Common table expressions

Introduction to BigQuery

Matthew Forrest

Field CTO

What is a CTE?

-- The WITH keywords start the CTE
WITH our_cte AS (
  SELECT column1, column2
  FROM table
)


/* The CTE is contained in parentheses and can be used as a 
table in the main query below */
SELECT *
FROM our_cte

Subqueries vs. CTEs

Subquery

SELECT *
FROM (
  SELECT order_id, status
  FROM ecommerce.ecomm_order_status
) AS subquery

CTE

WITH cte_name AS (
  SELECT order_id, status
  FROM ecommerce.ecomm_order_status
)
SELECT *
FROM cte_name

Writing CTEs

-- All CTEs begin with the WITH keyword
WITH cte_name AS


-- The CTE is contained in parentheses
(SELECT col_1, col_2
 FROM tablename WHERE col_1 > 10
)


-- The new CTE 'cte_name' is only accessible within the scope of this query

SELECT * FROM cte_name LIMIT 10

Using multiple CTEs

WITH orders AS (
  SELECT order_id, status
  FROM ecommerce.ecomm_order_status
),

-- Additional CTEs do not reuse the WTIH keyword and are separated by commas
order_details AS (
SELECT orders.*, details.order_items
FROM ecommerce.ecomm_orders details 
```

Using CTEs to filter data

WITH filtered_data AS (
  SELECT *
  FROM ecommerce.ecomm_orders
  -- Here, we filter the data in our CTE
  WHERE order_status = 'delivered'
)
SELECT order_id, order_status
FROM filtered_data

| order_id | order_status |
|----------|--------------|
| 1        | delivered    |
| 2        | delivered    |

Using CTEs to optimize queries

WITH precomputed_data AS (
  SELECT order_id, SUM(payment_value) AS total
  FROM ecommerce.ecomm_payments
  GROUP BY order_id
)
SELECT *
FROM precomputed_data
WHERE total > 1000

| order_id | total   |
|----------|---------|
| 1        | 50000   |
| 2        | 8000    |

Using CTEs to join data

WITH cte1 AS (
  SELECT order_id, status
  FROM ecommerce.ecomm_order_status
),
cte2 AS (
  SELECT order_id, order_items
  FROM ecommerce.ecomm_orders
)
SELECT *
FROM cte1
JOIN cte2 USING (order_id)

Let's practice!

Introduction to BigQuery