remove parse session filters

This commit is contained in:
Francis Cao 2024-08-01 22:57:54 -07:00
parent 3207b0ce06
commit d02dc6f997
4 changed files with 75 additions and 69 deletions

View File

@ -2,11 +2,11 @@ import { ClickHouseClient, createClient } from '@clickhouse/client';
import dateFormat from 'dateformat';
import debug from 'debug';
import { CLICKHOUSE } from 'lib/db';
import { PageParams, QueryFilters, QueryOptions } from './types';
import { EVENT_COLUMNS, DEFAULT_PAGE_SIZE, OPERATORS } from './constants';
import { fetchWebsite } from './load';
import { DEFAULT_PAGE_SIZE, OPERATORS } from './constants';
import { maxDate } from './date';
import { fetchWebsite } from './load';
import { filtersToArray } from './params';
import { PageParams, QueryFilters, QueryOptions } from './types';
export const CLICKHOUSE_DATE_FORMATS = {
second: '%Y-%m-%dT%H:%i:%S',
@ -100,26 +100,6 @@ function getFilterQuery(filters: QueryFilters = {}, options: QueryOptions = {})
return query.join('\n');
}
function getSessionFilterQuery(filters: QueryFilters = {}, options: QueryOptions = {}) {
const query = filtersToArray(filters, options).reduce((arr, { name, column, operator }) => {
if (column) {
if (EVENT_COLUMNS.includes(name)) {
arr.push(`and has(${column}, {${name}:String})`);
if (name === 'referrer') {
arr.push('and not has(referrer_domain, {websiteDomain:String})');
}
} else {
arr.push(`and ${mapFilter(column, operator, name)}`);
}
}
return arr;
}, []);
return query.join('\n');
}
function getDateQuery(filters: QueryFilters = {}) {
const { startDate, endDate } = filters;
@ -159,25 +139,6 @@ async function parseFilters(websiteId: string, filters: QueryFilters = {}, optio
};
}
async function parseSessionFilters(
websiteId: string,
filters: QueryFilters = {},
options?: QueryOptions,
) {
const website = await fetchWebsite(websiteId);
return {
filterQuery: getSessionFilterQuery(filters, options),
dateQuery: getDateQuery(filters),
params: {
...getFilterParams(filters),
websiteId,
startDate: maxDate(filters.startDate, new Date(website?.resetAt)),
websiteDomain: website.domain,
},
};
}
async function pagedQuery(
query: string,
queryParams: { [key: string]: any },
@ -260,7 +221,6 @@ export default {
getDateFormat,
getFilterQuery,
parseFilters,
parseSessionFilters,
pagedQuery,
findUnique,
findFirst,

View File

@ -74,8 +74,8 @@ async function clickhouseQuery(
sql = `
select
sum(t.c) as "pageviews",
count(distinct t.session_id) as "visitors",
count(distinct t.visit_id) as "visits",
uniq(t.session_id) as "visitors",
uniq(t.visit_id) as "visits",
sum(if(t.c = 1, 1, 0)) as "bounces",
sum(max_time-min_time) as "totaltime"
from (
@ -96,16 +96,24 @@ async function clickhouseQuery(
} else {
sql = `
select
sum(views) as "pageviews",
sum(t.c) as "pageviews",
uniq(session_id) as "visitors",
uniq(visit_id) as "visits",
sumIf(1, views = 1) as "bounces",
sumIf(1, t.c = 1) as "bounces",
sum(max_time-min_time) as "totaltime"
from website_event_stats_hourly "website_event"
from (select
session_id,
visit_id,
sum(views) c,
min(min_time) min_time,
max(max_time) max_time
from umami.website_event_stats_hourly "website_event"
where website_id = {websiteId:UUID}
and created_at between {startDate:DateTime64} and {endDate:DateTime64}
and event_type = {eventType:UInt32}
${filterQuery};
${filterQuery}
group by session_id, visit_id
) as t;
`;
}

