convert analytics queries

lib/constants.js

@@ -64,6 +64,7 @@ export const EVENT_COLORS = [
   '#ffec16',
 ];
 
+export const RELATIONAL = 'relational';
 export const POSTGRESQL = 'postgresql';
 export const MYSQL = 'mysql';
 export const CLICKHOUSE = 'clickhouse';

lib/db.js

@@ -7,8 +7,10 @@ import {
   POSTGRESQL,
   POSTGRESQL_DATE_FORMATS,
   CLICKHOUSE,
+  RELATIONAL,
 } from 'lib/constants';
 import moment from 'moment-timezone';
+import { CLICKHOUSE_DATE_FORMATS } from './constants';
 
 BigInt.prototype.toJSON = function () {
   return Number(this);
@@ -48,19 +50,48 @@ function initializePrisma(options) {
 }
 
 function initializeClickhouse() {
-  if (process.env.ANALYTICS_URL) {
+  if (!process.env.ANALYTICS_URL) {
     return null;
   }
 
+  const url = new URL(process.env.ANALYTICS_URL);
+
+  const database = url.pathname.replace('/', '');
+
   return new ClickHouse({
-    url: process.env.ANALYTICS_URL,
+    url: url.hostname,
+    port: Number(url.port),
+    basicAuth: url.password
+      ? {
+          username: url.username || 'default',
+          password: url.password,
+        }
+      : null,
     format: 'json',
+    config: {
+      database,
+    },
   });
+
+  // return new ClickHouse({
+  //   url: 'http://164.92.95.2',
+  //   port: 8123,
+  //   basicAuth: {
+  //     username: 'default',
+  //     password: 'shhhthisissupersecret!',
+  //   },
+  //   format: 'json',
+  //   config: {
+  //     database: 'umami_dev',
+  //   },
+  // });
 }
 
 const prisma = initializePrisma(options);
 const clickhouse = initializeClickhouse();
 
+console.log('clickhouse1: ', clickhouse);
+
 export { prisma, clickhouse };
 
 export function getDatabase() {
@@ -103,6 +134,10 @@ export function getDateStringQuery(data, unit) {
   }
 }
 
+export function getDateStringQueryClickhouse(data, unit) {
+  return `formatDateTime(${data}, '${CLICKHOUSE_DATE_FORMATS[unit]}')`;
+}
+
 export function getDateQuery(field, unit, timezone) {
   const db = getDatabase();
 
@@ -124,15 +159,31 @@ export function getDateQuery(field, unit, timezone) {
   }
 }
 
-export function getTimestampInterval(field) {
+export function getDateQueryClickhouse(field, unit, timezone) {
+  if (timezone) {
+    return `date_trunc('${unit}', ${field},'${timezone}')`;
+  }
+  return `date_trunc('${unit}', ${field})`;
+}
+
+export function getDateFormatClickhouse(date) {
+  return `parseDateTimeBestEffort('${date.toUTCString()}')`;
+}
+
+export function getBetweenDatesClickhouse(field, start_at, end_at) {
+  return `${field} between ${getDateFormatClickhouse(start_at)} 
+    and ${getDateFormatClickhouse(end_at)}`;
+}
+
+export function getTimestampInterval(maxColumn, minColumn) {
   const db = getDatabase();
 
   if (db === POSTGRESQL) {
-    return `floor(extract(epoch from max(${field}) - min(${field})))`;
+    return `floor(extract(epoch from max(${maxColumn}) - min(${minColumn})))`;
   }
 
   if (db === MYSQL) {
-    return `floor(unix_timestamp(max(${field})) - unix_timestamp(min(${field})))`;
+    return `floor(unix_timestamp(max(${maxColumn})) - unix_timestamp(min(${minColumn})))`;
   }
 }
 
@@ -149,6 +200,7 @@ export function getFilterQuery(table, filters = {}, params = []) {
         if (table === 'pageview' || table === 'event') {
           arr.push(`and ${table}.${key}=$${params.length + 1}`);
           params.push(decodeURIComponent(value));
+          console.log(params);
         }
         break;
 
@@ -213,6 +265,22 @@ export function parseFilters(table, filters = {}, params = []) {
   };
 }
 
+export function replaceQueryClickhouse(string, params = []) {
+  let formattedString = string;
+
+  params.forEach((a, i) => {
+    let replace = a;
+
+    if (typeof a === 'string' || a instanceof String) {
+      replace = `'${replace}'`;
+    }
+
+    formattedString = formattedString.replace(`$${i + 1}`, replace);
+  });
+
+  return formattedString;
+}
+
 export async function runQuery(query) {
   return query.catch(e => {
     throw e;
@@ -231,14 +299,24 @@ export async function rawQuery(query, params = []) {
   return runQuery(prisma.$queryRawUnsafe.apply(prisma, [sql, ...params]));
 }
 
