Merge Session/Event/Pageview CH

This commit is contained in:
Brian Cao 2022-09-12 09:55:34 -07:00
parent d377ef86e7
commit 52e036964b
18 changed files with 237 additions and 294 deletions

View file

@ -29,7 +29,7 @@ function clickhouseQuery(websites, start_at) {
return rawQuery(
`select
event_id,
event_uuid,
website_id,
session_id,
created_at,

View file

@ -33,17 +33,19 @@ async function relationalQuery(website_id, { session_id, url, event_name, event_
async function clickhouseQuery(
website_id,
{ event_uuid, session_uuid, url, event_name, event_data },
{ session: { country, ...sessionArgs }, event_uuid, url, event_name, event_data },
) {
const { getDateFormat, sendMessage } = kafka;
const params = {
event_uuid,
website_id,
session_uuid,
created_at: getDateFormat(new Date()),
url: url?.substring(0, URL_LENGTH),
event_name: event_name?.substring(0, EVENT_NAME_LENGTH),
event_data: JSON.stringify(event_data),
...sessionArgs,
country: country ? country : null,
};
await sendMessage(params, 'event');

View file

@ -9,9 +9,9 @@ export async function getPageviewMetrics(...args) {
});
}
async function relationalQuery(website_id, start_at, end_at, column, table, filters = {}) {
async function relationalQuery(website_id, { startDate, endDate, column, table, filters = {} }) {
const { rawQuery, parseFilters } = prisma;
const params = [website_id, start_at, end_at];
const params = [website_id, startDate, endDate];
const { pageviewQuery, sessionQuery, eventQuery, joinSession } = parseFilters(
table,
column,
@ -34,26 +34,20 @@ async function relationalQuery(website_id, start_at, end_at, column, table, filt
);
}
async function clickhouseQuery(website_id, start_at, end_at, column, table, filters = {}) {
async function clickhouseQuery(website_id, { startDate, endDate, column, filters = {} }) {
const { rawQuery, parseFilters, getBetweenDates } = clickhouse;
const params = [website_id];
const { pageviewQuery, sessionQuery, eventQuery, joinSession } = parseFilters(
table,
column,
filters,
params,
'session_uuid',
);
const { pageviewQuery, sessionQuery, eventQuery } = parseFilters(column, filters, params);
return rawQuery(
`select ${column} x, count(*) y
from ${table}
${joinSession}
where ${table}.website_id= $1
and ${getBetweenDates(table + '.created_at', start_at, end_at)}
${pageviewQuery}
${joinSession && sessionQuery}
${eventQuery}
from event
where website_id= $1
${column !== 'event_name' ? `and event_name = ''` : ''}
and ${getBetweenDates('created_at', startDate, endDate)}
${pageviewQuery}
${sessionQuery}
${eventQuery}
group by x
order by y desc`,
params,

View file

@ -11,13 +11,15 @@ export async function getPageviewStats(...args) {
async function relationalQuery(
website_id,
start_at,
end_at,
timezone = 'utc',
unit = 'day',
count = '*',
filters = {},
sessionKey = 'session_id',
{
start_at,
end_at,
timezone = 'utc',
unit = 'day',
count = '*',
filters = {},
sessionKey = 'session_id',
},
) {
const { getDateQuery, parseFilters, rawQuery } = prisma;
const params = [website_id, start_at, end_at];
@ -44,23 +46,11 @@ async function relationalQuery(
async function clickhouseQuery(
website_id,
start_at,
end_at,
timezone = 'UTC',
unit = 'day',
count = '*',
filters = {},
sessionKey = 'session_uuid',
{ start_at, end_at, timezone = 'UTC', unit = 'day', count = '*', filters = {} },
) {
const { parseFilters, rawQuery, getDateStringQuery, getDateQuery, getBetweenDates } = clickhouse;
const params = [website_id];
const { pageviewQuery, sessionQuery, joinSession } = parseFilters(
'pageview',
null,
filters,
params,
sessionKey,
);
const { pageviewQuery, sessionQuery } = parseFilters(null, filters, params);
return rawQuery(
`select
@ -69,11 +59,11 @@ async function clickhouseQuery(
from
(select
${getDateQuery('created_at', unit, timezone)} t,
count(${count !== '*' ? `${count}${sessionKey}` : count}) y
from pageview
${joinSession}
where pageview.website_id= $1
and ${getBetweenDates('pageview.created_at', start_at, end_at)}
count(${count !== '*' ? 'session_uuid' : count}) y
from event
where website_id= $1
and ${getBetweenDates('created_at', start_at, end_at)}
${pageviewQuery}
${sessionQuery}
group by t) g

View file

@ -32,8 +32,9 @@ async function clickhouseQuery(websites, start_at) {
session_id,
created_at,
url
from pageview
where website_id in (${websites.join[',']}
from event
where event_name = ''
and website_id in (${websites.join[',']}
and created_at >= ${clickhouse.getDateFormat(start_at)})`,
);
}

View file

@ -10,7 +10,7 @@ export async function savePageView(...args) {
});
}
async function relationalQuery(website_id, { session_id, url, referrer }) {
async function relationalQuery(website_id, { session: { session_id }, url, referrer }) {
return prisma.client.pageview.create({
data: {
website_id,
@ -21,15 +21,19 @@ async function relationalQuery(website_id, { session_id, url, referrer }) {
});
}
async function clickhouseQuery(website_id, { session_uuid, url, referrer }) {
async function clickhouseQuery(
website_id,
{ session: { country, ...sessionArgs }, url, referrer },
) {
const { getDateFormat, sendMessage } = kafka;
const params = {
website_id: website_id,
session_uuid: session_uuid,
created_at: getDateFormat(new Date()),
url: url?.substring(0, URL_LENGTH),
referrer: referrer?.substring(0, URL_LENGTH),
...sessionArgs,
country: country ? country : null,
};
await sendMessage(params, 'pageview');
await sendMessage(params, 'event');
}

