feat: add probe stats scripts

.gitignore

@@ -10,3 +10,7 @@ data/IP_BLACKLIST.json
 data/AWS_IP_RANGES.json
 data/GCP_IP_RANGES.json
 .eslintcache
+probes-stats/known-result.csv
+probes-stats/known-result.json
+probes-stats/all-result.csv
+probes-stats/all-result.json

package.json

@@ -105,6 +105,8 @@
 		"clean": "rimraf coverage",
 		"download:jsons": "tsx src/lib/download-jsons.ts",
 		"prepare": "npm run download:jsons; husky install || echo 'Failed to install husky'",
+		"stats:known": "NEW_RELIC_ENABLED=false NEW_RELIC_LOG_ENABLED=false tsx probes-stats/known.ts",
+		"stats:all": "NEW_RELIC_ENABLED=false NEW_RELIC_LOG_ENABLED=false tsx probes-stats/all.ts",
 		"start": "NODE_ENV=production node --max_old_space_size=3584 --max-semi-space-size=128 --experimental-loader newrelic/esm-loader.mjs dist/index.js",
 		"start:dev": "NODE_ENV=development FAKE_PROBE_IP=1 NEW_RELIC_ENABLED=false tsx src/index.ts",
 		"start:test": "NODE_ENV=test FAKE_PROBE_IP=1 NEW_RELIC_ENABLED=false tsx src/index.ts",

probes-stats/all.ts

@@ -0,0 +1,73 @@
+/**
+ * How to run:
+ * 1. Run redis instance
+ * 2. Run with fulfilled keys: `ADMIN_KEY= IPINFO_API_KEY= MAXMIND_ACCOUNT_ID= MAXMIND_LICENSE_KEY= npm run stats:all`
+ */
+
+import got from 'got';
+import fs from 'node:fs';
+import { getRedisClient, initRedis } from '../src/lib/redis/client.js';
+import { LocationInfo, createGeoipClient } from '../src/lib/geoip/client.js';
+import { FastlyBundledResponse, fastlyLookup } from '../src/lib/geoip/providers/fastly.js';
+import { ipinfoLookup } from '../src/lib/geoip/providers/ipinfo.js';
+import { maxmindLookup } from '../src/lib/geoip/providers/maxmind.js';
+import { populateMemList as populateIpWhiteList } from '../src/lib/geoip/whitelist.js';
+
+await populateIpWhiteList();
+await initRedis();
+const geoIpClient = createGeoipClient();
+
+type ResultItem = {
+	ip: string,
+	ipinfo: string,
+	maxmind: string,
+	fastly: string,
+	algorithm: string,
+};
+
+const getData = async () => {
+	const result: ResultItem[] = [];
+	const probes = await got(`https://api.globalping.io/v1/probes?adminkey=${process.env['ADMIN_KEY']}`).json<{ipAddress: string}[]>();
+
+	await Promise.all(probes.map(async (probe) => {
+		const ip = probe.ipAddress;
+
+		const ipinfo = await geoIpClient.lookupWithCache<LocationInfo>(`geoip:ipinfo:${ip}`, async () => ipinfoLookup(ip)).catch(err => ({ normalizedCity: err.message.toUpperCase() }));
+		const maxmind = await geoIpClient.lookupWithCache<LocationInfo>(`geoip:maxmind:${ip}`, async () => maxmindLookup(ip)).catch(err => ({ normalizedCity: err.message.toUpperCase() }));
+		const fastly = await geoIpClient.lookupWithCache<FastlyBundledResponse>(`geoip:fastly:${ip}`, async () => fastlyLookup(ip)).catch(err => ({ location: { normalizedCity: err.message.toUpperCase() } }));
+
+		const updatedRow = {
+			ip,
+			ipinfo: ipinfo.normalizedCity,
+			maxmind: maxmind.normalizedCity,
+			fastly: fastly.location.normalizedCity,
+		} as ResultItem;
+
+		try {
+			const location = await geoIpClient.lookup(ip);
+			updatedRow['algorithm'] = location.normalizedCity;
+		} catch (err) {
+			updatedRow['algorithm'] = err.message.toUpperCase();
+		}
+
+		result.push(updatedRow);
+	}));
+
+	return result;
+};
+
+const generateFiles = (data: ResultItem[]) => {
+	const csvContent = [
+		'ip,ipinfo,maxmind,fastly,algorithm',
+		...data.map(row => `${row.ip},${row.ipinfo},${row.maxmind},${row.fastly},${row.algorithm}`),
+	].join('\n');
+	fs.writeFileSync('./probes-stats/all-result.json', JSON.stringify(data, null, 2));
+	fs.writeFileSync('./probes-stats/all-result.csv', csvContent);
+};
+
+(async () => {
+	const data = await getData();
+	generateFiles(data);
+	const redis = getRedisClient();
+	await redis.disconnect();
+})();