-export function runAnalyticsQuery(relational, clickhouse) {
+export async function rawQueryClickhouse(query, params = [], debug = false) {
+  let formattedQuery = replaceQueryClickhouse(query, params);
+
+  if (debug || process.env.LOG_QUERY) {
+    console.log(formattedQuery);
+  }
+
+  return clickhouse.query(formattedQuery).toPromise();
+}
+
+export async function runAnalyticsQuery(queries) {
   const db = getAnalyticsDatabase();
 
   if (db === POSTGRESQL || db === MYSQL) {
-    return relational();
+    return queries[`${RELATIONAL}`]();
   }
 
   if (db === CLICKHOUSE) {
-    return runQuery(clickhouse());
+    return queries[`${CLICKHOUSE}`]();
   }
 }

queries/analytics/event/getEventMetrics.js

@@ -1,17 +1,23 @@
+import { CLICKHOUSE, RELATIONAL } from 'lib/constants';
 import {
+  rawQueryClickhouse,
+  getBetweenDatesClickhouse,
   getDateQuery,
+  getDateQueryClickhouse,
   getDateStringQuery,
   getFilterQuery,
   rawQuery,
   runAnalyticsQuery,
-  clickhouse,
 } from 'lib/db';
 
-export function getEventMetrics(...args) {
-  return runAnalyticsQuery(relationalQuery(...args), clickhouseQuery(...args));
+export async function getEventMetrics(...args) {
+  return runAnalyticsQuery({
+    [`${RELATIONAL}`]: () => relationalQuery(...args),
+    [`${CLICKHOUSE}`]: () => clickhouseQuery(...args),
+  });
 }
 
-function relationalQuery(
+async function relationalQuery(
   website_id,
   start_at,
   end_at,
@@ -38,28 +44,28 @@ function relationalQuery(
   );
 }
 
-function clickhouseQuery(
+async function clickhouseQuery(
   website_id,
   start_at,
   end_at,
-  timezone = 'utc',
+  timezone = 'UTC',
   unit = 'day',
   filters = {},
 ) {
-  const params = [website_id, start_at, end_at];
+  const params = [website_id];
 
-  return clickhouse.query(
+  return rawQueryClickhouse(
     `
     select
       event_value x,
-      ${getDateStringQuery(getDateQuery('created_at', unit, timezone), unit)} t,
+      ${getDateQueryClickhouse('created_at', unit, timezone)} t,
       count(*) y
     from event
-    where website_id=$1
-    and created_at between $2 and $3
+    where website_id= $1
+      and ${getBetweenDatesClickhouse('created_at', start_at, end_at)}
       ${getFilterQuery('event', filters, params)}
-    group by 1, 2
-    order by 2
+    group by x, t
+    order by t
     `,
     params,
   );

queries/analytics/event/getEvents.js

@@ -1,6 +1,20 @@
-import { prisma, runQuery } from 'lib/db';
+import { CLICKHOUSE, RELATIONAL } from 'lib/constants';
+import {
+  rawQueryClickhouse,
+  getDateFormatClickhouse,
+  prisma,
+  runAnalyticsQuery,
+  runQuery,
+} from 'lib/db';
 
-export async function getEvents(websites, start_at) {
+export function getEvents(...args) {
+  return runAnalyticsQuery({
+    [`${RELATIONAL}`]: () => relationalQuery(...args),
+    [`${CLICKHOUSE}`]: () => clickhouseQuery(...args),
+  });
+}
+
+function relationalQuery(websites, start_at) {
   return runQuery(
     prisma.event.findMany({
       where: {
@@ -16,3 +30,20 @@ export async function getEvents(websites, start_at) {
     }),
   );
 }
+
+function clickhouseQuery(websites, start_at) {
+  return rawQueryClickhouse(
+    `
+    select
+      event_id,
+      website_id, 
+      session_id,
+      created_at,
+      url,
+      event_type
+    from event
+    where website_id in (${websites.join[',']}
+      and created_at >= ${getDateFormatClickhouse(start_at)})
+    `,
+  );
+}

queries/analytics/pageview/getPageviewMetrics.js

@@ -1,6 +1,20 @@
-import { parseFilters, rawQuery } from 'lib/db';
+import { CLICKHOUSE, RELATIONAL } from 'lib/constants';
+import {
+  rawQueryClickhouse,
+  runAnalyticsQuery,
+  parseFilters,
+  rawQuery,
+  getBetweenDatesClickhouse,
+} from 'lib/db';
 
