Basic queries

The general form of that statement is:

[WITH with_query [, ...]]
SELECT select_expr [, ...]
FROM from_item
[JOIN join_item [, ...]]
[WHERE condition]
[GROUP BY grouping_element [, ...]]
[HAVING condition]

WITH clause

The with clauses allow you to give names to subquery which you can then reference. The syntax for a with clause is:

WITH query_name AS (subquery) [,...]

For example, using the nexmark source, you can create datasets for bids and price and then join.

WITH bids AS
    (SELECT bid.auction AS auction, bid.price AS price
        FROM nexmark where bid is not null),
auctions AS
    (SELECT auction.id AS id
        FROM nexmark where auction is not null)
SELECT * FROM bids bids
        JOIN auctions auctions
            ON bids.auction = auctions.id;

Worked example

Given a bids table:

auction_id	price
1	10.00
2	25.00
1	12.00
3	30.00

And an auctions table:

id	item
1	Laptop
2	Phone
3	Tablet

The following query uses CTEs to join bids with their auction items:

WITH bid_data AS (
    SELECT auction_id, price FROM bids
),
auction_data AS (
    SELECT id, item FROM auctions
)
SELECT bid_data.auction_id, bid_data.price, auction_data.item
FROM bid_data
JOIN auction_data ON bid_data.auction_id = auction_data.id;

Result:

auction_id	price	item
1	10.00	Laptop
2	25.00	Phone
1	12.00	Laptop
3	30.00	Tablet

Each bid row is matched to its corresponding auction by joining auction_id to id, producing one output row per bid with the item name attached.

SELECT clause

The select cause is a comma-separated list of expressions, with an optional alias. Column names must be unique.

SELECT select_expr [, ...]

FROM clause

The FROM clause specifies the primary source of data. It will be either a table name or subquery. The table name can be either a saved source, a table created in the WITH clause or a table created via CREATE TABLE and inserted into. Tables can be given aliases, but will default to their name as the alias for things like joins.

FROM from_item

JOIN clause

The JOIN clause allows you to join multiple tables together. See the join documentation for more details.

WHERE clause

The WHERE clause allows you to filter the data with a boolean condition. This predicate is applied to the incoming rows, so cannot include conditions on the resulting columns.

WHERE condition

Worked example

Given an events table:

event_id	event_type	user_id
1	pageview	alice
2	click	bob
3	pageview	charlie
4	purchase	alice
5	pageview	bob

The following query filters to only pageview events:

SELECT event_id, event_type, user_id
FROM events
WHERE event_type = 'pageview';

Result:

event_id	event_type	user_id
1	pageview	alice
3	pageview	charlie
5	pageview	bob

Only rows where event_type = 'pageview' pass the filter. The click and purchase events are excluded.

GROUP BY clause

The GROUP BY clause is used to compute aggregates over some set of fields. All GROUP BY queries will implicitly include a time window, and if the input doesn’t already have a time window, it should be specified as one of the grouping fields. For example,

SELECT
    count(*) AS bids,
    count(distinct auction_id) AS distinct_auctions,
    tumble(interval '1 minute') AS window
FROM BIDS GROUP BY 3

Worked example

Given a bids table with events arriving over time:

auction_id	price
1	10.00
2	25.00
1	12.00
3	30.00

The following query counts the number of bids and distinct auctions per 1-minute tumbling window:

SELECT
    count(*) AS bid_count,
    count(distinct auction_id) AS distinct_auctions,
    tumble(interval '1 minute') AS window
FROM bids
GROUP BY 3;

Result (assuming all events fall within the same 1-minute window):

bid_count	distinct_auctions	window
4	3	(2024-01-01T00:00:00, 2024-01-01T00:01:00)

The tumble(interval '1 minute') creates fixed, non-overlapping 1-minute windows. All rows within each window are aggregated together. See the windows documentation for more on tumbling and other window types.

HAVING clause

The HAVING clause allows filtering on the result of aggregations (as opposed to WHERE, which filters on the inputs to the aggregation). For example:

SELECT user_id, count(*) as count
FROM events
WHERE event_type = 'pageview'
GROUP BY user_id
HAVING count > 5;

This query counts pageview events by user, then returns those users with more than 5 pageviews.

Worked example

Given the same events table:

event_id	event_type	user_id
1	pageview	alice
2	click	bob
3	pageview	charlie
4	purchase	alice
5	pageview	bob

The following query counts all events per user, then keeps only users with more than 1 event:

SELECT user_id, count(*) AS event_count
FROM events
GROUP BY user_id
HAVING event_count > 1;

Result:

user_id	event_count
alice	2
bob	2

alice has 2 events (pageview + purchase) - included
bob has 2 events (click + pageview) - included
charlie has 1 event (pageview) - excluded by the HAVING filter

UNNEST operator

The UNNEST operator allows you to unnest arrays into multiple rows. This can be used as a normal scalar function with the following restrictions:

It may only appear in the SELECT clause
Only one array may be unnested per select statement For example,

SELECT
    UNNEST(make_array(1, 2, 3)) as x
FROM BIDS;

which will produce the following output:

+---+
| x |
+---+
| 1 |
| 2 |
| 3 |
+---+

Worked example

Given a bids table:

auction_id	price
1	10.00
2	25.00

The following query unnests a fixed array for each input row:

SELECT
    auction_id,
    UNNEST(make_array('low', 'medium', 'high')) AS tier
FROM bids;

Result:

auction_id	tier
1	low
1	medium
1	high
2	low
2	medium
2	high

Each input row is expanded into multiple output rows - one for each element in the unnested array. The other columns are repeated for each expanded row.

Ingest Engine

Get started

Connectors

Streaming SQL

Operate

Reference

Concepts

WITH clause

Worked example

SELECT clause

FROM clause

JOIN clause

WHERE clause

Worked example

GROUP BY clause

Worked example

HAVING clause

Worked example

UNNEST operator

Worked example

​WITH clause

​Worked example

​SELECT clause

​FROM clause

​JOIN clause

​WHERE clause

​Worked example

​GROUP BY clause

​Worked example

​HAVING clause

​Worked example

​UNNEST operator

​Worked example

WITH clause

Worked example

SELECT clause

FROM clause

JOIN clause

WHERE clause

Worked example

GROUP BY clause

Worked example

HAVING clause

Worked example

UNNEST operator

Worked example