probes-stats/known-probes.json

@@ -0,0 +1,8 @@
+[
+	{
+		"ip": "exampleIp",
+		"city": "exampleCity",
+		"country": "exampleCountry",
+		"note": "exampleNote"
+	}
+]

probes-stats/known.ts

@@ -0,0 +1,93 @@
+/**
+ * How to run:
+ * 1. Run redis instance
+ * 2. Replace content of the known-probes.json with the known probes info in the same format
+ * 3. Run with fulfilled keys: `IPINFO_API_KEY= MAXMIND_ACCOUNT_ID= MAXMIND_LICENSE_KEY= npm run stats:known`
+ */
+
+import fs from 'node:fs';
+import { getRedisClient, initRedis } from '../src/lib/redis/client.js';
+import { LocationInfo, createGeoipClient } from '../src/lib/geoip/client.js';
+import { normalizeCityName } from '../src/lib/geoip/utils.js';
+import { FastlyBundledResponse, fastlyLookup } from '../src/lib/geoip/providers/fastly.js';
+import { ipinfoLookup } from '../src/lib/geoip/providers/ipinfo.js';
+import { maxmindLookup } from '../src/lib/geoip/providers/maxmind.js';
+import { populateMemList as populateIpWhiteList } from '../src/lib/geoip/whitelist.js';
+import sheet from './known-probes.json' assert { type: 'json' };
+
+await populateIpWhiteList();
+await initRedis();
+const geoIpClient = createGeoipClient();
+
+type ResultItem = {
+	ip: string,
+	city: string,
+	ipinfo: string,
+	maxmind: string,
+	fastly: string,
+	algorithm: string,
+	country: string,
+	note: string,
+};
+
+const getData = async () => {
+	const result: ResultItem[] = [];
+
+	await Promise.all(sheet.map(async (row) => {
+		const normalizedCity = normalizeCityName(row.city);
+
+		const ipinfo = await geoIpClient.lookupWithCache<LocationInfo>(`geoip:ipinfo:${row.ip}`, async () => ipinfoLookup(row.ip));
+		const maxmind = await geoIpClient.lookupWithCache<LocationInfo>(`geoip:maxmind:${row.ip}`, async () => maxmindLookup(row.ip));
+		const fastly = await geoIpClient.lookupWithCache<FastlyBundledResponse>(`geoip:fastly:${row.ip}`, async () => fastlyLookup(row.ip));
+
+		const updatedRow = {
+			...row,
+			city: normalizedCity,
+			ipinfo: ipinfo.normalizedCity,
+			maxmind: maxmind.normalizedCity,
+			fastly: fastly.location.normalizedCity,
+		} as ResultItem;
+
+		try {
+			const location = await geoIpClient.lookup(row.ip);
+			updatedRow['algorithm'] = location.normalizedCity;
+		} catch (err) {
+			updatedRow['algorithm'] = err.message.toUpperCase();
+		}
+
+		result.push(updatedRow);
+	}));
+
+	return result;
+};
+
+const getAccuracy = (result) => {
+	const ipinfoTrueCount = result.filter(row => row.ipinfo === row.city).length;
+	const maxmindTrueCount = result.filter(row => row.maxmind === row.city).length;
+	const fastlyTrueCount = result.filter(row => row.fastly === row.city).length;
+	const algorithmTrueCount = result.filter(row => row.algorithm === row.city).length;
+	return {
+		ipinfo: (ipinfoTrueCount / result.length).toFixed(2),
+		maxmind: (maxmindTrueCount / result.length).toFixed(2),
+		fastly: (fastlyTrueCount / result.length).toFixed(2),
+		algorithm: (algorithmTrueCount / result.length).toFixed(2),
+	};
+};
+
+const generateFiles = (data: ResultItem[], acc: {ipinfo: string, maxmind: string, fastly: string, algorithm: string}) => {
+	const csvContent = [
+		'ip,real city,ipinfo,maxmind,fastly,algorithm',
+		`accuracy:,1,${acc.ipinfo},${acc.maxmind},${acc.fastly},${acc.algorithm}`,
+		...data.map(row => `${row.ip},${row.city},${row.ipinfo},${row.maxmind},${row.fastly},${row.algorithm}`),
+	].join('\n');
+	fs.writeFileSync('./probes-stats/known-result.json', JSON.stringify(data, null, 2));
+	fs.writeFileSync('./probes-stats/known-result.csv', csvContent);
+};
+
+(async () => {
+	const data = await getData();
+	const acc = getAccuracy(data);
+	generateFiles(data, acc);
+	const redis = getRedisClient();
+	await redis.disconnect();
+})();

src/lib/geoip/client.ts

@@ -162,7 +162,7 @@ export default class GeoipClient {
 		return _.omit(best, 'provider');
 	}
 
-	private async lookupWithCache<T> (key: string, fn: () => Promise<T>): Promise<T> {
+	public async lookupWithCache<T> (key: string, fn: () => Promise<T>): Promise<T> {
 		const cached = await this.cache.get<T>(key);
 
 		if (cached) {