-export function getPageviewMetrics(website_id, start_at, end_at, field, table, filters = {}) {
+export async function getPageviewMetrics(...args) {
+  return runAnalyticsQuery({
+    [`${RELATIONAL}`]: () => relationalQuery(...args),
+    [`${CLICKHOUSE}`]: () => clickhouseQuery(...args),
+  });
+}
+
+async function relationalQuery(website_id, start_at, end_at, field, table, filters = {}) {
   const params = [website_id, start_at, end_at];
   const { pageviewQuery, sessionQuery, eventQuery, joinSession } = parseFilters(
     table,
@@ -24,3 +38,28 @@ export function getPageviewMetrics(website_id, start_at, end_at, field, table, f
     params,
   );
 }
+
+async function clickhouseQuery(website_id, start_at, end_at, field, table, filters = {}) {
+  const params = [website_id];
+  const { pageviewQuery, sessionQuery, eventQuery, joinSession } = parseFilters(
+    table,
+    filters,
+    params,
+  );
+
+  return rawQueryClickhouse(
+    `
+    select ${field} x, count(*) y
+    from ${table}
+      ${joinSession}
+    where ${table}.website_id= $1
+    and ${getBetweenDatesClickhouse(table + '.created_at', start_at, end_at)}
+    ${pageviewQuery}
+    ${joinSession && sessionQuery}
+    ${eventQuery}
+    group by x
+    order by y desc
+    `,
+    params,
+  );
+}

queries/analytics/pageview/getPageviewStats.js

@@ -1,6 +1,24 @@
-import { parseFilters, rawQuery, getDateQuery, getDateStringQuery } from 'lib/db';
+import { CLICKHOUSE, RELATIONAL } from 'lib/constants';
+import {
+  rawQueryClickhouse,
+  getBetweenDatesClickhouse,
+  getDateQuery,
+  getDateQueryClickhouse,
+  getDateStringQuery,
+  getDateStringQueryClickhouse,
+  parseFilters,
+  rawQuery,
+  runAnalyticsQuery,
+} from 'lib/db';
 
-export function getPageviewStats(
+export async function getPageviewStats(...args) {
+  return runAnalyticsQuery({
+    [`${RELATIONAL}`]: () => relationalQuery(...args),
+    [`${CLICKHOUSE}`]: () => clickhouseQuery(...args),
+  });
+}
+
+async function relationalQuery(
   website_id,
   start_at,
   end_at,
@@ -32,3 +50,37 @@ export function getPageviewStats(
     params,
   );
 }
+
+async function clickhouseQuery(
+  website_id,
+  start_at,
+  end_at,
+  timezone = 'UTC',
+  unit = 'day',
+  count = '*',
+  filters = {},
+) {
+  const params = [website_id];
+  const { pageviewQuery, sessionQuery, joinSession } = parseFilters('pageview', filters, params);
+
+  return rawQueryClickhouse(
+    `
+    select
+      ${getDateStringQueryClickhouse('g.t', unit)} as t, 
+      g.y as y
+    from
+      (select 
+        ${getDateQueryClickhouse('created_at', unit, timezone)} t,
+        count(${count}) y
+      from pageview
+        ${joinSession}
+      where pageview.website_id= $1
+      and ${getBetweenDatesClickhouse('pageview.created_at', start_at, end_at)}
+      ${pageviewQuery}
+      ${sessionQuery}
+      group by t) g
+    order by t
+    `,
+    params,
+  );
+}

queries/analytics/pageview/getPageviews.js

@@ -1,6 +1,20 @@
-import { prisma, runQuery } from 'lib/db';
+import { CLICKHOUSE, RELATIONAL } from 'lib/constants';
+import {
+  rawQueryClickhouse,
+  getDateFormatClickhouse,
+  prisma,
+  runAnalyticsQuery,
+  runQuery,
+} from 'lib/db';
 
