Event data properties refactor and design

This commit is contained in:
Francis Cao committed 2026-04-23 11:15:59 -07:00
1 parent 93ed3cb89f
commit be6154cb68
22 files changed
+1211 -125

No files matched your search

+7 -17
View File
@@ -13,7 +13,7 @@ export async function getEventDataFields(...args: [websiteId: string, filters: Q
}
async function relationalQuery(websiteId: string, filters: QueryFilters) {
const { rawQuery, parseFilters, getDateSQL } = prisma;
const { rawQuery, parseFilters } = prisma;
const { filterQuery, cohortQuery, joinSessionQuery, queryParams } = parseFilters({
...filters,
websiteId,
@@ -24,11 +24,6 @@ async function relationalQuery(websiteId: string, filters: QueryFilters) {
select
data_key as "propertyName",
data_type as "dataType",
case
when data_type = 2 then replace(string_value, '.0000', '')
when data_type = 4 then ${getDateSQL('date_value', 'hour')}
else string_value
end as "value",
count(*) as "total"
from event_data
join website_event on website_event.event_id = event_data.website_event_id
@@ -39,9 +34,8 @@ async function relationalQuery(websiteId: string, filters: QueryFilters) {
where event_data.website_id = {{websiteId::uuid}}
and event_data.created_at between {{startDate}} and {{endDate}}
${filterQuery}
group by data_key, data_type, value
order by 2 desc
limit 100
group by data_key, data_type
order by "total" desc, "propertyName" asc
`,
queryParams,
FUNCTION_NAME,
@@ -51,7 +45,7 @@ async function relationalQuery(websiteId: string, filters: QueryFilters) {
async function clickhouseQuery(
websiteId: string,
filters: QueryFilters,
): Promise<{ propertyName: string; dataType: number; propertyValue: string; total: number }[]> {
): Promise<{ propertyName: string; dataType: number; total: number }[]> {
const { rawQuery, parseFilters } = clickhouse;
const { filterQuery, cohortQuery, queryParams } = parseFilters({ ...filters, websiteId });
@@ -60,13 +54,10 @@ async function clickhouseQuery(
select
data_key as propertyName,
data_type as dataType,
multiIf(data_type = 2, replaceAll(string_value, '.0000', ''),
data_type = 4, toString(date_trunc('hour', date_value)),
string_value) as "value",
count(*) as "total"
from event_data
any left join (
select *
select *
from website_event
where website_id = {websiteId:UUID}
and created_at between {startDate:DateTime64} and {endDate:DateTime64}
@@ -78,9 +69,8 @@ async function clickhouseQuery(
where event_data.website_id = {websiteId:UUID}
and event_data.created_at between {startDate:DateTime64} and {endDate:DateTime64}
${filterQuery}
group by data_key, data_type, value
order by 2 desc
limit 100
group by data_key, data_type
order by total desc, propertyName asc
`,
queryParams,
FUNCTION_NAME,
@@ -0,0 +1,104 @@
import clickhouse from '@/lib/clickhouse';
import { CLICKHOUSE, PRISMA, runQuery } from '@/lib/db';
import prisma from '@/lib/prisma';
import type { QueryFilters } from '@/lib/types';
const FUNCTION_NAME = 'getEventDataNumericSeries';
export async function getEventDataNumericSeries(
...args: [
websiteId: string,
eventName: string,
propertyName: string,
metric: 'sum' | 'avg',
filters: QueryFilters,
]
) {
return runQuery({
[PRISMA]: () => relationalQuery(...args),
[CLICKHOUSE]: () => clickhouseQuery(...args),
});
}
async function relationalQuery(
websiteId: string,
eventName: string,
propertyName: string,
metric: 'sum' | 'avg',
filters: QueryFilters,
) {
const { timezone = 'utc', unit = 'day' } = filters;
const { rawQuery, getDateSQL, parseFilters } = prisma;
const { filterQuery, cohortQuery, joinSessionQuery, queryParams } = parseFilters({
...filters,
websiteId,
});
const aggFn = metric === 'avg' ? 'avg' : 'sum';
return rawQuery(
`
select
${getDateSQL('event_data.created_at', unit, timezone)} t,
${aggFn}(cast(event_data.number_value as decimal)) y
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}}
and website_event.event_type = 2
and website_event.event_name = {{eventName}}
${cohortQuery}
${joinSessionQuery}
where event_data.website_id = {{websiteId::uuid}}
and event_data.created_at between {{startDate}} and {{endDate}}
