fix broken bot filtering

prismelabs · Jun 13, 2024 · 8b9bef2 · 8b9bef2
1 parent f08ca3d
commit 8b9bef2
Show file tree

Hide file tree

Showing 4 changed files with 22 additions and 3 deletions.
diff --git a/pkg/handlers/events_pageviews.go b/pkg/handlers/events_pageviews.go
@@ -54,6 +54,12 @@ func ProvidePostEventsPageViews(
 
 		newSession := !equalBytes(referrerUri.Host(), pageView.PageUri.Host())
 		if newSession {
+			// Filter bot.
+			client := uaParserService.ParseUserAgent(utils.UnsafeString(userAgent))
+			if client.IsBot {
+				return fiber.NewError(fiber.StatusBadRequest, "bot session filtered")
+			}
+
 			sessionUuid, err := uuid.NewV7()
 			if err != nil {
 				return fmt.Errorf("failed to generate session uuid: %w", err)
@@ -62,7 +68,7 @@ func ProvidePostEventsPageViews(
 			pageView.Session = event.Session{
 				PageUri:     &pageView.PageUri,
 				ReferrerUri: &referrerUri,
-				Client:      uaParserService.ParseUserAgent(utils.UnsafeString(userAgent)),
+				Client:      client,
 				CountryCode: ipGeolocatorService.FindCountryCodeForIP(c.IP()),
 				VisitorId:   visitorId,
 				SessionUuid: sessionUuid,

diff --git a/pkg/services/uaparser/client.go b/pkg/services/uaparser/client.go
@@ -14,5 +14,6 @@ type Client struct {
 func (c Client) MarshalZerologObject(e *zerolog.Event) {
 	e.Str("browser_family", c.BrowserFamily).
 		Str("operating_system", c.OperatingSystem).
-		Str("device", c.Device)
+		Str("device", c.Device).
+		Bool("is_bot", c.IsBot)
 }
diff --git a/tests/bun/events/events_pageviews.test.ts b/tests/bun/events/events_pageviews.test.ts
@@ -82,6 +82,19 @@ test('internal traffic with no session associated is rejected', async () => {
   expect(response.status).toBe(400)
 })
 
+test('robot user agent is rejected', async () => {
+  const response = await fetch(PRISME_PAGEVIEWS_URL, {
+    method: 'POST',
+    headers: {
+      Origin: 'https://mywebsite.localhost',
+      'X-Forwarded-For': faker.internet.ip(),
+      Referer: 'https://mywebsite.localhost/foo?bar=baz#qux',
+      'User-Agent': 'Googlebot'
+    }
+  })
+  expect(response.status).toBe(400)
+})
+
 test('registered domain in Origin header and valid referrer is accepted', async () => {
   const response = await fetch(PRISME_PAGEVIEWS_URL, {
     method: 'POST',

diff --git a/tests/bun/trusted-proxy/metrics.test.ts b/tests/bun/trusted-proxy/metrics.test.ts
@@ -55,7 +55,6 @@ async function fetchMetrics (): Promise<string> {
 }
 
 async function parseMetrics (prometheusMetrics: string): Promise<Metrics> {
-  console.log(prometheusMetrics)
   const metrics: Metrics = {
     counter: {},
     summary: {},