-export async function getPageviews(websites, start_at) {
+export async function getPageviews(...args) {
+  return runAnalyticsQuery({
+    [`${RELATIONAL}`]: () => relationalQuery(...args),
+    [`${CLICKHOUSE}`]: () => clickhouseQuery(...args),
+  });
+}
+
+async function relationalQuery(websites, start_at) {
   return runQuery(
     prisma.pageview.findMany({
       where: {
@@ -16,3 +30,19 @@ export async function getPageviews(websites, start_at) {
     }),
   );
 }
+
+async function clickhouseQuery(websites, start_at) {
+  return rawQueryClickhouse(
+    `
+      select
+        view_id,
+        website_id,
+        session_id,
+        created_at,
+        url
+      from pageview
+      where website_id in (${websites.join[',']}
+      and created_at >= ${getDateFormatClickhouse(start_at)})
+    `,
+  );
+}

queries/analytics/session/getSessionByUuid.js

@@ -1,6 +1,14 @@
-import { prisma, runQuery } from 'lib/db';
+import { CLICKHOUSE, RELATIONAL } from 'lib/constants';
+import { rawQueryClickhouse, prisma, runAnalyticsQuery, runQuery } from 'lib/db';
 
-export async function getSessionByUuid(session_uuid) {
+export async function getSessionByUuid(...args) {
+  return runAnalyticsQuery({
+    [`${RELATIONAL}`]: () => relationalQuery(...args),
+    [`${CLICKHOUSE}`]: () => clickhouseQuery(...args),
+  });
+}
+
+async function relationalQuery(session_uuid) {
   return runQuery(
     prisma.session.findUnique({
       where: {
@@ -9,3 +17,27 @@ export async function getSessionByUuid(session_uuid) {
     }),
   );
 }
+
+async function clickhouseQuery(session_uuid) {
+  const params = [session_uuid];
+
+  return rawQueryClickhouse(
+    `
+    select 
+      session_id, 
+      session_uuid, 
+      website_id, 
+      created_at, 
+      hostname, 
+      browser, 
+      os, 
+      device, 
+      screen,
+      "language", 
+      country 
+    from session
+    where session_id = $1
+    `,
+    params,
+  );
+}

queries/analytics/session/getSessionMetrics.js

@@ -1,6 +1,20 @@
-import { parseFilters, rawQuery } from 'lib/db';
+import { CLICKHOUSE, RELATIONAL } from 'lib/constants';
+import {
+  getBetweenDatesClickhouse,
+  parseFilters,
+  rawQuery,
+  rawQueryClickhouse,
+  runAnalyticsQuery,
+} from 'lib/db';
 
-export function getSessionMetrics(website_id, start_at, end_at, field, filters = {}) {
+export async function getSessionMetrics(...args) {
+  return runAnalyticsQuery({
+    [`${RELATIONAL}`]: () => relationalQuery(...args),
+    [`${CLICKHOUSE}`]: () => clickhouseQuery(...args),
+  });
+}
+
+async function relationalQuery(website_id, start_at, end_at, field, filters = {}) {
   const params = [website_id, start_at, end_at];
   const { pageviewQuery, sessionQuery, joinSession } = parseFilters('pageview', filters, params);
 
@@ -23,3 +37,27 @@ export function getSessionMetrics(website_id, start_at, end_at, field, filters =
     params,
   );
 }
+
+async function clickhouseQuery(website_id, start_at, end_at, field, filters = {}) {
+  const params = [website_id];
+  const { pageviewQuery, sessionQuery, joinSession } = parseFilters('pageview', filters, params);
+
+  return rawQueryClickhouse(
+    `
+    select ${field} x, count(*) y
+    from session as x
+    where x.session_id in (
+      select pageview.session_id
+      from pageview
+        ${joinSession}
+      where pageview.website_id=$1
+      and ${getBetweenDatesClickhouse('pageview.created_at', start_at, end_at)}
+      ${pageviewQuery}
+      ${sessionQuery}
+    )
+    group by x
+    order by y desc
+    `,
+    params,
+  );
+}

queries/analytics/session/getSessions.js

@@ -1,6 +1,20 @@
-import { prisma, runQuery } from 'lib/db';
+import { CLICKHOUSE, RELATIONAL } from 'lib/constants';
+import {
+  getDateFormatClickhouse,
+  prisma,
+  rawQueryClickhouse,
+  runAnalyticsQuery,
+  runQuery,
+} from 'lib/db';
 
