From c130c2a751cb44bef0df7f99c70f265c583c6162 Mon Sep 17 00:00:00 2001
From: Uku Taht <Uku.taht@gmail.com>
Date: Wed, 6 Nov 2024 13:12:41 +0200
Subject: [PATCH] Fixes for `?source` and `?ref` params (#4781)

* Store `?source` and `?ref` params in `utm_source` field

* Remove dead code
---
 lib/plausible/ingestion/acquisition.ex        | 155 ++++++++----------
 lib/plausible/ingestion/event.ex              |  43 +++--
 .../api/external_controller_test.exs          |  35 +++-
 3 files changed, 132 insertions(+), 101 deletions(-)

diff --git a/lib/plausible/ingestion/acquisition.ex b/lib/plausible/ingestion/acquisition.ex
index d72dc11cb..c65a09cbf 100644
--- a/lib/plausible/ingestion/acquisition.ex
+++ b/lib/plausible/ingestion/acquisition.ex
@@ -45,58 +45,66 @@ defmodule Plausible.Ingestion.Acquisition do
                      |> then(&(@custom_source_categories ++ &1))
                      |> Enum.into(%{})
 
-  def get_channel(request, source) do
-    source = source && String.downcase(source)
+  def get_channel(source, utm_medium, utm_campaign, utm_source, click_id_param) do
+    get_channel_lowered(
+      String.downcase(source || ""),
+      String.downcase(utm_medium || ""),
+      String.downcase(utm_campaign || ""),
+      String.downcase(utm_source || ""),
+      click_id_param
+    )
+  end
 
+  defp get_channel_lowered(source, utm_medium, utm_campaign, utm_source, click_id_param) do
     cond do
-      cross_network?(request) -> "Cross-network"
-      paid_shopping?(request, source) -> "Paid Shopping"
-      paid_search?(request, source) -> "Paid Search"
-      paid_social?(request, source) -> "Paid Social"
-      paid_video?(request, source) -> "Paid Video"
-      display?(request) -> "Display"
-      paid_other?(request) -> "Paid Other"
-      organic_shopping?(request, source) -> "Organic Shopping"
-      organic_social?(request, source) -> "Organic Social"
-      organic_video?(request, source) -> "Organic Video"
+      cross_network?(utm_campaign) -> "Cross-network"
+      paid_shopping?(source, utm_campaign, utm_medium) -> "Paid Shopping"
+      paid_search?(source, utm_medium, utm_source, click_id_param) -> "Paid Search"
+      paid_social?(source, utm_medium, utm_source) -> "Paid Social"
+      paid_video?(source, utm_medium, utm_source) -> "Paid Video"
+      display?(utm_medium) -> "Display"
+      paid_other?(utm_medium) -> "Paid Other"
+      organic_shopping?(source, utm_campaign) -> "Organic Shopping"
+      organic_social?(source, utm_medium) -> "Organic Social"
+      organic_video?(source, utm_medium) -> "Organic Video"
       search_source?(source) -> "Organic Search"
-      email?(request, source) -> "Email"
-      affiliates?(request) -> "Affiliates"
-      audio?(request) -> "Audio"
-      sms?(request) -> "SMS"
-      mobile_push_notifications?(request, source) -> "Mobile Push Notifications"
-      referral?(request, source) -> "Referral"
+      email?(source, utm_source, utm_medium) -> "Email"
+      affiliates?(utm_medium) -> "Affiliates"
+      audio?(utm_medium) -> "Audio"
+      sms?(utm_source, utm_medium) -> "SMS"
+      mobile_push_notifications?(source, utm_medium) -> "Mobile Push Notifications"
+      referral?(source, utm_medium) -> "Referral"
       true -> "Direct"
     end
   end
 
-  defp cross_network?(request) do
-    String.contains?(query_param(request, "utm_campaign"), "cross-network")
+  defp cross_network?(utm_campaign) do
+    String.contains?(utm_campaign, "cross-network")
   end
 
-  defp paid_shopping?(request, source) do
-    (shopping_source?(source) or shopping_campaign?(request)) and paid_medium?(request)
+  defp paid_shopping?(source, utm_campaign, utm_medium) do
+    (shopping_source?(source) or shopping_campaign?(utm_campaign)) and paid_medium?(utm_medium)
   end
 
-  defp paid_search?(request, source) do
-    (search_source?(source) and paid_medium?(request)) or
-      (search_source?(source) and paid_source?(request)) or
-      (source == "google" and !!request.query_params["gclid"]) or
-      (source == "bing" and !!request.query_params["msclkid"])
+  defp paid_search?(source, utm_medium, utm_source, click_id_param) do
+    (search_source?(source) and paid_medium?(utm_medium)) or
+      (search_source?(source) and paid_source?(utm_source)) or
+      (source == "google" and click_id_param == "gclid") or
+      (source == "bing" and click_id_param == "msclkid")
   end
 
-  defp paid_social?(request, source) do
-    (social_source?(source) and paid_medium?(request)) or
-      (social_source?(source) and paid_source?(request))
+  defp paid_social?(source, utm_medium, utm_source) do
+    (social_source?(source) and paid_medium?(utm_medium)) or
+      (social_source?(source) and paid_source?(utm_source))
   end
 
