]{1,3})/) != null ||
    addr.match(/^169\.254\.([0-9]{1,3})\.([0-9]{1,3})/) != null ||
    addr.match(/^fc00:/) != null || addr.match(/^fe80:/) != null;
};

utils.ntoa4 = function(n) {
  n = n.toString();
  n = '' + (n>>>24&0xff) + '.' + (n>>>16&0xff) + '.' + (n>>>8&0xff) + '.' + (n&0xff);

  return n;
};

utils.ntoa6 = function(n) {
  var a = "[";

  for (var i = 0; i<n.length; i++) {
    a += (n[i]>>>16).toString(16) + ':';
    a += (n[i]&0xffff).toString(16) + ':';
  }

  a = a.replace(/:$/, ']').replace(/:0+/g, ':').replace(/::+/, '::');

  return a;
};

// __ main __ //

function mkdir(name) {
	var dir = path.dirname(name);
	if (!fs.existsSync(dir)) {
		fs.mkdirSync(dir);
	}
}

function fetch(downloadUrl, cb) {
	function getOptions() {
		if (process.env.http_proxy) {
			var options = url.parse(process.env.http_proxy);

			options.path = downloadUrl;
			options.headers = {
				Host: url.parse(downloadUrl).host
			};

			return options;
		} else {
			return url.parse(downloadUrl);
		}
	}

	function onResponse(response) {
		var status = response.statusCode;

		if (status !== 200) {
			console.log('ERROR'.red + ': HTTP Request Failed [%d %s]', status, http.STATUS_CODES[status]);
			client.abort();
			process.exit();
		}

		var tmpFilePipe;
		var tmpFileStream = fs.createWriteStream(tmpFile);

		if (gzip) {
			tmpFilePipe = response.pipe(zlib.createGunzip()).pipe(tmpFileStream);
		} else {
			tmpFilePipe = response.pipe(tmpFileStream);
		}

		tmpFilePipe.on('close', function() {
			console.log(' DONE'.green);
			cb(tmpFile, fileName);
		});
	}

	var fileName = downloadUrl.split('/').pop();
	var gzip = (fileName.indexOf('.gz') !== -1);

	if (gzip) {
		fileName = fileName.replace('.gz', '');
	}

	var tmpFile = path.join(tmpPath, fileName);

	mkdir(tmpFile);

	var client = http.get(getOptions(), onResponse);

	process.stdout.write('Retrieving ' + fileName + ' ...');
}

function extract(wantedFiles, tmpFile, tmpFileName, cb) {
	// force cast to array
	if (typeof(wantedFiles) === "string")
		wantedFiles = new Array(wantedFiles);
	
	if (tmpFileName.indexOf('.zip') === -1) {
		cb();
	} else {
		process.stdout.write('Extracting ' + tmpFileName + ' ...');

		var pipeSteam = fs.createReadStream(tmpFile).pipe(unzip.Parse());
		pipeSteam.on('entry', function(entry) {
			var fileName = entry.path.toString().split('/').pop();
			if (wantedFiles.indexOf(fileName) > -1)
				entry.pipe(fs.createWriteStream(path.join(path.dirname(tmpFile), fileName)));
			else
				entry.autodrain();
		});
		pipeSteam.on('finish', function() {
			console.log(' DONE'.green);
			cb();
		});
	}
}

function processCountryData(src, dest, cb) {
	function processLine(line) {
		if (line.match(/^Copyright/) || !line.match(/\d/)) {
			return;
		}
		var fields = line.split(/, */);

		if (fields.length < 6) {
			console.log("weird line: %s::", line);
			return;
		}

		var sip;
		var eip;
		var cc = fields[4].replace(/"/g, '');
		var b;
		var bsz;
		var i;

		if (fields[0].match(/:/)) {
			// IPv6
			bsz = 34;
			sip = utils.aton6(fields[0]);
			eip = utils.aton6(fields[1]);

			b = new Buffer(bsz);
			for (i = 0; i < sip.length; i++) {
				b.writeUInt32BE(sip[i], i * 4);
			}

			for (i = 0; i < eip.length; i++) {
				b.writeUInt32BE(eip[i], 16 + (i * 4));
			}
		} else {
			// IPv4
			bsz = 10;

			sip = parseInt(fields[2].replace(/"/g, ''), 10);
			eip = parseInt(fields[3].replace(/"/g, ''), 10);

			b = new Buffer(bsz);
			b.fill(0);
			b.writeUInt32BE(sip, 0);
			b.writeUInt32BE(eip, 4);
		}

		b.write(cc, bsz - 2);

		fs.writeSync(datFile, b, 0, bsz, null);
	}

	var dataFile = path.join(dataPath, dest);
	var tmpDataFile = path.join(tmpPath, src);

	rimraf(dataFile);
	mkdir(dataFile);

	process.stdout.write('Processing Data (may take a moment) ...');
	var datFile = fs.openSync(dataFile, "w");
	var csvStream = new LineInputStream(fs.createReadStream(tmpDataFile), /[\r\n]+/);

