finish expanded queries and ui.

This commit is contained in:
Francis Cao 2025-08-07 09:47:18 -07:00
parent 0a0c1f27c6
commit 38f251ead5
12 changed files with 119 additions and 113 deletions

View file

@ -1,5 +1,5 @@
import clickhouse from '@/lib/clickhouse';
import { EVENT_TYPE, FILTER_COLUMNS, 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 { QueryFilters } from '@/lib/types';
@ -99,7 +99,7 @@ async function clickhouseQuery(
filters: QueryFilters,
): Promise<{ x: string; y: number }[]> {
const { type, limit = 500, offset = 0 } = parameters;
const column = FILTER_COLUMNS[type] || type;
let column = FILTER_COLUMNS[type] || type;
const { rawQuery, parseFilters } = clickhouse;
const { filterQuery, cohortQuery, queryParams } = parseFilters({
...filters,
@ -112,21 +112,24 @@ async function clickhouseQuery(
if (column === 'referrer_domain') {
excludeDomain = `and referrer_domain != hostname and referrer_domain != ''`;
if (type === 'grouped') {
column = toClickHouseGroupedReferrer(GROUPED_DOMAINS);
}
}
if (type === 'entry' || type === 'exit') {
const aggregrate = type === 'entry' ? 'min' : 'max';
const aggregrate = type === 'entry' ? 'argMin' : 'argMax';
column = `x.${column}`;
entryExitQuery = `
JOIN (select visit_id,
${aggregrate}(created_at) target_created_at
${aggregrate}(url_path, created_at) url_path
from website_event
where website_id = {websiteId:UUID}
and created_at between {startDate:DateTime64} and {endDate:DateTime64}
and event_type = {eventType:UInt32}
group by visit_id) x
ON x.visit_id = website_event.visit_id
and x.target_created_at = website_event.created_at`;
ON x.visit_id = website_event.visit_id`;
}
return rawQuery(
@ -164,3 +167,19 @@ async function clickhouseQuery(
{ ...queryParams, ...parameters },
);
}
export function toClickHouseGroupedReferrer(
domains: any[],
column: string = 'referrer_domain',
): string {
return [
'CASE',
...domains.map(group => {
const matches = Array.isArray(group.match) ? group.match : [group.match];
const formattedArray = matches.map(m => `'${m}'`).join(', ');
return ` WHEN multiSearchAny(${column}, [${formattedArray}]) != 0 THEN '${group.domain}'`;
}),
" ELSE 'Other'",
'END',
].join('\n');
}

View file

@ -30,7 +30,7 @@ async function relationalQuery(
filters: QueryFilters,
): Promise<PageviewMetricsData[]> {
const { type, limit = 500, offset = 0 } = parameters;
const column = FILTER_COLUMNS[type] || type;
let column = FILTER_COLUMNS[type] || type;
const { rawQuery, parseFilters } = prisma;
const { filterQuery, joinSessionQuery, cohortQuery, queryParams } = parseFilters(
{
@ -50,20 +50,21 @@ async function relationalQuery(
}
if (type === 'entry' || type === 'exit') {
const aggregrate = type === 'entry' ? 'min' : 'max';
const order = type === 'entry' ? 'asc' : 'desc';
column = `x.${column}`;
entryExitQuery = `
join (
select visit_id,
${aggregrate}(created_at) target_created_at
select distinct on (visit_id)
visit_id,
url_path
from website_event
where website_event.website_id = {{websiteId::uuid}}
and website_event.created_at between {{startDate}} and {{endDate}}
and event_type = {{eventType}}
group by visit_id
order by visit_id, created_at ${order}
) x
on x.visit_id = website_event.visit_id
and x.target_created_at = website_event.created_at
`;
}
@ -95,7 +96,7 @@ async function clickhouseQuery(
filters: QueryFilters,
): Promise<{ x: string; y: number }[]> {
const { type, limit = 500, offset = 0 } = parameters;
const column = FILTER_COLUMNS[type] || type;
let column = FILTER_COLUMNS[type] || type;
const { rawQuery, parseFilters } = clickhouse;
const { filterQuery, cohortQuery, queryParams } = parseFilters({
...filters,
@ -114,18 +115,18 @@ async function clickhouseQuery(
}
if (type === 'entry' || type === 'exit') {
const aggregrate = type === 'entry' ? 'min' : 'max';
const aggregrate = type === 'entry' ? 'argMin' : 'argMax';
column = `x.${column}`;
entryExitQuery = `
JOIN (select visit_id,
${aggregrate}(created_at) target_created_at
${aggregrate}(url_path, created_at) url_path
from website_event
where website_id = {websiteId:UUID}
and created_at between {startDate:DateTime64} and {endDate:DateTime64}
and event_type = {eventType:UInt32}
group by visit_id) x
ON x.visit_id = website_event.visit_id
and x.target_created_at = website_event.created_at`;
ON x.visit_id = website_event.visit_id`;
}
sql = `