-  defp paid_video?(request, source) do
-    (video_source?(source) and paid_medium?(request)) or
-      (video_source?(source) and paid_source?(request))
+  defp paid_video?(source, utm_medium, utm_source) do
+    (video_source?(source) and paid_medium?(utm_medium)) or
+      (video_source?(source) and paid_source?(utm_source))
   end
 
-  defp display?(request) do
-    query_param(request, "utm_medium") in [
+  defp display?(utm_medium) do
+    utm_medium in [
       "display",
       "banner",
       "expandable",
@@ -105,17 +113,17 @@ defmodule Plausible.Ingestion.Acquisition do
     ]
   end
 
-  defp paid_other?(request) do
-    paid_medium?(request)
+  defp paid_other?(utm_medium) do
+    paid_medium?(utm_medium)
   end
 
-  defp organic_shopping?(request, source) do
-    shopping_source?(source) or shopping_campaign?(request)
+  defp organic_shopping?(source, utm_campaign) do
+    shopping_source?(source) or shopping_campaign?(utm_campaign)
   end
 
-  defp organic_social?(request, source) do
+  defp organic_social?(source, utm_medium) do
     social_source?(source) or
-      query_param(request, "utm_medium") in [
+      utm_medium in [
         "social",
         "social-network",
         "social-media",
@@ -125,89 +133,68 @@ defmodule Plausible.Ingestion.Acquisition do
       ]
   end
 
-  defp organic_video?(request, source) do
-    video_source?(source) or String.contains?(query_param(request, "utm_medium"), "video")
+  defp organic_video?(source, utm_medium) do
+    video_source?(source) or String.contains?(utm_medium, "video")
   end
 
-  defp referral?(request, source) do
-    query_param(request, "utm_medium") in ["referral", "app", "link"] or
-      !!source
+  defp referral?(source, utm_medium) do
+    utm_medium in ["referral", "app", "link"] or source !== ""
   end
 
   @email_tags ["email", "e-mail", "e_mail", "e mail", "newsletter"]
-  defp email?(request, source) do
+  defp email?(source, utm_source, utm_medium) do
     email_source?(source) or
-      String.contains?(query_param(request, "utm_source"), @email_tags) or
-      String.contains?(query_param(request, "utm_medium"), @email_tags)
+      String.contains?(utm_source, @email_tags) or
+      String.contains?(utm_medium, @email_tags)
   end
 
-  defp affiliates?(request) do
-    query_param(request, "utm_medium") == "affiliate"
+  defp affiliates?(utm_medium) do
+    utm_medium == "affiliate"
   end
 
-  defp audio?(request) do
-    query_param(request, "utm_medium") == "audio"
+  defp audio?(utm_medium) do
+    utm_medium == "audio"
   end
 
-  defp sms?(request) do
-    query_param(request, "utm_source") == "sms" or
-      query_param(request, "utm_medium") == "sms"
+  defp sms?(utm_source, utm_medium) do
+    utm_source == "sms" or utm_medium == "sms"
   end
 
-  defp mobile_push_notifications?(request, source) do
-    medium = query_param(request, "utm_medium")
-
-    String.ends_with?(medium, "push") or
-      String.contains?(medium, ["mobile", "notification"]) or
+  defp mobile_push_notifications?(source, utm_medium) do
+    String.ends_with?(utm_medium, "push") or
+      String.contains?(utm_medium, ["mobile", "notification"]) or
       source == "firebase"
   end
 
-  defp shopping_source?(nil), do: false
-
   defp shopping_source?(source) do
     @source_categories[source] == "SOURCE_CATEGORY_SHOPPING"
   end
 
-  defp search_source?(nil), do: false
-
   defp search_source?(source) do
     @source_categories[source] == "SOURCE_CATEGORY_SEARCH"
   end
 
-  defp social_source?(nil), do: false
-
   defp social_source?(source) do
     @source_categories[source] == "SOURCE_CATEGORY_SOCIAL"
   end
 
-  defp video_source?(nil), do: false
-
   defp video_source?(source) do
     @source_categories[source] == "SOURCE_CATEGORY_VIDEO"
   end
 
-  defp email_source?(nil), do: false
-
   defp email_source?(source) do
     @source_categories[source] == "SOURCE_CATEGORY_EMAIL"
   end
 
-  defp shopping_campaign?(request) do
-    campaign_name = query_param(request, "utm_campaign")
-    Regex.match?(~r/^(.*(([^a-df-z]|^)shop|shopping).*)$/, campaign_name)
+  defp shopping_campaign?(utm_campaign) do
+    Regex.match?(~r/^(.*(([^a-df-z]|^)shop|shopping).*)$/, utm_campaign)
   end
 
-  defp paid_medium?(request) do
-    medium = query_param(request, "utm_medium")
-    Regex.match?(~r/^(.*cp.*|ppc|retargeting|paid.*)$/, medium)
+  defp paid_medium?(utm_medium) do
+    Regex.match?(~r/^(.*cp.*|ppc|retargeting|paid.*)$/, utm_medium)
   end
 
