refactor session replays to be per visit. sessions persist too long resulting in low-quality recordings

This commit is contained in:
Francis Cao
2026-03-03 22:59:25 -08:00
parent b2d45613b5
commit 0e1d6cef6b
26 changed files with 248 additions and 83 deletions
+21 -12
View File
@@ -6,6 +6,8 @@ import prisma from '@/lib/prisma';
const FUNCTION_NAME = 'getReplayChunks';
export interface ReplayChunk {
sessionId: string;
visitId: string;
events: any[];
chunkIndex: number;
eventCount: number;
@@ -13,20 +15,19 @@ export interface ReplayChunk {
endedAt: Date;
}
export async function getReplayChunks(
websiteId: string,
sessionId: string,
): Promise<ReplayChunk[]> {
export async function getReplayChunks(websiteId: string, visitId: string): Promise<ReplayChunk[]> {
return runQuery({
[PRISMA]: () => relationalQuery(websiteId, sessionId),
[CLICKHOUSE]: () => clickhouseQuery(websiteId, sessionId),
[PRISMA]: () => relationalQuery(websiteId, visitId),
[CLICKHOUSE]: () => clickhouseQuery(websiteId, visitId),
});
}
async function relationalQuery(websiteId: string, sessionId: string): Promise<ReplayChunk[]> {
async function relationalQuery(websiteId: string, visitId: string): Promise<ReplayChunk[]> {
const { rawQuery } = prisma;
const chunks: {
sessionId: string;
visitId: string;
events: Buffer;
chunkIndex: number;
eventCount: number;
@@ -35,6 +36,8 @@ async function relationalQuery(websiteId: string, sessionId: string): Promise<Re
}[] = await rawQuery(
`
select
session_id as "sessionId",
visit_id as "visitId",
events,
chunk_index as "chunkIndex",
event_count as "eventCount",
@@ -42,10 +45,10 @@ async function relationalQuery(websiteId: string, sessionId: string): Promise<Re
ended_at as "endedAt"
from session_replay
where website_id = {{websiteId::uuid}}
and session_id = {{sessionId::uuid}}
and visit_id = {{visitId::uuid}}
order by chunk_index asc
`,
{ websiteId, sessionId },
{ websiteId, visitId },
FUNCTION_NAME,
);
@@ -55,11 +58,13 @@ async function relationalQuery(websiteId: string, sessionId: string): Promise<Re
}));
}
async function clickhouseQuery(websiteId: string, sessionId: string): Promise<ReplayChunk[]> {
async function clickhouseQuery(websiteId: string, visitId: string): Promise<ReplayChunk[]> {
const { rawQuery } = clickhouse;
const results = await rawQuery<
{
sessionId: string;
visitId: string;
events: string;
chunk_index: number;
event_count: number;
@@ -69,6 +74,8 @@ async function clickhouseQuery(websiteId: string, sessionId: string): Promise<Re
>(
`
select
session_id as sessionId,
visit_id as visitId,
events,
chunk_index,
event_count,
@@ -76,14 +83,16 @@ async function clickhouseQuery(websiteId: string, sessionId: string): Promise<Re
ended_at
from session_replay
where website_id = {websiteId:UUID}
and session_id = {sessionId:UUID}
and visit_id = {visitId:UUID}
order by chunk_index asc
`,
{ websiteId, sessionId },
{ websiteId, visitId },
FUNCTION_NAME,
);
return results.map(row => ({
sessionId: row.sessionId,
visitId: row.visitId,
events: JSON.parse(row.events),
chunkIndex: row.chunk_index,
eventCount: row.event_count,
+31 -18
View File
@@ -5,14 +5,16 @@ import type { QueryFilters } from '@/lib/types';
const FUNCTION_NAME = 'getSessionReplays';
export function getSessionReplays(...args: [websiteId: string, filters: QueryFilters]) {
export function getSessionReplays(
...args: [websiteId: string, filters: QueryFilters, sessionId?: string]
) {
return runQuery({
[PRISMA]: () => relationalQuery(...args),
[CLICKHOUSE]: () => clickhouseQuery(...args),
});
}
async function relationalQuery(websiteId: string, filters: QueryFilters) {
async function relationalQuery(websiteId: string, filters: QueryFilters, sessionId?: string) {
const { pagedRawQuery, parseFilters } = prisma;
const { search } = filters;
const { filterQuery, cohortQuery, queryParams } = parseFilters({
@@ -23,14 +25,19 @@ async function relationalQuery(websiteId: string, filters: QueryFilters) {
const joinQuery =
filterQuery || cohortQuery
? `join (select *
? `join (select distinct website_id, session_id, visit_id
from website_event
${cohortQuery}
where website_id = {{websiteId::uuid}}
and created_at between {{startDate}} and {{endDate}}) website_event
and created_at between {{startDate}} and {{endDate}}
${filterQuery}) website_event
on website_event.website_id = sr.website_id
and website_event.session_id = sr.session_id`
