Set exit_page only on pageviews (#3870)

* Set exit_page only on pageviews

* Update tests

* Update entry_page on first pageview

* Update CHANGELOG.md

CHANGELOG.md

@@ -41,6 +41,7 @@ All notable changes to this project will be documented in this file.
 - Remove `DISABLE_AUTH` deprecation warning plausible/analytics#3904
 
 ### Changed
+- A visits `entry_page` and `exit_page` is only set and updated for pageviews, not custom events
 - Limit the number of Goal Conversions shown on the dashboard and render a "Details" link when there are more entries to show
 - Show Outbound Links / File Downloads / 404 Pages / Cloaked Links instead of Goal Conversions when filtering by the corresponding goal
 - Require custom properties to be explicitly added from Site Settings > Custom Properties in order for them to show up on the dashboard

lib/plausible/session/cache_store.ex

@@ -42,7 +42,12 @@ defmodule Plausible.Session.CacheStore do
       session
       | user_id: event.user_id,
         timestamp: event.timestamp,
-        exit_page: event.pathname,
+        entry_page:
+          if(session.entry_page == "" and event.name == "pageview",
+            do: event.pathname,
+            else: session.entry_page
+          ),
+        exit_page: if(event.name == "pageview", do: event.pathname, else: session.exit_page),
         is_bounce: false,
         duration: Timex.diff(event.timestamp, session.start, :second) |> abs,
         pageviews:
@@ -58,8 +63,8 @@ defmodule Plausible.Session.CacheStore do
       hostname: event.hostname,
       site_id: event.site_id,
       user_id: event.user_id,
-      entry_page: event.pathname,
-      exit_page: event.pathname,
+      entry_page: if(event.name == "pageview", do: event.pathname, else: ""),
+      exit_page: if(event.name == "pageview", do: event.pathname, else: ""),
       is_bounce: true,
       duration: 0,
       pageviews: if(event.name == "pageview", do: 1, else: 0),

test/plausible/session/cache_store_test.exs

@@ -89,6 +89,69 @@ defmodule Plausible.Session.CacheStoreTest do
     assert session.events == 2
   end
 
+  test "initial pageview-specific attributes" do
+    site_id = new_site_id()
+
+    event =
+      build(:event,
+        name: "custom_event",
+        site_id: site_id,
+        pathname: "/path/1",
+        user_id: 1
+      )
+
+    flush([event])
+
+    session = get_session(site_id)
+
+    assert session.exit_page == ""
+    assert session.events == 1
+    assert session.pageviews == 0
+  end
+
+  test "updating pageview-specific attributes" do
+    site_id = new_site_id()
+
+    event1 =
+      build(:event,
+        name: "custom_event",
+        site_id: site_id,
+        pathname: "/path/1",
+        user_id: 1
+      )
+
+    event2 =
+      build(:pageview,
+        pathname: "/path/2",
+        site_id: site_id,
+        user_id: 1
+      )
+
+    event3 =
+      build(:pageview,
+        pathname: "/path/3",
+        site_id: site_id,
+        user_id: 1
+      )
+
+    event4 =
+      build(:event,
+        name: "custom_event",
+        site_id: site_id,
+        pathname: "/path/4",
+        user_id: 1
+      )
+
+    flush([event1, event2, event3, event4])
+
+    session = get_session(site_id)
+
+    assert session.entry_page == "/path/2"
+    assert session.exit_page == "/path/3"
+    assert session.events == 4
+    assert session.pageviews == 2
+  end
+
   test "calculates duration correctly for out-of-order events", %{buffer: buffer} do
     timestamp = Timex.now()
     event1 = build(:event, name: "pageview", timestamp: timestamp |> Timex.shift(seconds: 10))
@@ -103,18 +166,6 @@ defmodule Plausible.Session.CacheStoreTest do
   end
 
   describe "collapse order" do
-    defp new_site_id() do
-      [[site_id]] =
-        Plausible.ClickhouseRepo.query!("select max(site_id) + rand() from sessions_v2 FINAL").rows
-
-      site_id
-    end
-
-    defp flush(events) do
-      for e <- events, do: CacheStore.on_event(e, %{}, nil)
-      Plausible.Session.WriteBuffer.flush()
-    end
-
     test "across parts" do
       e = build(:event, name: "pageview", site_id: new_site_id())
 
@@ -137,8 +188,7 @@ defmodule Plausible.Session.CacheStoreTest do
         %{e | pathname: "/exit"}
       ])
 
-      session_q = from s in Plausible.ClickhouseSessionV2, where: s.site_id == ^e.site_id
-      session = Plausible.ClickhouseRepo.one!(session_q)
+      session = get_session(e.site_id)
 
       refute session.is_bounce
       assert session.entry_page == "/"
@@ -167,4 +217,21 @@ defmodule Plausible.Session.CacheStoreTest do
       assert session.events == 4
     end
   end
+
+  defp new_site_id() do
+    [[site_id]] =
+      Plausible.ClickhouseRepo.query!("select max(site_id) + rand() from sessions_v2 FINAL").rows
+
+    site_id
+  end
+
+  defp flush(events) do
+    for e <- events, do: CacheStore.on_event(e, %{}, nil)
+    Plausible.Session.WriteBuffer.flush()
+  end
+
+  defp get_session(site_id) do
+    session_q = from s in Plausible.ClickhouseSessionV2, where: s.site_id == ^site_id
+    Plausible.ClickhouseRepo.one!(session_q)
+  end
 end

test/plausible_web/controllers/CSVs/30d-filter-goal/entry_pages.csv

@@ -1,2 +1,2 @@
 name,conversions,conversion_rate
-/,1,25.0
+,1,100.0

test/plausible_web/controllers/CSVs/30d-filter-goal/exit_pages.csv

@@ -1,2 +1,2 @@
 name,conversions,conversion_rate
-/,1,33.3
+,1,100.0

test/plausible_web/controllers/CSVs/30d/entry_pages.csv

@@ -1,2 +1,3 @@
 name,unique_entrances,total_entrances,visit_duration
-/,4,4,15
+/,3,3,20
+,1,1,0

test/plausible_web/controllers/CSVs/30d/exit_pages.csv

@@ -1,3 +1,4 @@
 name,unique_exits,total_exits,exit_rate
-/,3,3,100.0
+/,2,2,66.0
+,1,1,
 /some-other-page,1,1,100.0

test/plausible_web/controllers/CSVs/30d/pages.csv

@@ -1,3 +1,3 @@
 name,visitors,pageviews,bounce_rate,time_on_page
-/,4,3,75,
+/,4,3,67,
 /some-other-page,1,1,,60.0

test/plausible_web/controllers/CSVs/6m/entry_pages.csv

@@ -1,2 +1,3 @@
 name,unique_entrances,total_entrances,visit_duration
-/,5,5,12
+/,4,4,15
+,1,1,0

test/plausible_web/controllers/CSVs/6m/exit_pages.csv

@@ -1,3 +1,4 @@
 name,unique_exits,total_exits,exit_rate
-/,4,4,100.0
+/,3,3,75.0
+,1,1,
 /some-other-page,1,1,100.0

test/plausible_web/controllers/CSVs/6m/pages.csv

@@ -1,3 +1,3 @@
 name,visitors,pageviews,bounce_rate,time_on_page
-/,5,4,80,
+/,5,4,75,
 /some-other-page,1,1,,60.0