View file

@ -19,6 +19,14 @@ async function relationalQuery(website_id, data) {
},
select: {
session_id: true,
session_uuid: true,
hostname: true,
browser: true,
os: true,
screen: true,
language: true,
country: true,
device: true,
},
})
.then(async res => {
@ -35,20 +43,21 @@ async function clickhouseQuery(
{ session_uuid, hostname, browser, os, screen, language, country, device },
) {
const { getDateFormat, sendMessage } = kafka;
const params = {
session_uuid: session_uuid,
website_id: website_id,
session_uuid,
website_id,
created_at: getDateFormat(new Date()),
hostname: hostname,
browser: browser,
os: os,
device: device,
screen: screen,
language: language,
hostname,
browser,
os,
device,
screen,
language,
country: country ? country : null,
};
await sendMessage(params, 'session');
await sendMessage(params, 'event');
if (redis.client) {
await redis.client.set(`session:${session_uuid}`, 1);

View file

@ -31,7 +31,7 @@ async function clickhouseQuery(session_uuid) {
const params = [session_uuid];
return rawQuery(
`select
`select distinct
session_uuid,
website_id,
created_at,
@ -42,7 +42,7 @@ async function clickhouseQuery(session_uuid) {
screen,
language,
country
from session
from event
where session_uuid = $1`,
params,
)

View file

@ -9,15 +9,10 @@ export async function getSessionMetrics(...args) {
});
}
async function relationalQuery(website_id, start_at, end_at, field, filters = {}) {
async function relationalQuery(website_id, { startDate, endDate, field, filters = {} }) {
const { parseFilters, rawQuery } = prisma;
const params = [website_id, start_at, end_at];
const { pageviewQuery, sessionQuery, joinSession } = parseFilters(
'pageview',
null,
filters,
params,
);
const params = [website_id, startDate, endDate];
const { pageviewQuery, sessionQuery, joinSession } = parseFilters(null, filters, params);
return rawQuery(
`select ${field} x, count(*) y
@ -37,29 +32,19 @@ async function relationalQuery(website_id, start_at, end_at, field, filters = {}
);
}
async function clickhouseQuery(website_id, start_at, end_at, field, filters = {}) {
async function clickhouseQuery(website_id, { startDate, endDate, field, filters = {} }) {
const { parseFilters, getBetweenDates, rawQuery } = clickhouse;
const params = [website_id];
const { pageviewQuery, sessionQuery, joinSession } = parseFilters(
'pageview',
null,
filters,
params,
'session_uuid',
);
const { pageviewQuery, sessionQuery } = parseFilters(null, filters, params);
return rawQuery(
`select ${field} x, count(*) y
from session as x
where x.session_uuid in (
select pageview.session_uuid
from pageview
${joinSession}
where pageview.website_id=$1
and ${getBetweenDates('pageview.created_at', start_at, end_at)}
from event as x
where website_id=$1
and event_name = ''
and ${getBetweenDates('created_at', startDate, endDate)}
${pageviewQuery}
${sessionQuery}
)
group by x
order by y desc`,
params,

View file

@ -32,7 +32,7 @@ async function clickhouseQuery(websites, start_at) {
const { rawQuery, getDateFormat } = clickhouse;
return rawQuery(
`select
`select distinct
session_uuid,
website_id,
created_at,
@ -43,7 +43,7 @@ async function clickhouseQuery(websites, start_at) {
screen,
language,
country
from session
from event
where ${websites && websites.length > 0 ? `(website_id in (${websites.join[',']})` : '0 = 0'}
and created_at >= ${getDateFormat(start_at)}`,
);

View file

@ -29,7 +29,7 @@ async function clickhouseQuery(website_id) {
return rawQuery(
`select count(distinct session_uuid) x
from pageview
from event
where website_id = $1
and created_at >= ${getDateFormat(subMinutes(new Date(), 5))}`,
params,

View file

@ -44,13 +44,7 @@ async function relationalQuery(website_id, start_at, end_at, filters = {}) {
async function clickhouseQuery(website_id, start_at, end_at, filters = {}) {
const { rawQuery, getDateQuery, getBetweenDates, parseFilters } = clickhouse;
const params = [website_id];
const { pageviewQuery, sessionQuery, joinSession } = parseFilters(
'pageview',
null,
filters,
params,
'session_uuid',
);
const { pageviewQuery, sessionQuery } = parseFilters(null, filters, params);
return rawQuery(
`select
@ -59,18 +53,18 @@ async function clickhouseQuery(website_id, start_at, end_at, filters = {}) {
sum(if(t.c = 1, 1, 0)) as "bounces",
sum(if(max_time < min_time + interval 1 hour, max_time-min_time, 0)) as "totaltime"
from (
select pageview.session_uuid,
${getDateQuery('pageview.created_at', 'day')} time_series,
select session_uuid,
${getDateQuery('created_at', 'day')} time_series,
count(*) c,
min(created_at) min_time,
max(created_at) max_time
from pageview
${joinSession}
where pageview.website_id = $1
and ${getBetweenDates('pageview.created_at', start_at, end_at)}
from event
where event_name = ''
and website_id = $1
and ${getBetweenDates('created_at', start_at, end_at)}
${pageviewQuery}
${sessionQuery}
group by pageview.session_uuid, time_series
group by session_uuid, time_series
) t;`,
params,
);