and website_event.session_id = sr.session_id
and website_event.visit_id = sr.visit_id`
: '';
const sessionFilter = sessionId ? 'and sr.session_id = {{sessionId::uuid}}' : '';
const searchQuery = search
? `and (session.distinct_id ilike {{search}}
or session.city ilike {{search}}
@@ -42,7 +49,8 @@ async function relationalQuery(websiteId: string, filters: QueryFilters) {
return pagedRawQuery(
`
select
sr.session_id as "id",
sr.visit_id as "id",
sr.session_id as "sessionId",
sr.website_id as "websiteId",
session.browser,
session.os,
@@ -58,13 +66,13 @@ async function relationalQuery(websiteId: string, filters: QueryFilters) {
from session_replay sr
join session on session.session_id = sr.session_id
and session.website_id = sr.website_id
${cohortQuery}
${joinQuery}
where sr.website_id = {{websiteId::uuid}}
and sr.created_at between {{startDate}} and {{endDate}}
${filterQuery}
${sessionFilter}
${searchQuery}
group by sr.session_id,
group by sr.visit_id,
sr.session_id,
sr.website_id,
session.browser,
session.os,
@@ -73,13 +81,13 @@ async function relationalQuery(websiteId: string, filters: QueryFilters) {
session.city
order by max(sr.created_at) desc
`,
queryParams,
{ ...queryParams, sessionId },
filters,
FUNCTION_NAME,
);
}
async function clickhouseQuery(websiteId: string, filters: QueryFilters) {
async function clickhouseQuery(websiteId: string, filters: QueryFilters, sessionId?: string) {
const { pagedRawQuery, parseFilters } = clickhouse;
const { search } = filters;
const { queryParams, cohortQuery, filterQuery } = parseFilters({
@@ -87,6 +95,8 @@ async function clickhouseQuery(websiteId: string, filters: QueryFilters) {
websiteId,
});
const sessionFilter = sessionId ? 'and session_replay.session_id = {sessionId:UUID}' : '';
const searchQuery = search
? `and ((positionCaseInsensitive(distinct_id, {search:String}) > 0)
or (positionCaseInsensitive(city, {search:String}) > 0)
@@ -98,7 +108,8 @@ async function clickhouseQuery(websiteId: string, filters: QueryFilters) {
return pagedRawQuery(
`
select
session_replay.session_id as id,
session_replay.visit_id as id,
session_replay.session_id as sessionId,
session_replay.website_id as websiteId,
website_event.browser,
website_event.os,
@@ -113,22 +124,24 @@ async function clickhouseQuery(websiteId: string, filters: QueryFilters) {
max(session_replay.created_at) as createdAt
from session_replay
join (
select *
select distinct website_id, session_id, visit_id, browser, os, device, country, city
from website_event
${cohortQuery}
where website_id = {websiteId:UUID}
and created_at between {startDate:DateTime64} and {endDate:DateTime64}
${filterQuery}
${searchQuery}
) website_event
on website_event.session_id = session_replay.session_id
and website_event.website_id = session_replay.website_id
${cohortQuery}
and website_event.visit_id = session_replay.visit_id
where session_replay.website_id = {websiteId:UUID}
and session_replay.created_at between {startDate:DateTime64} and {endDate:DateTime64}
${filterQuery}
${searchQuery}
group by session_replay.session_id, session_replay.website_id, website_event.browser, website_event.os, website_event.device, website_event.country, website_event.city
${sessionFilter}
group by session_replay.visit_id, session_replay.session_id, session_replay.website_id, website_event.browser, website_event.os, website_event.device, website_event.country, website_event.city
order by max(created_at) desc
`,
queryParams,
{ ...queryParams, sessionId },
filters,
FUNCTION_NAME,
);
+5
View File
@@ -7,6 +7,7 @@ import prisma from '@/lib/prisma';
export interface SaveRecordingArgs {
websiteId: string;
sessionId: string;
visitId: string;
chunkIndex: number;
events: any[];
eventCount: number;
@@ -24,6 +25,7 @@ export async function saveRecording(args: SaveRecordingArgs) {
async function relationalQuery({
websiteId,
sessionId,
visitId,
chunkIndex,
events,
eventCount,
@@ -37,6 +39,7 @@ async function relationalQuery({
id: uuid(),
websiteId,
sessionId,
visitId,
chunkIndex,
events: compressed as any,
eventCount,
@@ -49,6 +52,7 @@ async function relationalQuery({
async function clickhouseQuery({
websiteId,
sessionId,
visitId,
chunkIndex,
events,
eventCount,
@@ -62,6 +66,7 @@ async function clickhouseQuery({
replay_id: uuid(),
website_id: websiteId,
session_id: sessionId,
visit_id: visitId,
chunk_index: chunkIndex,
events: JSON.stringify(events),
event_count: eventCount,