feat: use ip2location for hosting detection, allow manual overrides

Author: MartinKolarik

Diff for: src/lib/geoip/client.ts

@@ -14,6 +14,7 @@ import { fastlyLookup } from './providers/fastly.js';
 import { maxmindLookup } from './providers/maxmind.js';
 import { ipmapLookup } from './providers/ipmap.js';
 import { type Ip2LocationBundledResponse, ip2LocationLookup } from './providers/ip2location.js';
+import { isHostingOverrides } from './overrides.js';
 
 export type LocationInfo = Omit<ProbeLocation, 'region'>;
 type Provider = 'ipmap' | 'ip2location' | 'ipinfo' | 'maxmind' | 'fastly';
@@ -42,7 +43,7 @@ export default class GeoipClient {
 				this.lookupWithCache<LocationInfo>(`geoip:fastly:${addr}`, async () => fastlyLookup(addr)),
 			])
 			.then(([ ipinfo, ip2location, maxmind, ipmap, fastly ]) => {
-				isHosting = ipinfo.status === 'fulfilled' ? ipinfo.value.isHosting : undefined;
+				isHosting = ip2location.status === 'fulfilled' ? ip2location.value.isHosting : undefined;
 				const fulfilled: (LocationInfoWithProvider | null)[] = [];
 
 				// Providers here are pushed in a desc prioritized order
@@ -74,6 +75,13 @@ export default class GeoipClient {
 			throw new ProbeError(`unresolvable geoip: ${addr}`);
 		}
 
+		for (const override of isHostingOverrides) {
+			if (override.normalizedNetwork.test(networkMatch.normalizedNetwork)) {
+				isHosting = override.isHosting;
+				break;
+			}
+		}
+
 		const region = getRegionByCountry(match.country);
 
 		return {

Diff for: src/lib/geoip/overrides.ts

@@ -0,0 +1,12 @@
+export const isHostingOverrides = [
+	// https://baxetgroup.com/
+	{
+		normalizedNetwork: /baxet group/,
+		isHosting: true,
+	},
+	// https://www.psychz.net/
+	{
+		normalizedNetwork: /psychz networks/,
+		isHosting: true,
+	},
+];

Diff for: src/lib/geoip/providers/ip2location.ts

@@ -25,13 +25,17 @@ type Ip2LocationResponse = {
 	longitude?: number;
 	as?: string;
 	is_proxy?: boolean;
+	usage_type?: string;
 };
 
 export type Ip2LocationBundledResponse = {
 	location: LocationInfo,
+	isHosting: boolean | undefined,
 	isProxy: boolean,
 };
 
+const HOSTING_USAGE_TYPES = [ 'CDN', 'DCH' ];
+
 export const ip2LocationLookup = async (addr: string): Promise<Ip2LocationBundledResponse> => {
 	const result = await got(`https://api.ip2location.io`, {
 		searchParams: {
@@ -58,6 +62,7 @@ export const ip2LocationLookup = async (addr: string): Promise<Ip2LocationBundle
 
 	return {
 		location,
+		isHosting: result.usage_type ? HOSTING_USAGE_TYPES.includes(result.usage_type) : undefined,
 		isProxy: result.is_proxy ?? false,
 	};
 };

Diff for: test/mocks/nock-geoip.json

@@ -8,7 +8,8 @@
 			"longitude": -96.8067,
 			"asn": "20001",
 			"as": "The Constant Company LLC",
-			"is_proxy": false
+			"is_proxy": false,
+			"usage_type": "DCH"
 		},
 		"argentina": {
 			"country_code": "AR",
@@ -28,7 +29,8 @@
 			"longitude": -74.0060,
 			"asn": "80001",
 			"as": "The Constant Company LLC",
-			"is_proxy": false
+			"is_proxy": false,
+			"usage_type": "CDN"
 		},
 		"washington": {
 			"country_code": "US",
@@ -37,7 +39,8 @@
 			"latitude": 38.89539,
 			"longitude": -77.039476,
 			"asn": "40676",
-			"as": "Psychz Networks"
+			"as": "Psychz Networks",
+			"usage_type": "ISP"
 		},
 		"falkenstein": {
 			"country_code": "DE",
@@ -102,7 +105,8 @@
 			"longitude": -96.8067,
 			"asn": "20001",
 			"as": "The Constant Company LLC",
-			"is_proxy": true
+			"is_proxy": true,
+			"usage_type": "DCH"
 		},
 		"noVpn": {
 			"country_code": "US",
@@ -111,7 +115,8 @@
 			"latitude": 32.7831,
 			"longitude": -96.8067,
 			"asn": "20001",
-			"as": "The Constant Company LLC"
+			"as": "The Constant Company LLC",
+			"usage_type": "DCH"
 		}
 	},
 	"ipmap": {

Diff for: test/tests/unit/geoip/client.test.ts

@@ -161,7 +161,7 @@ describe('geoip service', () => {
 			longitude: -58.3772,
 			network: 'InterBS S.R.L. (BAEHOST)',
 			normalizedNetwork: 'interbs s.r.l. (baehost)',
-			isHosting: undefined,
+			isHosting: true,
 		});
 	});
 
@@ -427,7 +427,7 @@ describe('geoip service', () => {
 				longitude: -77.039476,
 				network: 'Psychz Networks',
 				normalizedNetwork: 'psychz networks',
-				isHosting: undefined,
+				isHosting: true,
 			});
 
 			nockGeoIpProviders({ ip2location: 'empty', ipmap: 'empty', maxmind: 'empty', ipinfo: 'washington', fastly: 'empty' });
@@ -446,7 +446,7 @@ describe('geoip service', () => {
 				longitude: -77.039476,
 				network: 'Verizon Business',
 				normalizedNetwork: 'verizon business',
-				isHosting: false,
+				isHosting: undefined,
 			});
 		});