[web] add LRU cache for geoip details (#34521)

GitOrigin-RevId: 0c96b53228b3e4fd764ac09e029db8b3f2b45de7

services/web/app/src/infrastructure/GeoIpLookup.mjs

@@ -3,12 +3,26 @@
 import settings from '@overleaf/settings'
 import logger from '@overleaf/logger'
 import { fetchJson } from '@overleaf/fetch-utils'
+import LRU from 'lru-cache'
 
 /**
  * @typedef {import('../../../types/subscription/currency').CurrencyCode} CurrencyCode
  */
 
 const DEFAULT_CURRENCY_CODE = /** @type {const} */ 'USD'
+const cache = new LRU({
+  max: settings.apis.geoIpLookup.cacheSize,
+})
+
+/**
+ * Cache details per /24 subnet, which is the smallest subnet routed on the public internet.
+ * IPv6 is not supported by GCP. We could cache by /48.
+ * @param {string} ip
+ */
+function networkCacheKey(ip) {
+  const octets = ip.split('.')
+  return octets.length === 4 ? octets.slice(0, 3).join('.') : ip
+}
 
 /** @type {Record<string, CurrencyCode>} */
 const currencyMappings = {
@@ -99,10 +113,17 @@ async function getDetails(ip, callback) {
     return
   }
   ip = ip.trim().split(' ')[0]
+  const cacheKey = networkCacheKey(ip)
+  const cached = cache.get(cacheKey)
+  if (cached) {
+    return cached
+  }
   const url = new URL(settings.apis.geoIpLookup.url)
   url.pathname += ip
   logger.debug({ ip, url }, 'getting geo ip details')
-  return await fetchJson(url, { signal: AbortSignal.timeout(1_000) })
+  const details = await fetchJson(url, { signal: AbortSignal.timeout(1_000) })
+  cache.set(cacheKey, details)
+  return details
 }
 
 /**

services/web/config/settings.defaults.js

@@ -232,6 +232,9 @@ module.exports = {
         '127.0.0.1'
       }:3003`,
     },
+    geoIpLookup: {
+      cacheSize: intFromEnv('GEO_IP_LOOKUP_CACHE_SIZE', 10_000),
+    },
     docstore: {
       url: `http://${process.env.DOCSTORE_HOST || '127.0.0.1'}:3016`,
       pubUrl: `http://${process.env.DOCSTORE_HOST || '127.0.0.1'}:3016`,

services/web/test/unit/src/infrastructure/GeoIpLookup.test.mjs

@@ -31,6 +31,7 @@ describe('GeoIpLookup', function () {
       apis: {
         geoIpLookup: {
           url: 'http://lookup.com/',
+          cacheSize: 10_000,
         },
       },
     }
@@ -87,6 +88,20 @@ describe('GeoIpLookup', function () {
           new URL(ctx.settings.apis.geoIpLookup.url + ctx.ipAddress)
         )
       })
+
+      it('should cache lookups by /24 network', async function (ctx) {
+        const first = await ctx.GeoIpLookup.promises.getDetails('12.34.56.78')
+        const second = await ctx.GeoIpLookup.promises.getDetails('12.34.56.99')
+        ctx.fetchUtils.fetchJson.should.have.been.calledOnce
+        assert.equal(first, second)
+        assert.deepEqual(second, ctx.stubbedResponse)
+      })
+
+      it('should not share cache entries across different /24 networks', async function (ctx) {
+        await ctx.GeoIpLookup.promises.getDetails('12.34.56.78')
+        await ctx.GeoIpLookup.promises.getDetails('12.34.99.1')
+        ctx.fetchUtils.fetchJson.should.have.been.calledTwice
+      })
     })
   })