From d18f83d5e2d2ce853c0b377d570436f83e040e7a Mon Sep 17 00:00:00 2001
From: Alexandre Negrel <alexandre@negrel.dev>
Date: Wed, 28 Feb 2024 11:39:24 +0100
Subject: [PATCH] fix: properly clean pageview path

---
 pkg/event/pageview.go                     | 17 ++++++++--------
 tests/bun/events/events_pageviews.test.ts | 24 +++++++++++++++++++++++
 2 files changed, 33 insertions(+), 8 deletions(-)

diff --git a/pkg/event/pageview.go b/pkg/event/pageview.go
index 6067ca2..fd9a414 100644
--- a/pkg/event/pageview.go
+++ b/pkg/event/pageview.go
@@ -2,6 +2,7 @@ package event
 
 import (
 	"net/url"
+	"path"
 	"time"
 
 	"github.com/prismelabs/analytics/pkg/services/ipgeolocator"
@@ -22,10 +23,12 @@ type PageView struct {
 }
 
 // NewPageView creates a new PageView event.
-func NewPageView(pvUrl *url.URL,
+func NewPageView(
+	pvUrl *url.URL,
 	cli uaparser.Client,
 	pageReferrer string,
-	countryCode ipgeolocator.CountryCode) (PageView, error) {
+	countryCode ipgeolocator.CountryCode,
+) (PageView, error) {
 	domain, err := ParseDomainName(pvUrl.Hostname())
 	if err != nil {
 		return PageView{}, err
@@ -36,17 +39,15 @@ func NewPageView(pvUrl *url.URL,
 		return PageView{}, err
 	}
 
-	path := pvUrl.Path
-	if path == "" {
-		path = "/"
-	} else if path[len(path)-1] == '/' && len(path) > 1 {
-		path = path[:len(path)-1]
+	pageviewPath := pvUrl.Path
+	if pageviewPath == "" {
+		pageviewPath = "/"
 	}
 
 	return PageView{
 		Timestamp:      time.Now().UTC(),
 		DomainName:     domain,
-		PathName:       path,
+		PathName:       path.Clean(pageviewPath),
 		Client:         cli,
 		ReferrerDomain: referrerDomain,
 		CountryCode:    countryCode,
diff --git a/tests/bun/events/events_pageviews.test.ts b/tests/bun/events/events_pageviews.test.ts
index ad1179b..ef3d1d1 100644
--- a/tests/bun/events/events_pageviews.test.ts
+++ b/tests/bun/events/events_pageviews.test.ts
@@ -205,6 +205,30 @@ test('valid pageview with US IP address', async () => {
   })
 })
 
+test('valid pageview with dirty path', async () => {
+  const response = await fetch(PRISME_PAGEVIEWS_URL, {
+    method: 'POST',
+    headers: {
+      'X-Forwarded-For': '8.8.8.8', // Google public DNS
+      Referer: 'http://foo.mywebsite.localhost///another/../another/foo?bar=baz#qux'
+    }
+  })
+  expect(response.status).toBe(200)
+
+  const data = await getLatestPageview()
+
+  expect(data).toMatchObject({
+    timestamp: expect.stringMatching(TIMESTAMP_REGEX),
+    domain: 'foo.mywebsite.localhost',
+    path: '/another/foo',
+    operating_system: 'Other',
+    browser_family: 'Other',
+    device: 'Other',
+    referrer_domain: 'direct',
+    country_code: 'US'
+  })
+})
+
 async function getLatestPageview (): Promise<any> {
   // Wait for clickhouse to ingest batch.
   Bun.sleepSync(1000)