Custom singular tests in dbt - Cheat Sheet & Quick Revision

Recall & Review

beginner

What is a custom singular test in dbt?

A custom singular test in dbt is a user-defined check that runs a single SQL query returning zero or more rows. If the query returns any rows, the test fails. It helps validate specific data quality rules beyond built-in tests.

beginner

How do you define a custom singular test in dbt?

You create a SQL file in the 'tests' directory with a query that returns rows when the test fails, typically named `<model_name>__<test_name>.sql`. dbt automatically runs it during `dbt test`.

beginner

Example: What does this custom singular test do?

select * from {{ model }} where order_amount < 0

This test checks if any orders have a negative amount. If any rows are returned, it means there are invalid negative order amounts, so the test fails.

intermediate

Why use custom singular tests instead of built-in tests in dbt?

Custom singular tests let you check complex or unique business rules that built-in tests can't cover. They provide flexibility to enforce specific data quality needs.

beginner

Where do you place custom singular test SQL files in a dbt project?

Place them inside the 'tests' folder at the root or inside a subfolder like 'tests/custom'. This keeps tests organized and discoverable by dbt.

What does a custom singular test in dbt return when it fails?

A One or more rows

B No rows

C A boolean value

D An error message

A custom singular test returns rows when the test fails. If no rows are returned, the test passes.

Where should you place your custom singular test SQL files in a dbt project?

A In the 'models' folder

B In the 'tests' folder

C In the 'macros' folder

D In the root folder

Custom singular test SQL files belong in the 'tests' folder so dbt can find and run them.

Which of these is a reason to create a custom singular test?

A To check for null values in a column

B To enforce a unique constraint

C To create a new model

D To validate a complex business rule not covered by built-in tests

Custom singular tests are used to validate complex or unique business rules beyond built-in tests.

How do you run custom singular tests in dbt?

A Using 'dbt test'

B Using 'dbt run'

C Using 'dbt compile'

D Using 'dbt seed'

'dbt test' runs all tests including custom singular tests.

What happens if a custom singular test query returns zero rows?

A The test is skipped

B The test fails

C The test passes

D An error is raised

If the query returns zero rows, it means no issues were found and the test passes.

Explain what a custom singular test is in dbt and how it works.

Think about how dbt knows a test passed or failed based on query results.

Describe the steps to create and use a custom singular test in a dbt project.

Consider both the file location and how dbt executes the test.

Practice

1. What is the main purpose of a custom singular test in dbt?

easy

A. To automatically generate documentation for your models

B. To write your own SQL query that checks data quality and returns rows only if there are issues

C. To schedule dbt runs at specific times

D. To create new tables from existing data

Solution

1. Step 1: Understand the role of custom singular tests

   Custom singular tests are SQL queries that check data quality by returning rows only when problems exist.

2. Step 2: Compare options with this definition

   Only To write your own SQL query that checks data quality and returns rows only if there are issues describes writing a SQL query that returns rows if there are data issues, matching the purpose of custom singular tests.

3. Final Answer:

   To write your own SQL query that checks data quality and returns rows only if there are issues -> Option B

4. Quick Check:

   Custom singular test = SQL check returning problem rows [OK]

Hint: Custom singular tests return rows only when data has problems [OK]

Common Mistakes:

• Confusing tests with documentation generation
• Thinking tests create tables
• Assuming tests schedule runs

2. Which of the following is the correct way to define a custom singular test in your schema.yml file?

easy

A. tests:
- my_custom_test.sql

B. tests:
- my_custom_test:
sql: my_custom_test.sql

C. tests:
- name: my_custom_test
test: my_custom_test

D. tests:
- my_custom_test

Solution

1. Step 1: Recall the schema.yml syntax for custom singular tests

   Custom singular tests are referenced by their filename (without .sql) in the tests list of schema.yml.

2. Step 2: Match options to this syntax

   tests: - my_custom_test correctly references the test file tests/my_custom_test.sql. Other options use incorrect structure, extra keys, or include .sql.

3. Final Answer:

   tests: - my_custom_test -> Option D

4. Quick Check:

   schema.yml test syntax = - test_filename_without_sql [OK]

Hint: Reference tests by name (no .sql) in tests: list [OK]

Common Mistakes:

• Using 'name' or 'test' keys
• Including .sql extension
• Using map/dict structure

3. Given the following custom singular test SQL in tests/check_positive_values.sql:

SELECT * FROM {{ ref('orders') }} WHERE amount <= 0

What will be the output if all amounts in the orders table are positive?

medium

A. An empty result with zero rows

B. A table with all rows where amount is less than or equal to zero

C. An error because of invalid SQL syntax

D. A count of rows with amount less than or equal to zero

Solution

1. Step 1: Understand the test SQL logic

   The test selects rows where amount is less than or equal to zero.

2. Step 2: Analyze the data condition

   If all amounts are positive, no rows satisfy the condition, so the query returns zero rows.

3. Final Answer:

   An empty result with zero rows -> Option A

4. Quick Check:

   All positive amounts means zero rows returned [OK]

Hint: No matching rows means test passes with empty output [OK]

Common Mistakes:

• Expecting a count instead of rows
• Thinking it returns all rows
• Assuming SQL syntax error

4. You wrote a custom singular test SQL file but when running dbt test, it fails with a syntax error. Which of the following is the most likely cause?

medium

A. The model referenced in {{ ref() }} does not exist

B. The test SQL returns zero rows

C. The SQL file is missing the required SELECT statement

D. The test is not listed in schema.yml

Solution

1. Step 1: Identify causes of SQL syntax errors

   Syntax errors happen when SQL is malformed, such as missing SELECT statements.

2. Step 2: Evaluate options for syntax error causes

   The SQL file is missing the required SELECT statement directly relates to SQL syntax. Other options cause runtime or configuration errors, not syntax errors.

3. Final Answer:

   The SQL file is missing the required SELECT statement -> Option C

4. Quick Check:

   Syntax error = malformed SQL like missing SELECT [OK]

Hint: Syntax errors usually mean SQL is incomplete or malformed [OK]

Common Mistakes:

• Confusing missing test listing with syntax error
• Assuming zero rows cause syntax errors
• Ignoring missing model references

5. You want to create a custom singular test that checks if any user has a NULL email in the users table. Which SQL query should you write in your test file?

hard

A. SELECT * FROM {{ ref('users') }} WHERE email IS NULL

B. SELECT COUNT(*) FROM {{ ref('users') }} WHERE email IS NULL

C. SELECT email FROM {{ ref('users') }} WHERE email IS NOT NULL

D. SELECT * FROM {{ ref('users') }} WHERE email = ''

Solution

1. Step 1: Understand the test goal

   The test should return rows where email is NULL to detect missing emails.

2. Step 2: Choose the SQL that returns rows with NULL emails

   SELECT * FROM {{ ref('users') }} WHERE email IS NULL returns rows only when there are NULL emails (0 rows = pass). COUNT(*) always returns one row, failing even with zero NULLs. IS NOT NULL selects good rows (opposite). = '' checks empty strings, not NULLs.

3. Final Answer:

   SELECT * FROM {{ ref('users') }} WHERE email IS NULL -> Option A

4. Quick Check:

   Return rows with NULL email = SELECT * FROM {{ ref('users') }} WHERE email IS NULL [OK]

Hint: Use SELECT * WHERE column IS NULL to find missing values [OK]

Common Mistakes:

• Using COUNT(*) instead of returning rows
• Checking for empty string instead of NULL
• Selecting non-NULL emails