and event_data.data_key = {{propertyName}}
and event_data.data_type = 2
${filterQuery}
group by 1
order by 1
`,
{ ...queryParams, eventName, propertyName },
FUNCTION_NAME,
);
}
async function clickhouseQuery(
websiteId: string,
eventName: string,
propertyName: string,
metric: 'sum' | 'avg',
filters: QueryFilters,
): Promise<{ t: string; y: number }[]> {
const { timezone = 'UTC', unit = 'day' } = filters;
const { rawQuery, getDateSQL, parseFilters } = clickhouse;
const { filterQuery, cohortQuery, queryParams } = parseFilters({ ...filters, websiteId });
const aggFn = metric === 'avg' ? 'avg' : 'sum';
return rawQuery(
`
select
${getDateSQL('event_data.created_at', unit, timezone)} as t,
${aggFn}(event_data.number_value) as y
from event_data
any left join (
select *
from website_event
where website_id = {websiteId:UUID}
and created_at between {startDate:DateTime64} and {endDate:DateTime64}
and event_type = 2
and event_name = {eventName:String}) website_event
on website_event.event_id = event_data.event_id
and website_event.session_id = event_data.session_id
and website_event.website_id = event_data.website_id
${cohortQuery}
where event_data.website_id = {websiteId:UUID}
and event_data.created_at between {startDate:DateTime64} and {endDate:DateTime64}
and event_data.data_key = {propertyName:String}
and event_data.data_type = 2
${filterQuery}
group by t
order by t
`,
{ ...queryParams, eventName, propertyName },
FUNCTION_NAME,
);
}
+173
View File
@@ -0,0 +1,173 @@
import clickhouse from '@/lib/clickhouse';
import { DEFAULT_PAGE_SIZE } from '@/lib/constants';
import { CLICKHOUSE, PRISMA, runQuery } from '@/lib/db';
import prisma from '@/lib/prisma';
import type { QueryFilters } from '@/lib/types';
const FUNCTION_NAME = 'getEventDataPivot';
export async function getEventDataPivot(
...args: [websiteId: string, eventName: string, filters: QueryFilters]
) {
return runQuery({
[PRISMA]: () => relationalQuery(...args),
[CLICKHOUSE]: () => clickhouseQuery(...args),
});
}
async function relationalQuery(websiteId: string, eventName: string, filters: QueryFilters) {
const { rawQuery, parseFilters } = prisma;
const { page = 1, pageSize } = filters;
const size = +pageSize || DEFAULT_PAGE_SIZE;
const offset = +size * (+page - 1);
const { filterQuery, cohortQuery, joinSessionQuery, queryParams } = parseFilters({
...filters,
websiteId,
});
const countResult = await rawQuery(
`
select count(distinct website_event.event_id) as num
from website_event
join event_data on event_data.website_event_id = website_event.event_id
and event_data.website_id = {{websiteId::uuid}}
and event_data.created_at between {{startDate}} and {{endDate}}
${cohortQuery}
${joinSessionQuery}
where website_event.website_id = {{websiteId::uuid}}
and website_event.created_at between {{startDate}} and {{endDate}}
and website_event.event_name = {{eventName}}
${filterQuery}
`,
{ ...queryParams, eventName },
);
const count = countResult[0].num;
const rows = await rawQuery(
`
with paged_events as (
select website_event.event_id
from website_event
join event_data on event_data.website_event_id = website_event.event_id
and event_data.website_id = {{websiteId::uuid}}
and event_data.created_at between {{startDate}} and {{endDate}}
${cohortQuery}
${joinSessionQuery}
where website_event.website_id = {{websiteId::uuid}}
and website_event.created_at between {{startDate}} and {{endDate}}
and website_event.event_name = {{eventName}}
${filterQuery}
group by website_event.event_id
order by max(website_event.created_at) desc
limit ${size} offset ${offset}
)
select
website_event.event_id as "eventId",
website_event.session_id as "sessionId",
website_event.event_name as "eventName",
website_event.url_path as "urlPath",
website_event.created_at as "createdAt",
event_data.data_key as "dataKey",
coalesce(
case when event_data.data_type = 1 then event_data.string_value end,
case when event_data.data_type = 2 then cast(event_data.number_value as varchar) end,
case when event_data.data_type = 4 then cast(event_data.date_value as varchar) end,
