Fix excludeBounce query, simplify bounce definition for SPA websites, query routing for performance, add tests

src/queries/sql/pageviews/getPageviewExpandedMetrics.ts

@@ -1,5 +1,10 @@
 import clickhouse from '@/lib/clickhouse';
-import { BOUNCE_THRESHOLD, EVENT_TYPE, FILTER_COLUMNS, GROUPED_DOMAINS, SESSION_COLUMNS } from '@/lib/constants';
+import {
+  EVENT_TYPE,
+  FILTER_COLUMNS,
+  GROUPED_DOMAINS,
+  SESSION_COLUMNS,
+} from '@/lib/constants';
 import { CLICKHOUSE, PRISMA, runQuery } from '@/lib/db';
 import prisma from '@/lib/prisma';
 import type { QueryFilters } from '@/lib/types';
@@ -46,6 +51,7 @@ async function relationalQuery(
       },
       { joinSession: SESSION_COLUMNS.includes(type) },
     );
+  const needsBounceEvents = filters.excludeBounce !== true;
   const fullPathSearchQuery =
     type === 'fullPath' && filters.search
       ? `and (case when website_event.url_query != '' then website_event.url_path || '?' || website_event.url_query else website_event.url_path end) ilike {{fullPathSearch}}`
@@ -88,6 +94,21 @@ async function relationalQuery(
     type === 'fullPath'
       ? `case when website_event.url_query != '' then website_event.url_path || '?' || website_event.url_query else website_event.url_path end`
       : column;
+  const bounceQuery = needsBounceEvents
+    ? `sum(case when t.c = 1 and coalesce(e.has_custom_event, 0) = 0 then 1 else 0 end) as "bounces",`
+    : '0 as "bounces",';
+  const visitEventsJoin = needsBounceEvents
+    ? `left join (
+      select session_id, visit_id, 1 as "has_custom_event"
+      from website_event
+      where website_id = {{websiteId::uuid}}
+        and created_at between {{startDate}} and {{endDate}}
+        and event_type = ${EVENT_TYPE.customEvent}
+      group by 1, 2
+    ) as e
+      on e.session_id = t.session_id
+      and e.visit_id = t.visit_id`
+    : '';
 
   return rawQuery(
     `
@@ -96,7 +117,7 @@ async function relationalQuery(
       sum(t.c) as "pageviews",
       count(distinct t.session_id) as "visitors",
       count(distinct t.visit_id) as "visits",
-      sum(case when t.c = 1 and t.events_count < ${BOUNCE_THRESHOLD} then 1 else 0 end) as "bounces",
+      ${bounceQuery}
       sum(${getTimestampDiffSQL('t.min_time', 't.max_time')}) as "totaltime"
     from (
       select
@@ -105,16 +126,7 @@ async function relationalQuery(
         website_event.visit_id,
         count(*) as "c",
         min(website_event.created_at) as "min_time",
-        max(website_event.created_at) as "max_time",
-        max((
-          select count(*)
-          from website_event we2
-          where we2.website_id = website_event.website_id
-            and we2.session_id = website_event.session_id
-            and we2.visit_id = website_event.visit_id
-            and we2.created_at between {{startDate}} and {{endDate}}
-            and we2.event_type = ${EVENT_TYPE.customEvent}
-        )) as "events_count"
+        max(website_event.created_at) as "max_time"
       from website_event
       ${cohortQuery}
       ${excludeBounceQuery}
@@ -128,6 +140,7 @@ async function relationalQuery(
         ${filterQuery}
       group by ${groupByColumn}, website_event.session_id, website_event.visit_id
     ) as t
+    ${visitEventsJoin}
     where name != ''
     group by name
     order by visitors desc, visits desc
@@ -154,6 +167,7 @@ async function clickhouseQuery(
     ...filters,
     websiteId,
   });
+  const needsBounceEvents = filters.excludeBounce !== true;
   const fullPathSearchQuery =
     type === 'fullPath' && filters.search
       ? `and positionCaseInsensitive(if(url_query != '', concat(url_path, '?', url_query), url_path), {fullPathSearch:String}) > 0`
@@ -189,6 +203,19 @@ async function clickhouseQuery(
   } else if (type === 'fullPath') {
     selectColumn = `if(url_query != '', concat(url_path, '?', url_query), url_path)`;
   }
+  const bounceQuery = needsBounceEvents
+    ? `sumIf(1, t.c = 1 and ifNull(e.has_custom_event, 0) = 0) as "bounces",`
+    : '0 as "bounces",';
+  const visitEventsJoin = needsBounceEvents
+    ? `left join (
+      select session_id, visit_id, toUInt8(1) as has_custom_event
+      from website_event
+      where website_id = {websiteId:UUID}
+        and created_at between {startDate:DateTime64} and {endDate:DateTime64}
+        and event_type = ${EVENT_TYPE.customEvent}
+      group by session_id, visit_id
+    ) as e using (session_id, visit_id)`
+    : '';
 
   return rawQuery(
     `
@@ -197,7 +224,7 @@ async function clickhouseQuery(
       sum(t.c) as "pageviews",
       uniq(t.session_id) as "visitors",
       uniq(t.visit_id) as "visits",
-      sumIf(1, t.c = 1 and ifNull(e.events_count, 0) < ${BOUNCE_THRESHOLD}) as "bounces",
+      ${bounceQuery}
       sum(max_time-min_time) as "totaltime"
     from (
       select
@@ -220,14 +247,7 @@ async function clickhouseQuery(
         ${filterQuery}
       group by name, session_id, visit_id
     ) as t
-    left join (
-      select session_id, visit_id, toUInt32(count()) as events_count
-      from website_event
-      where website_id = {websiteId:UUID}
-        and created_at between {startDate:DateTime64} and {endDate:DateTime64}
-        and event_type = ${EVENT_TYPE.customEvent}
-      group by session_id, visit_id
-    ) as e using (session_id, visit_id)
+    ${visitEventsJoin}
     group by name
     order by visitors desc, visits desc
     limit ${limit}