-export async function getSessions(websites, start_at) {
+export async function getSessions(...args) {
+  return runAnalyticsQuery({
+    [`${RELATIONAL}`]: () => relationalQuery(...args),
+    [`${CLICKHOUSE}`]: () => clickhouseQuery(...args),
+  });
+}
+
+async function relationalQuery(websites, start_at) {
   return runQuery(
     prisma.session.findMany({
       where: {
@@ -16,3 +30,25 @@ export async function getSessions(websites, start_at) {
     }),
   );
 }
+
+async function clickhouseQuery(websites, start_at) {
+  return rawQueryClickhouse(
+    `
+    select
+      session_id,
+      session_uuid,
+      website_id,
+      created_at,
+      hostname,
+      browser,
+      os,
+      device,
+      screen,
+      "language",
+      country
+    from session
+    where website_id in (${websites.join[',']}
+      and created_at >= ${getDateFormatClickhouse(start_at)})
+    `,
+  );
+}

queries/analytics/stats/getActiveVisitors.js

@@ -1,7 +1,15 @@
-import { rawQuery } from 'lib/db';
 import { subMinutes } from 'date-fns';
+import { CLICKHOUSE, RELATIONAL } from 'lib/constants';
+import { getDateFormatClickhouse, rawQuery, rawQueryClickhouse, runAnalyticsQuery } from 'lib/db';
 
-export function getActiveVisitors(website_id) {
+export async function getActiveVisitors(...args) {
+  return runAnalyticsQuery({
+    [`${RELATIONAL}`]: () => relationalQuery(...args),
+    [`${CLICKHOUSE}`]: () => clickhouseQuery(...args),
+  });
+}
+
+async function relationalQuery(website_id) {
   const date = subMinutes(new Date(), 5);
   const params = [website_id, date];
 
@@ -9,9 +17,23 @@ export function getActiveVisitors(website_id) {
     `
     select count(distinct session_id) x
     from pageview
-    where website_id=$1
+    where website_id = $1
     and created_at >= $2
     `,
     params,
   );
 }
+
+async function clickhouseQuery(website_id) {
+  const params = [website_id];
+
+  return rawQueryClickhouse(
+    `
+    select count(distinct session_id) x
+    from pageview
+    where website_id = $1
+    and created_at >= ${getDateFormatClickhouse(subMinutes(new Date(), 5))}
+    `,
+    params,
+  );
+}

queries/analytics/stats/getWebsiteStats.js

@@ -1,6 +1,23 @@
-import { parseFilters, rawQuery, getDateQuery, getTimestampInterval } from 'lib/db';
+import { CLICKHOUSE, RELATIONAL } from 'lib/constants';
+import {
+  getDateQuery,
+  getBetweenDatesClickhouse,
+  getDateQueryClickhouse,
+  getTimestampInterval,
+  parseFilters,
+  rawQuery,
+  rawQueryClickhouse,
+  runAnalyticsQuery,
+} from 'lib/db';
 
-export function getWebsiteStats(website_id, start_at, end_at, filters = {}) {
+export async function getWebsiteStats(...args) {
+  return runAnalyticsQuery({
+    [`${RELATIONAL}`]: () => relationalQuery(...args),
+    [`${CLICKHOUSE}`]: () => clickhouseQuery(...args),
+  });
+}
+
+async function relationalQuery(website_id, start_at, end_at, filters = {}) {
   const params = [website_id, start_at, end_at];
   const { pageviewQuery, sessionQuery, joinSession } = parseFilters('pageview', filters, params);
 
@@ -9,12 +26,18 @@ export function getWebsiteStats(website_id, start_at, end_at, filters = {}) {
       select sum(t.c) as "pageviews",
         count(distinct t.session_id) as "uniques",
         sum(case when t.c = 1 then 1 else 0 end) as "bounces",
+        sum(case when m2 < m1 + interval '1 hour' then ${getTimestampInterval(
+          'm2',
+          'm1',
+        )} else 0 end) as "totaltime"
         sum(t.time) as "totaltime"
       from (
-         select pageview.session_id,
+        select 
+          pageview.session_id,
           ${getDateQuery('pageview.created_at', 'hour')},
           count(*) c,
-           ${getTimestampInterval('pageview.created_at')} as "time"
+          min(created_at) m1,
+          max(created_at) m2
         from pageview
             ${joinSession}
         where pageview.website_id=$1
@@ -27,3 +50,33 @@ export function getWebsiteStats(website_id, start_at, end_at, filters = {}) {
     params,
   );
 }
+
+async function clickhouseQuery(website_id, start_at, end_at, filters = {}) {
+  const params = [website_id];
+  const { pageviewQuery, sessionQuery, joinSession } = parseFilters('pageview', filters, params);
+
+  return rawQueryClickhouse(
+    `
+      select 
+        sum(t.c) as "pageviews",
+        count(distinct t.session_id) as "uniques",
+        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_id,
+          ${getDateQueryClickhouse('pageview.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 ${getBetweenDatesClickhouse('pageview.created_at', start_at, end_at)}
+          ${pageviewQuery}
+          ${sessionQuery}
+        group by pageview.session_id, time_series
+      ) t;
+    `,
+    params,
+  );
+}