''
) as "value",
event_data.data_type as "dataType"
from event_data
join website_event on website_event.event_id = event_data.website_event_id
and website_event.website_id = {{websiteId::uuid}}
join paged_events on paged_events.event_id = event_data.website_event_id
where event_data.website_id = {{websiteId::uuid}}
and event_data.created_at between {{startDate}} and {{endDate}}
order by website_event.created_at desc
`,
{ ...queryParams, eventName },
FUNCTION_NAME,
);
// Pivot flat rows into one record per event
const eventMap = new Map<
string,
{ eventId: string; sessionId: string; eventName: string; urlPath: string; createdAt: Date; propertyKeys: string[]; propertyValues: string[] }
>();
for (const { eventId, sessionId, eventName: name, urlPath, createdAt, dataKey, value } of rows) {
if (!eventMap.has(eventId)) {
eventMap.set(eventId, { eventId, sessionId, eventName: name, urlPath, createdAt, propertyKeys: [], propertyValues: [] });
}
const entry = eventMap.get(eventId);
entry.propertyKeys.push(dataKey);
entry.propertyValues.push(value ?? '');
}
return { data: [...eventMap.values()], count, page: +page, pageSize: size };
}
async function clickhouseQuery(websiteId: string, eventName: string, filters: QueryFilters) {
const { rawQuery, parseFilters } = clickhouse;
const { page = 1, pageSize } = filters;
const size = +pageSize || DEFAULT_PAGE_SIZE;
const offset = +size * (+page - 1);
const { filterQuery, cohortQuery, queryParams } = parseFilters({ ...filters, websiteId });
const count = await rawQuery(
`
select count() as num
from umami.event_data_pivot
any left join (
select *
from website_event
where website_id = {websiteId:UUID}
and created_at between {startDate:DateTime64} and {endDate:DateTime64}
and event_type = 2
and event_name = {eventName:String}) website_event
on website_event.event_id = event_data_pivot.event_id
and website_event.session_id = event_data_pivot.session_id
and website_event.website_id = event_data_pivot.website_id
${cohortQuery}
where event_data_pivot.website_id = {websiteId:UUID}
and event_data_pivot.created_at between {startDate:DateTime64} and {endDate:DateTime64}
${filterQuery}
`,
{ ...queryParams, eventName },
).then((res: any) => res[0].num);
const data = await rawQuery(
`
select
event_id as eventId,
session_id as sessionId,
event_name as eventName,
url_path as urlPath,
created_at as createdAt,
groupArrayMerge(property_keys) as propertyKeys,
groupArrayMerge(property_values) as propertyValues
from umami.event_data_pivot
any left join (
select *
from website_event
where website_id = {websiteId:UUID}
and created_at between {startDate:DateTime64} and {endDate:DateTime64}
and event_type = 2
and event_name = {eventName:String}) website_event
on website_event.event_id = event_data_pivot.event_id
and website_event.session_id = event_data_pivot.session_id
and website_event.website_id = event_data_pivot.website_id
${cohortQuery}
where event_data_pivot.website_id = {websiteId:UUID}
and event_data_pivot.created_at between {startDate:DateTime64} and {endDate:DateTime64}
${filterQuery}
group by event_id, session_id, event_name, url_path, created_at
order by created_at desc
limit ${size} offset ${offset}
`,
{ ...queryParams, eventName },
FUNCTION_NAME,
);
return { data, count, page: +page, pageSize: size };
}
@@ -31,6 +31,7 @@ async function relationalQuery(
select
website_event.event_name as "eventName",
event_data.data_key as "propertyName",
event_data.data_type as "dataType",
count(*) as "total"
from event_data
join website_event on website_event.event_id = event_data.website_event_id
@@ -41,8 +42,8 @@ async function relationalQuery(
where event_data.website_id = {{websiteId::uuid}}
and event_data.created_at between {{startDate}} and {{endDate}}
${filterQuery}
group by website_event.event_name, event_data.data_key
order by 3 desc
group by website_event.event_name, event_data.data_key, event_data.data_type
order by 4 desc
limit 500
`,
queryParams,
@@ -53,7 +54,7 @@ async function relationalQuery(
async function clickhouseQuery(
