implement cohorts to clickhouse/pg library and all relevant queries

This commit is contained in:
Francis Cao 2025-07-03 12:06:49 -07:00
parent a753809a74
commit e75d009df3
22 changed files with 220 additions and 10983 deletions

View file

@ -14,7 +14,7 @@ export async function getEventDataFields(
async function relationalQuery(websiteId: string, filters: QueryFilters) {
const { rawQuery, parseFilters, getDateSQL } = prisma;
const { filterQuery, params } = await parseFilters(websiteId, filters);
const { filterQuery, cohortQuery, params } = await parseFilters(websiteId, filters);
return rawQuery(
`
@ -29,6 +29,9 @@ async function relationalQuery(websiteId: string, filters: QueryFilters) {
count(*) as "total"
from event_data
join website_event on website_event.event_id = event_data.website_event_id
and website_event.website_id = {{websiteId::uuid}}
and website_event.created_at between {{startDate}} and {{endDate}}
${cohortQuery}
where event_data.website_id = {{websiteId::uuid}}
and event_data.created_at between {{startDate}} and {{endDate}}
${filterQuery}
@ -45,7 +48,7 @@ async function clickhouseQuery(
filters: QueryFilters,
): Promise<{ propertyName: string; dataType: number; propertyValue: string; total: number }[]> {
const { rawQuery, parseFilters } = clickhouse;
const { filterQuery, params } = await parseFilters(websiteId, filters);
const { filterQuery, cohortQuery, params } = await parseFilters(websiteId, filters);
return rawQuery(
`
@ -56,7 +59,8 @@ async function clickhouseQuery(
data_type = 4, toString(date_trunc('hour', date_value)),
string_value) as "value",
count(*) as "total"
from event_data
from event_data website_event
${cohortQuery}
where website_id = {websiteId:UUID}
and created_at between {startDate:DateTime64} and {endDate:DateTime64}
${filterQuery}

View file

@ -17,7 +17,7 @@ async function relationalQuery(
filters: QueryFilters & { propertyName?: string },
) {
const { rawQuery, parseFilters } = prisma;
const { filterQuery, params } = await parseFilters(websiteId, filters, {
const { filterQuery, cohortQuery, params } = await parseFilters(websiteId, filters, {
columns: { propertyName: 'data_key' },
});
@ -29,6 +29,9 @@ async function relationalQuery(
count(*) as "total"
from event_data
join website_event on website_event.event_id = event_data.website_event_id
and website_event.website_id = {{websiteId::uuid}}
and website_event.created_at between {{startDate}} and {{endDate}}
${cohortQuery}
where event_data.website_id = {{websiteId::uuid}}
and event_data.created_at between {{startDate}} and {{endDate}}
${filterQuery}
@ -45,7 +48,7 @@ async function clickhouseQuery(
filters: QueryFilters & { propertyName?: string },
): Promise<{ eventName: string; propertyName: string; total: number }[]> {
const { rawQuery, parseFilters } = clickhouse;
const { filterQuery, params } = await parseFilters(websiteId, filters, {
const { filterQuery, cohortQuery, params } = await parseFilters(websiteId, filters, {
columns: { propertyName: 'data_key' },
});
@ -55,7 +58,8 @@ async function clickhouseQuery(
event_name as eventName,
data_key as propertyName,
count(*) as total
from event_data
from event_data website_event
${cohortQuery}
where website_id = {websiteId:UUID}
and created_at between {startDate:DateTime64} and {endDate:DateTime64}
${filterQuery}

View file

@ -18,7 +18,7 @@ export async function getEventDataStats(
async function relationalQuery(websiteId: string, filters: QueryFilters) {
const { rawQuery, parseFilters } = prisma;
const { filterQuery, params } = await parseFilters(websiteId, filters);
const { filterQuery, cohortQuery, params } = await parseFilters(websiteId, filters);
return rawQuery(
`
@ -32,8 +32,12 @@ async function relationalQuery(websiteId: string, filters: QueryFilters) {
data_key,
count(*) as "total"
from event_data
where website_id = {{websiteId::uuid}}
and created_at between {{startDate}} and {{endDate}}
join website_event on website_event.event_id = event_data.website_event_id
and website_event.website_id = {{websiteId::uuid}}
and website_event.created_at between {{startDate}} and {{endDate}}
${cohortQuery}
where event_data.website_id = {{websiteId::uuid}}
and event_data.created_at between {{startDate}} and {{endDate}}
${filterQuery}
group by website_event_id, data_key
) as t
@ -47,7 +51,7 @@ async function clickhouseQuery(
filters: QueryFilters,
): Promise<{ events: number; properties: number; records: number }[]> {
const { rawQuery, parseFilters } = clickhouse;
const { filterQuery, params } = await parseFilters(websiteId, filters);
const { filterQuery, cohortQuery, params } = await parseFilters(websiteId, filters);
return rawQuery(
`
@ -60,7 +64,8 @@ async function clickhouseQuery(
event_id,
data_key,
count(*) as "total"
from event_data
from event_data website_event
${cohortQuery}
where website_id = {websiteId:UUID}
and created_at between {startDate:DateTime64} and {endDate:DateTime64}
${filterQuery}