	csvStream.setEncoding('utf8');

	csvStream.on('line', processLine);

	csvStream.on('end', function() {
		console.log(' DONE'.green);
		cb();
	});
}

function processCityData(src, dest, cb) {
	function processLine(line) {
		if (line.match(/^Copyright/) || !line.match(/\d/)) {
			return;
		}

		var fields = line.replace(/"/g, '').split(/, */);
		var sip;
		var eip;
		var locId;
		var b;
		var bsz;

		var i;

		if (fields[0].match(/:/)) {
			// IPv6
			var offset = 0;

			var cc = fields[4];
			var city = fields[6];
			var lat = Math.round(parseFloat(fields[7]) * 10000);
			var lon = Math.round(parseFloat(fields[8]) * 10000);
			var rg = fields[5];

			bsz = 58;
			sip = utils.aton6(fields[0]);
			eip = utils.aton6(fields[1]);

			b = new Buffer(bsz);
			b.fill(0);

			for (i = 0; i < sip.length; i++) {
				b.writeUInt32BE(sip[i], offset);
				offset += 4;
			}

			for (i = 0; i < eip.length; i++) {
				b.writeUInt32BE(eip[i], offset);
				offset += 4;
			}

			b.write(cc, offset);
			b.write(rg, offset + 2);
			b.writeInt32BE(lat, offset + 4);
			b.writeInt32BE(lon, offset + 8);
			b.write(city, offset + 12);
		} else {
			// IPv4
			bsz = 12;

			sip = parseInt(fields[0], 10);
			eip = parseInt(fields[1], 10);
			locId = parseInt(fields[2], 10);

			b = new Buffer(bsz);
			b.fill(0);
			b.writeUInt32BE(sip>>>0, 0);
			b.writeUInt32BE(eip>>>0, 4);
			b.writeUInt32BE(locId>>>0, 8);
		}

		fs.writeSync(datFile, b, 0, b.length, null);
	}

	var dataFile = path.join(dataPath, dest);
	var tmpDataFile = path.join(tmpPath, src);

	rimraf(dataFile);

	process.stdout.write('Processing Data (may take a moment) ...');
	var datFile = fs.openSync(dataFile, "w");

	var csvStream = new LineInputStream(fs.createReadStream(tmpDataFile), /[\r\n]+/);
	csvStream.setEncoding('utf8');

	csvStream.on('line', processLine);
	csvStream.on('end', function() {
		cb();
	});
}

function processCityDataNames(src, dest, cb) {
	function processLine(line, i, a) {
		if (line.match(/^Copyright/) || !line.match(/\d/)) {
			return;
		}

		var fields = line.replace(/"/g, '').split(/, */);
		var cc = fields[1];
		var rg = fields[2];
		var city = fields[3];
		var lat = Math.round(parseFloat(fields[5]) * 10000);
		var lon = Math.round(parseFloat(fields[6]) * 10000);
		var b;
		var sz = 32;

		b = new Buffer(sz);
		b.fill(0);
		b.write(cc, 0);
		b.write(rg, 2);
		b.writeInt32BE(lat, 4);
		b.writeInt32BE(lon, 8);
		b.write(city, 12);

		fs.writeSync(datFile, b, 0, b.length, null);
	}

	var dataFile = path.join(dataPath, dest);
	var tmpDataFile = path.join(tmpPath, src);

	rimraf(dataFile);

	var datFile = fs.openSync(dataFile, "w");

	var csvStream = new LineInputStream(fs.createReadStream(tmpDataFile), /[\r\n]+/);
	csvStream.setEncoding('utf8');

	csvStream.on('line', processLine);
	csvStream.on('end', function() {
		cb();
	});
}

function processData(type, src, dest, cb) {
	if (type === 'country') {
		processCountryData(src, dest, cb);
	} else if (type === 'city-extended') {
		processCityData(src[0], dest[0], function() {
			processCityDataNames(src[1], dest[1], function() {
				console.log(' DONE'.green);
				cb();
			});
		});
	} else {
		processCityData(src, dest, function() {
			console.log(' DONE'.green);
			cb();
		});
	}
}

function main() {
	
	rimraf(tmpPath);
	mkdir(tmpPath);
	
	async.forEachSeries(databases, function(database, nextDatabase) {
		fetch(database.url, function(tmpFile, tmpFileName) {
			extract(database.src, tmpFile, tmpFileName, function() {
				processData(database.type, database.src, database.dest, function() {
					console.log();
					nextDatabase();
				});
			});
		});
	}, function(err) {
		console.log();
	
		if (err) {
			console.log('Failed to Update Databases from MaxMind.'.red);
			process.exit();
		} else {
			console.log('Successfully Updated Databases from MaxMind.'.green);
		}
	});
}

if (require.main === module) {
	main();
} else {
	module.exports = main;
}