websiteId: string,
filters: QueryFilters & { propertyName?: string },
): Promise<{ eventName: string; propertyName: string; total: number }[]> {
): Promise<{ eventName: string; propertyName: string; dataType: number; total: number }[]> {
const { rawQuery, parseFilters } = clickhouse;
const { filterQuery, cohortQuery, queryParams } = parseFilters(
{ ...filters, websiteId },
@@ -67,6 +68,7 @@ async function clickhouseQuery(
select
event_name as eventName,
data_key as propertyName,
data_type as dataType,
count(*) as total
from event_data
any left join (
@@ -82,8 +84,8 @@ async function clickhouseQuery(
where event_data.website_id = {websiteId:UUID}
and event_data.created_at between {startDate:DateTime64} and {endDate:DateTime64}
${filterQuery}
group by event_name, data_key
order by 1, 3 desc
group by event_name, data_key, data_type
order by 1, 4 desc
limit 500
`,
queryParams,
@@ -0,0 +1,96 @@
import clickhouse from '@/lib/clickhouse';
import { CLICKHOUSE, PRISMA, runQuery } from '@/lib/db';
import prisma from '@/lib/prisma';
import type { QueryFilters } from '@/lib/types';
const FUNCTION_NAME = 'getEventDataPropertySeries';
export async function getEventDataPropertySeries(
...args: [websiteId: string, eventName: string, propertyName: string, filters: QueryFilters]
) {
return runQuery({
[PRISMA]: () => relationalQuery(...args),
[CLICKHOUSE]: () => clickhouseQuery(...args),
});
}
async function relationalQuery(
websiteId: string,
eventName: string,
propertyName: string,
filters: QueryFilters,
) {
const { timezone = 'utc', unit = 'day' } = filters;
const { rawQuery, getDateSQL, parseFilters } = prisma;
const { filterQuery, cohortQuery, joinSessionQuery, queryParams } = parseFilters({
...filters,
websiteId,
});
return rawQuery(
`
select
event_data.string_value as x,
${getDateSQL('event_data.created_at', unit, timezone)} t,
count(*) y
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}}
and website_event.event_type = 2
and website_event.event_name = {{eventName}}
${cohortQuery}
${joinSessionQuery}
where event_data.website_id = {{websiteId::uuid}}
and event_data.created_at between {{startDate}} and {{endDate}}
and event_data.data_key = {{propertyName}}
and event_data.data_type = 1
${filterQuery}
group by 1, 2
order by 2
`,
{ ...queryParams, eventName, propertyName },
FUNCTION_NAME,
);
}
async function clickhouseQuery(
websiteId: string,
eventName: string,
propertyName: string,
filters: QueryFilters,
): Promise<{ x: string; t: string; y: number }[]> {
const { timezone = 'UTC', unit = 'day' } = filters;
const { rawQuery, getDateSQL, parseFilters } = clickhouse;
const { filterQuery, cohortQuery, queryParams } = parseFilters({ ...filters, websiteId });
return rawQuery(
`
select
event_data.string_value as x,
${getDateSQL('event_data.created_at', unit, timezone)} as t,
count() as y
from event_data
any left join (
select *
from website_event
where website_id = {websiteId:UUID}
and created_at between {startDate:DateTime64} and {endDate:DateTime64}
and event_type = 2
and event_name = {eventName:String}) website_event
on website_event.event_id = event_data.event_id
and website_event.session_id = event_data.session_id
and website_event.website_id = event_data.website_id
${cohortQuery}
where event_data.website_id = {websiteId:UUID}
and event_data.created_at between {startDate:DateTime64} and {endDate:DateTime64}
and event_data.data_key = {propertyName:String}
and event_data.data_type = 1
${filterQuery}
group by x, t
order by t
`,
{ ...queryParams, eventName, propertyName },
FUNCTION_NAME,
);
}
+3
View File
@@ -1,7 +1,10 @@
export * from './events/getEventDataById';
export * from './events/getEventDataEvents';
export * from './events/getEventDataFields';
export * from './events/getEventDataNumericSeries';
export * from './events/getEventDataPivot';
export * from './events/getEventDataProperties';
export * from './events/getEventDataPropertySeries';
export * from './events/getEventDataStats';
export * from './events/getEventDataUsage';
export * from './events/getEventDataValues';