More speed and less memory usage via code generation

.gitignore

@@ -1,3 +1,6 @@
 mongodb
 node_modules
 .DS_Store
+.idea
+*.iml
+generated-*.js

README.markdown

@@ -17,28 +17,40 @@ Benchmarks on my 2011 Macbook Air whilst running lots of software.  The test too
 
 1. Comes with the [standard CSV database by MaxMind](http://www.maxmind.com/app/geolite) which may require updating.
 
-## How to use
+
+## Install
+
+Simple run
+
+    npm install geoip-native
+
+## Install from source
 1. git clone https://github.com/benlowry/node-geoip-native
 2. cd node-geoip-native
-3. node test.js
-
-or just ```npm install geoip-native```
+3. npm install
+4. node benchmark.js
 
 ## Methods
 
 Node GeoIP Native provides methods for:
 
 1. ```lookup``` performs the lookup, takes the ip address as a parameter
 
-## Examples
+## Testing
+First, you have to install nodeunit (https://github.com/caolan/nodeunit)
+
+	$> npm install nodeunit
+
+Second, run the unit test:
 
-	var geoip = require("geoip-native");
-	var ip = "123.123.123.123";
-	geoip.lookup(ip);
-	console.log("country: " + ip.name + " / " + ip.code);
+	$> ./node_modules/nodeunit/bin/nodeunit test/unit/lookupTest.js
+
+## Examples
 
-	// in practice you'd want:
-	// ip = request.headers["x-forwarded-for"] || request.connection.remoteAddress,
+    var geoip = require("geoip-native");
+    var ip = geoip.lookup("134.12.12.123");
+    console.log("numeric ip value: " + ip.ipstart);
+    console.log("country: " + ip.name + " / " + ip.code);
 
 ### What's missing
 Be neat to expand this to include cities.

generate_sources.js

@@ -0,0 +1,124 @@
+"use strict";
+
+var countries = [];
+var countryNamesAndCodes = [];
+
+function read_csv_file_and_prepare_data() {
+
+  function load_CSV_file() {
+    var fs = require("fs");
+    var data = fs.readFileSync(__dirname + "/GeoIPCountryWhois.csv")
+    var buffer = "";
+    buffer += data.toString();
+    return buffer;
+  }
+
+  function extractPartsFromCsv(line) {
+    var matches = line.match(/("(?:[^"]|"")*"|[^,]*)/g);
+    var result = [];
+    for (var i = 0; i < matches.length; i++) {
+      var part = matches[i].replace(/"/g, "").trim();
+      if (part.length > 0) {
+        result.push(part);
+      }
+    }
+    return result;
+  }
+
+  var entries = load_CSV_file().split("\n");
+  var offsetCounter = 0;
+  var countryIndex = 0;
+  var countrySet = {};
+  var lastIpRangeEnd = 0;
+
+  for (var i = 0; i < entries.length; i++) {
+    var parts = extractPartsFromCsv(entries[i]);
+    if (parts.length > 5) {
+      var countryName = parts[5].trim();
+      var countryCode = parts[4];
+      if (!countrySet[countryName]) {
+        countryIndex = offsetCounter++;
+        countrySet[countryName] = {index: countryIndex};
+        countryNamesAndCodes.push(countryName);
+        countryNamesAndCodes.push(countryCode);
+      } else {
+        countryIndex = countrySet[countryName].index;
+      }
+      countries.push(createCountryInformation(parseInt(parts[2]), parseInt(parts[3]), countryCode, countryName, (countryIndex * 2)));
+      lastIpRangeEnd = parseInt(parts[3]);
+    }
+  }
+
+  // add a special country, which indicates the END
+  countryIndex = offsetCounter;
+  countrySet["UNKNOWN"] = {index: countryIndex};
+  countryNamesAndCodes.push("UNKNOWN");
+  countryNamesAndCodes.push("N/A");
+  countries.push(
+     createCountryInformation(lastIpRangeEnd + 1, lastIpRangeEnd + 1, "N/A", "UNKNOWN", countryIndex * 2)
+  );
+
+  countries.sort(function (a, b) {
+    return a.ipstart - b.ipstart;
+  });
+
+}
+
+function createCountryInformation(ipStart, ipEnd, countryCode, countryName, index) {
+  return {
+    ipstart: ipStart,
+    ipend: ipEnd,
+    code: countryCode,
+    name: countryName,
+    index: index
+  };
+}
+
+
+function write_sourceFile_countryNamesAndCodes() {
+  var data = "";
+  data += "// AUTOGENERATED CODE - DO NOT MODIFY!   " + new Date() + " \n";
+  data += "var countryNamesAndCodes = [\n";
+  for (var i = 0, len = countryNamesAndCodes.length >> 1; i < len; i++) {
+    var countryName = countryNamesAndCodes[i << 1].replace("'", "\\'");
+    var countryCode = countryNamesAndCodes[(i << 1) + 1].replace("'", "\\'");
+    data += "'" + countryName + "','" + countryCode + "',\n";
+  }
+  data += "];\n";
+  data += "module.exports = {\n";
+  data += "   countryNamesAndCodes:countryNamesAndCodes\n";
+  data += "};\n";
+
+  var fs = require("fs");
+  fs.writeFileSync('generated-namesandcodes.js', data, 'utf8');
+}
+
+function write_sourceFile_countries() {
+  var data = "";
+  data += "// AUTOGENERATED CODE - DO NOT MODIFY!   " + new Date() + " \n";
+  data += "var countries = [\n";
+  for (var i = 0, len = countries.length; i < len; i++) {
+    var country = countries[i];
+    data += "{s:" + country.ipstart + ",";
+    data += "e:" + country.ipend + ",";
+    data += "i:" + country.index + "},";
+    data += "//" + countryNamesAndCodes[country.index + 1] + "\n";
+  }
+  data += "];\n";
+  data += "module.exports = {\n";
+  data += "   countries:countries\n";
+  data += "};\n";
+
+  var fs = require("fs");
+  fs.writeFileSync('generated-countries.js', data, 'utf8');
+}
+
+/**
+ * Prepare the data.  This uses the standard free GeoIP CSV database
+ * from MaxMind, you should be able to update it at any time by just
+ * overwriting GeoIPCountryWhois.csv with a new version.
+ */
+
+read_csv_file_and_prepare_data()
+write_sourceFile_countryNamesAndCodes();
+write_sourceFile_countries();

