FLATTEN for nested data in Snowflake - Mini Project: Build & Apply

Flatten Nested Data Using FLATTEN in Snowflake

📖 Scenario: You work with a Snowflake database that stores customer orders. Each order has a list of items purchased, stored as a nested array. You need to extract each item as a separate row to analyze item sales.

🎯 Goal: Build a Snowflake SQL query that uses the FLATTEN function to convert nested arrays of order items into individual rows for easier analysis.

📋 What You'll Learn

Create a table called orders with columns order_id (integer) and items (variant) containing nested arrays

Insert sample data with nested arrays of items

Write a query using FLATTEN to extract each item from the nested items array

Select order_id and the flattened item values as separate rows

💡 Why This Matters

🌍 Real World

Many cloud data platforms store nested JSON data. Flattening helps analyze each nested element as a separate row for reporting and insights.

💼 Career

Data engineers and analysts often use FLATTEN in Snowflake to work with semi-structured data efficiently.

1

Create the orders table with nested items data

Create a table called orders with columns order_id as INTEGER and items as VARIANT. Insert two rows with order_id values 1 and 2. For items, insert the JSON arrays ["apple", "banana"] for order 1 and ["orange", "grape", "melon"] for order 2.

Snowflake

-- Create the orders table and insert two rows with nested items arrays
-- Your code here

Hint

Use CREATE OR REPLACE TABLE to create the table. Use PARSE_JSON to insert JSON arrays into the items column.

2

Add a query to flatten the items array

Write a SELECT query that uses the FLATTEN function on the items column from the orders table. Alias the FLATTEN function as f. Select order_id and f.value as item.

Snowflake

CREATE OR REPLACE TABLE orders (
  order_id INTEGER,
  items VARIANT
);

INSERT INTO orders (order_id, items) VALUES
  (1, PARSE_JSON('["apple", "banana"]')),
  (2, PARSE_JSON('["orange", "grape", "melon"]'));

-- Write a SELECT query using FLATTEN on items
-- Your code here

Hint

Use LATERAL FLATTEN(input => items) f in the FROM clause to expand the nested array.

3

Filter items to only include those starting with the letter 'g'

Modify the SELECT query to add a WHERE clause that filters the flattened item values to only include those starting with the letter 'g'. Use the LIKE operator with 'g%'.

Snowflake

CREATE OR REPLACE TABLE orders (
  order_id INTEGER,
  items VARIANT
);

INSERT INTO orders (order_id, items) VALUES
  (1, PARSE_JSON('["apple", "banana"]')),
  (2, PARSE_JSON('["orange", "grape", "melon"]'));

SELECT order_id, f.value AS item
FROM orders,
LATERAL FLATTEN(input => items) f
-- Add a WHERE clause to filter items starting with 'g'
-- Your code here

Hint

Use WHERE f.value LIKE 'g%' to filter items starting with 'g'.

4

Complete the query by ordering results by order_id and item

Add an ORDER BY clause to the SELECT query to sort the results by order_id ascending and then by item ascending.

Snowflake

CREATE OR REPLACE TABLE orders (
  order_id INTEGER,
  items VARIANT
);

INSERT INTO orders (order_id, items) VALUES
  (1, PARSE_JSON('["apple", "banana"]')),
  (2, PARSE_JSON('["orange", "grape", "melon"]'));

SELECT order_id, f.value AS item
FROM orders,
LATERAL FLATTEN(input => items) f
WHERE f.value LIKE 'g%'
-- Add ORDER BY clause to sort by order_id and item
-- Your code here

Hint

Use ORDER BY order_id ASC, item ASC to sort the results.

Practice

1. What does the FLATTEN function do in Snowflake when working with nested data?

easy

A. It encrypts nested data for security.

B. It compresses data to save storage space.

C. It converts nested arrays or objects into simple rows.

D. It creates a backup of nested data.

Solution

1. Step 1: Understand the purpose of FLATTEN

   FLATTEN is designed to take nested arrays or objects and turn them into individual rows so they are easier to query.

2. Step 2: Compare options to FLATTEN's function

   Options A, B, and D describe encryption, compression, and backup, which are unrelated to FLATTEN's role.

3. Final Answer:

   It converts nested arrays or objects into simple rows. -> Option C

4. Quick Check:

   FLATTEN = convert nested data to rows [OK]

Hint: FLATTEN breaks nested data into rows for easy reading [OK]

Common Mistakes:

• Thinking FLATTEN compresses or encrypts data
• Confusing FLATTEN with backup or storage functions
• Assuming FLATTEN changes data format instead of structure

2. Which of the following is the correct syntax to use FLATTEN on a JSON column named data in Snowflake?

easy

A. SELECT FLATTEN(data) FROM table;

B. SELECT * FROM FLATTEN(input => data);

C. SELECT FLATTEN(input = data) FROM table;

D. SELECT * FROM table, LATERAL FLATTEN(input => data);

Solution

1. Step 1: Recall FLATTEN usage in FROM clause

   FLATTEN is used as a table function in the FROM clause with LATERAL to expand nested data.