View file

@ -20,7 +20,7 @@ async function relationalQuery(
filters: QueryFilters & { eventName?: string; propertyName?: string },
) {
const { rawQuery, parseFilters, getDateSQL } = prisma;
const { filterQuery, params } = await parseFilters(websiteId, filters);
const { filterQuery, cohortQuery, params } = await parseFilters(websiteId, filters);
return rawQuery(
`
@ -33,6 +33,9 @@ async function relationalQuery(
count(*) as "total"
from event_data
join website_event on website_event.event_id = event_data.website_event_id
and website_event.website_id = {{websiteId::uuid}}
and website_event.created_at between {{startDate}} and {{endDate}}
${cohortQuery}
where event_data.website_id = {{websiteId::uuid}}
and event_data.created_at between {{startDate}} and {{endDate}}
and event_data.data_key = {{propertyName}}
@ -51,7 +54,7 @@ async function clickhouseQuery(
filters: QueryFilters & { eventName?: string; propertyName?: string },
): Promise<{ value: string; total: number }[]> {
const { rawQuery, parseFilters } = clickhouse;
const { filterQuery, params } = await parseFilters(websiteId, filters);
const { filterQuery, cohortQuery, params } = await parseFilters(websiteId, filters);
return rawQuery(
`
@ -60,7 +63,8 @@ async function clickhouseQuery(
data_type = 4, toString(date_trunc('hour', date_value)),
string_value) as "value",
count(*) as "total"
from event_data
from event_data website_event
${cohortQuery}
where website_id = {websiteId:UUID}
and created_at between {startDate:DateTime64} and {endDate:DateTime64}
and data_key = {propertyName:String}

View file

@ -16,7 +16,7 @@ export async function getEventMetrics(
async function relationalQuery(websiteId: string, filters: QueryFilters) {
const { timezone = 'utc', unit = 'day' } = filters;
const { rawQuery, getDateSQL, parseFilters } = prisma;
const { filterQuery, joinSession, params } = await parseFilters(websiteId, {
const { filterQuery, cohortQuery, joinSession, params } = await parseFilters(websiteId, {
...filters,
eventType: EVENT_TYPE.customEvent,
});
@ -28,6 +28,7 @@ async function relationalQuery(websiteId: string, filters: QueryFilters) {
${getDateSQL('website_event.created_at', unit, timezone)} t,
count(*) y
from website_event
${cohortQuery}
${joinSession}
where website_event.website_id = {{websiteId::uuid}}
and website_event.created_at between {{startDate}} and {{endDate}}
@ -46,20 +47,21 @@ async function clickhouseQuery(
): Promise<{ x: string; t: string; y: number }[]> {
const { timezone = 'UTC', unit = 'day' } = filters;
const { rawQuery, getDateSQL, parseFilters } = clickhouse;
const { filterQuery, params } = await parseFilters(websiteId, {
const { filterQuery, cohortQuery, params } = await parseFilters(websiteId, {
...filters,
eventType: EVENT_TYPE.customEvent,
});
let sql = '';
if (filterQuery) {
if (filterQuery || cohortQuery) {
sql = `
select
event_name x,
${getDateSQL('created_at', unit, timezone)} t,
count(*) y
from website_event
${cohortQuery}
where website_id = {websiteId:UUID}
and created_at between {startDate:DateTime64} and {endDate:DateTime64}
and event_type = {eventType:UInt32}

View file

@ -15,7 +15,7 @@ export function getWebsiteEvents(
async function relationalQuery(websiteId: string, filters: QueryFilters, pageParams?: PageParams) {
const { pagedRawQuery, parseFilters } = prisma;
const { search } = pageParams;
const { filterQuery, params } = await parseFilters(websiteId, {
const { filterQuery, cohortQuery, params } = await parseFilters(websiteId, {
...filters,
});
@ -38,6 +38,7 @@ async function relationalQuery(websiteId: string, filters: QueryFilters, pagePar
event_type as "eventType",
event_name as "eventName"
from website_event
${cohortQuery}
where website_id = {{websiteId::uuid}}
and created_at between {{startDate}} and {{endDate}}
${filterQuery}
@ -56,7 +57,7 @@ async function relationalQuery(websiteId: string, filters: QueryFilters, pagePar
async function clickhouseQuery(websiteId: string, filters: QueryFilters, pageParams?: PageParams) {
const { pagedQuery, parseFilters } = clickhouse;
const { params, dateQuery, filterQuery } = await parseFilters(websiteId, filters);
const { params, dateQuery, filterQuery, cohortQuery } = await parseFilters(websiteId, filters);
const { search } = pageParams;
return pagedQuery(
@ -75,6 +76,7 @@ async function clickhouseQuery(websiteId: string, filters: QueryFilters, pagePar
event_type as eventType,
event_name as eventName
from website_event
${cohortQuery}
where website_id = {websiteId:UUID}
${dateQuery}
${filterQuery}