-  defp paid_source?(request) do
-    query_param(request, "utm_source")
-    |> Plausible.Ingestion.Source.paid_source?()
-  end
-
-  defp query_param(request, name) do
-    String.downcase(request.query_params[name] || "")
+  defp paid_source?(utm_source) do
+    Plausible.Ingestion.Source.paid_source?(utm_source)
   end
 end
diff --git a/lib/plausible/ingestion/event.ex b/lib/plausible/ingestion/event.ex
index 313bf25df..7aa4d4c3b 100644
--- a/lib/plausible/ingestion/event.ex
+++ b/lib/plausible/ingestion/event.ex
@@ -118,8 +118,8 @@ defmodule Plausible.Ingestion.Event do
       drop_shield_rule_country: &drop_shield_rule_country/2,
       put_user_agent: &put_user_agent/2,
       put_basic_info: &put_basic_info/2,
-      put_referrer: &put_referrer/2,
-      put_utm_tags: &put_utm_tags/2,
+      put_source_info: &put_source_info/2,
+      put_channel: &put_channel/2,
       put_props: &put_props/2,
       put_revenue: &put_revenue/2,
       put_salts: &put_salts/2,
@@ -250,30 +250,41 @@ defmodule Plausible.Ingestion.Event do
     })
   end
 
-  defp put_referrer(%__MODULE__{} = event, _context) do
-    source = Plausible.Ingestion.Source.resolve(event.request)
-    channel = Plausible.Ingestion.Acquisition.get_channel(event.request, source)
-
-    update_session_attrs(event, %{
-      channel: channel,
-      referrer_source: source,
-      referrer: Plausible.Ingestion.Source.format_referrer(event.request.referrer),
-      click_id_param: get_click_id_param(event.request.query_params)
-    })
-  end
-
-  defp put_utm_tags(%__MODULE__{} = event, _context) do
+  defp put_source_info(%__MODULE__{} = event, _context) do
     query_params = event.request.query_params
 
+    tagged_source =
+      query_params["utm_source"] ||
+        query_params["source"] ||
+        query_params["ref"]
+
     update_session_attrs(event, %{
+      referrer_source: Plausible.Ingestion.Source.resolve(event.request),
+      referrer: Plausible.Ingestion.Source.format_referrer(event.request.referrer),
+      click_id_param: get_click_id_param(event.request.query_params),
+      utm_source: tagged_source,
       utm_medium: query_params["utm_medium"],
-      utm_source: query_params["utm_source"],
       utm_campaign: query_params["utm_campaign"],
       utm_content: query_params["utm_content"],
       utm_term: query_params["utm_term"]
     })
   end
 
+  defp put_channel(%__MODULE__{} = event, _context) do
+    session = event.clickhouse_session_attrs
+
+    channel =
+      Plausible.Ingestion.Acquisition.get_channel(
+        session[:referrer_source],
+        session[:utm_medium],
+        session[:utm_campaign],
+        session[:utm_source],
+        session[:click_id_param]
+      )
+
+    update_session_attrs(event, %{channel: channel})
+  end
+
   defp put_geolocation(%__MODULE__{} = event, _context) do
     case event.request.ip_classification do
       "anonymous_vpn_ip" ->
diff --git a/test/plausible_web/controllers/api/external_controller_test.exs b/test/plausible_web/controllers/api/external_controller_test.exs
index 7578b5681..dfb6c46c1 100644
--- a/test/plausible_web/controllers/api/external_controller_test.exs
+++ b/test/plausible_web/controllers/api/external_controller_test.exs
@@ -406,6 +406,39 @@ defmodule PlausibleWeb.Api.ExternalControllerTest do
       session = get_created_session(site)
 
       assert session.referrer_source == "betalist"
+      assert session.utm_source == "betalist"
+    end
+
+    test "?ref param behaves like ?utm_source", %{conn: conn, site: site} do
+      params = %{
+        name: "pageview",
+        url: "http://www.example.com/?ref=betalist",
+        domain: site.domain
+      }
+
+      conn
+      |> post("/api/event", params)
+
+      session = get_created_session(site)
+
+      assert session.referrer_source == "betalist"
+      assert session.utm_source == "betalist"
+    end
+
+    test "?source param behaves like ?utm_source", %{conn: conn, site: site} do
+      params = %{
+        name: "pageview",
+        url: "http://www.example.com/?source=betalist",
+        domain: site.domain
+      }
+
+      conn
+      |> post("/api/event", params)
+
+      session = get_created_session(site)
+
+      assert session.referrer_source == "betalist"
+      assert session.utm_source == "betalist"
     end
 
     test "if utm_source matches a capitalized form from ref_inspector, the capitalized form is recorded",
@@ -2192,7 +2225,7 @@ defmodule PlausibleWeb.Api.ExternalControllerTest do
     } do
       params = %{
         name: "pageview",
-        url: "http://example.com?utm_source=Google-ads",
+        url: "http://example.com?source=Google-ads",
         domain: site.domain
       }