2. Step 2: Analyze each option's syntax

   SELECT * FROM table, LATERAL FLATTEN(input => data); correctly uses FROM table, LATERAL FLATTEN(input => data). Options A and C misuse FLATTEN as a scalar function. SELECT * FROM FLATTEN(input => data); misses the table reference.

3. Final Answer:

   SELECT * FROM table, LATERAL FLATTEN(input => data); -> Option D

4. Quick Check:

   FLATTEN used with LATERAL in FROM clause [OK]

Hint: Use FLATTEN with LATERAL in FROM clause for nested data [OK]

Common Mistakes:

• Using FLATTEN as a scalar function in SELECT
• Omitting LATERAL keyword
• Not specifying input parameter correctly

3. Given the JSON column data with value '{"items": ["apple", "banana", "cherry"]}', what will the query below return?

SELECT f.value FROM table, LATERAL FLATTEN(input => data:items) f;

medium

A. Rows with values: apple, banana, cherry

B. A single row with the entire array as a string

C. An error because data:items is invalid syntax

D. Rows with keys and values of the JSON object

Solution

1. Step 1: Understand FLATTEN on JSON array

   FLATTEN(input => data:items) expands the array under 'items' into multiple rows, each with one element.

2. Step 2: Analyze the query output

   The query selects f.value, which will be each element: 'apple', 'banana', 'cherry' as separate rows.

3. Final Answer:

   Rows with values: apple, banana, cherry -> Option A

4. Quick Check:

   FLATTEN on array returns each element as a row [OK]

Hint: FLATTEN on JSON array returns each element as a separate row [OK]

Common Mistakes:

• Expecting a single row with the whole array
• Confusing keys with values in FLATTEN output
• Misreading JSON path syntax

4. You wrote this query to flatten nested JSON data:

SELECT f.value FROM table, FLATTEN(input => data:items) f;

But it returns an error. What is the likely cause?

medium

A. Missing LATERAL keyword before FLATTEN

B. Incorrect JSON path syntax in input

C. FLATTEN cannot be used on JSON arrays

D. SELECT statement missing WHERE clause

Solution

1. Step 1: Identify FLATTEN usage requirements

   FLATTEN is a table function that requires LATERAL when used with another table to expand nested data.

2. Step 2: Check query syntax

   The query misses the LATERAL keyword before FLATTEN, causing a syntax error.

3. Final Answer:

   Missing LATERAL keyword before FLATTEN -> Option A

4. Quick Check:

   FLATTEN needs LATERAL in FROM clause [OK]

Hint: Always add LATERAL before FLATTEN in FROM clause [OK]

Common Mistakes:

• Forgetting LATERAL keyword
• Assuming FLATTEN works without LATERAL
• Blaming JSON path syntax instead of syntax structure

5. You have a table with a column orders storing nested JSON arrays of items per order. You want to list each item with its order ID. Which query correctly uses FLATTEN to achieve this?

hard

A. SELECT order_id, FLATTEN(orders) FROM orders_table;

B. SELECT order_id, f.value FROM orders_table, LATERAL FLATTEN(input => orders) f;

C. SELECT order_id, f.value FROM orders_table JOIN FLATTEN(input => orders) f ON TRUE;

D. SELECT order_id, f.value FROM orders_table, FLATTEN(orders) f;

Solution

1. Step 1: Understand how to join FLATTEN with table

   FLATTEN must be used with LATERAL in the FROM clause to expand nested arrays per row.

2. Step 2: Evaluate each option's correctness

   SELECT order_id, f.value FROM orders_table, LATERAL FLATTEN(input => orders) f; correctly uses FROM orders_table, LATERAL FLATTEN(input => orders) f, selecting order_id and each item value. Options A and D misuse FLATTEN syntax. SELECT order_id, f.value FROM orders_table JOIN FLATTEN(input => orders) f ON TRUE; uses JOIN incorrectly without LATERAL.

3. Final Answer:

   SELECT order_id, f.value FROM orders_table, LATERAL FLATTEN(input => orders) f; -> Option B

4. Quick Check:

   Use LATERAL FLATTEN with table to list nested items [OK]

Hint: Use LATERAL FLATTEN(input => column) to expand nested arrays per row [OK]

Common Mistakes:

• Using FLATTEN without LATERAL
• Trying to JOIN FLATTEN without LATERAL
• Incorrect FLATTEN syntax in SELECT clause