View File

@ -1,5 +1,5 @@
import clickhouse from 'lib/clickhouse';
import { EVENT_TYPE, FILTER_COLUMNS, SESSION_COLUMNS } from 'lib/constants';
import { EVENT_COLUMNS, EVENT_TYPE, FILTER_COLUMNS, SESSION_COLUMNS } from 'lib/constants';
import { CLICKHOUSE, PRISMA, runQuery } from 'lib/db';
import prisma from 'lib/prisma';
import { QueryFilters } from 'lib/types';
@ -64,15 +64,34 @@ async function clickhouseQuery(
offset: number = 0,
): Promise<{ x: string; y: number }[]> {
const column = FILTER_COLUMNS[type] || type;
const { parseSessionFilters, rawQuery } = clickhouse;
const { filterQuery, params } = await parseSessionFilters(websiteId, {
const { parseFilters, rawQuery } = clickhouse;
const { filterQuery, params } = await parseFilters(websiteId, {
...filters,
eventType: EVENT_TYPE.pageView,
});
const includeCountry = column === 'city' || column === 'subdivision1';
return rawQuery(
`
let sql = '';
if (EVENT_COLUMNS.some(item => Object.keys(filters).includes(item))) {
sql = `
select
${column} x,
count(distinct session_id) y
${includeCountry ? ', country' : ''}
from website_event
where website_id = {websiteId:UUID}
and created_at between {startDate:DateTime64} and {endDate:DateTime64}
and event_type = {eventType:UInt32}
${filterQuery}
group by x
${includeCountry ? ', country' : ''}
order by y desc
limit ${limit}
offset ${offset}
`;
} else {
sql = `
select
${column} x,
uniq(session_id) y
@ -87,9 +106,10 @@ async function clickhouseQuery(
order by y desc
limit ${limit}
offset ${offset}
`,
params,
).then(a => {
`;
}
return rawQuery(sql, params).then(a => {
return Object.values(a).map(a => {
return { x: a.x, y: Number(a.y), country: a.country };
});

View File

@ -1,7 +1,7 @@
import clickhouse from 'lib/clickhouse';
import { EVENT_COLUMNS, EVENT_TYPE } from 'lib/constants';
import { CLICKHOUSE, PRISMA, runQuery } from 'lib/db';
import prisma from 'lib/prisma';
import { EVENT_TYPE } from 'lib/constants';
import { QueryFilters } from 'lib/types';
export async function getSessionStats(...args: [websiteId: string, filters: QueryFilters]) {
@ -41,25 +41,24 @@ async function clickhouseQuery(
filters: QueryFilters,
): Promise<{ x: string; y: number }[]> {
const { timezone = 'UTC', unit = 'day' } = filters;
const { parseSessionFilters, rawQuery, getDateStringSQL, getDateSQL } = clickhouse;
const { filterQuery, params } = await parseSessionFilters(websiteId, {
const { parseFilters, rawQuery, getDateStringSQL, getDateSQL } = clickhouse;
const { filterQuery, params } = await parseFilters(websiteId, {
...filters,
eventType: EVENT_TYPE.pageView,
});
const table = unit === 'minute' ? 'website_event' : 'website_event_stats_hourly';
const columnQuery = unit === 'minute' ? 'count(distinct session_id)' : 'uniq(session_id)';
let sql = '';
return rawQuery(
`
if (EVENT_COLUMNS.some(item => Object.keys(filters).includes(item)) || unit === 'minute') {
sql = `
select
${getDateStringSQL('g.t', unit)} as x,
g.y as y
from (
select
${getDateSQL('created_at', unit, timezone)} as t,
${columnQuery} as y
from ${table} website_event
count(distinct session_id) as y
from website_event
where website_id = {websiteId:UUID}
and created_at between {startDate:DateTime64} and {endDate:DateTime64}
and event_type = {eventType:UInt32}
@ -67,9 +66,28 @@ async function clickhouseQuery(
group by t
) as g
order by t
`,
params,
).then(result => {
`;
} else {
sql = `
select
${getDateStringSQL('g.t', unit)} as x,
g.y as y
from (
select
${getDateSQL('created_at', unit, timezone)} as t,
uniq(session_id) as y
from website_event_stats_hourly website_event
where website_id = {websiteId:UUID}
and created_at between {startDate:DateTime64} and {endDate:DateTime64}
and event_type = {eventType:UInt32}
${filterQuery}
group by t
) as g
order by t
`;
}
return rawQuery(sql, params).then(result => {
return Object.values(result).map((a: any) => {
return { x: a.x, y: Number(a.y) };
});