geoip-native.js

@@ -0,0 +1,65 @@
+var countryNamesAndCodes = require('./generated-namesandcodes.js').countryNamesAndCodes;
+var countries = require('./generated-countries.js').countries;
+var countriesLength = countries.length;
+
+module.exports = geoip = {
+  lookup: function (ip) {
+    return _lookup(ip);
+  }
+};
+
+/**
+ * @param ip the ip we're looking for
+ * @return {*}
+ * @see http://en.wikipedia.org/wiki/Binary_search_algorithm   (Deferred detection of equality approach)
+ */
+function _lookup(ip) {
+
+  var parts = ip.split(".");
+  var target_ip = parseInt(parts[3], 10) +
+     (parseInt(parts[2], 10) * 256) +
+     (parseInt(parts[1], 10) * 65536) +
+     (parseInt(parts[0], 10) * 16777216);
+
+  var idxMin = 0;
+  var idxMiddle = 0;
+  var idxMax = countriesLength - 1;
+
+  while (idxMin < idxMax) {
+    idxMiddle = (idxMax + idxMin) >> 1;
+    if (!(idxMiddle < idxMax)) {
+      throw "assertion error: idxMiddle is not lower then idxMax"
+    }
+    if (countries[idxMiddle].s < target_ip) {
+      idxMin = idxMiddle + 1;
+    } else {
+      idxMax = idxMiddle;
+    }
+  }
+
+  var pickedCountry = countries[idxMin];
+  if ((idxMax == idxMin) && (pickedCountry.s == target_ip)) {
+    pickedCountry = countries[idxMin];
+    return createCountry(pickedCountry.s, pickedCountry.e, countryNamesAndCodes[pickedCountry.i], countryNamesAndCodes[pickedCountry.i + 1]);
+  }
+
+  if ((idxMiddle > 0) && (countries[idxMiddle - 1].s < target_ip) && (target_ip < countries[idxMiddle].s)) {
+    pickedCountry = countries[idxMiddle - 1]
+    return createCountry(pickedCountry.s, pickedCountry.e, countryNamesAndCodes[pickedCountry.i], countryNamesAndCodes[pickedCountry.i + 1]);
+  }
+
+  if ((idxMiddle < idxMax) && (countries[idxMiddle].s < target_ip) && (target_ip < countries[idxMiddle + 1].s)) {
+    pickedCountry = countries[idxMiddle]
+    return createCountry(pickedCountry.s, pickedCountry.e, countryNamesAndCodes[pickedCountry.i], countryNamesAndCodes[pickedCountry.i + 1]);
+  }
+  return createCountry(target_ip, target_ip, "UNKNOWN", "N/A");
+}
+
+function createCountry(ipstart, ipend, countryName, countryCode) {
+  return {
+    ipstart: ipstart,
+    ipend: ipend,
+    name: countryName,
+    code: countryCode
+  };
+}