DBT Unit Testing with Pytest Plugin

The Pytest plugin for DBT simplifies the task of writing unit tests for your DBT models, offering a straightforward and efficient testing procedure.

How to Install

Install this plugin by utilizing pip. Simply run the following command:

pip install pytest-dbt

Steps for Use

To use this plugin, follow these instructions:

1. Begin by creating a DBT model in your usual manner.
2. Next, execute dbt compile.
3. Now, either create a new pytest.ini file or modify an existing one. In this file, set the dbt_manifest_file_path to point towards the appropriate manifest file. See the example below:

[pytest]
dbt_manifest_file_path = target/manifest.json

4. The next step is to create your tests using pytest. As an example, let's assume you have a model called order_size and you want to verify it has three columns. You can create a test file test_order_size.py and add the following test:

customer_no_order = [
    {'order_id': 1 , "order_size": 'small'},
    {'order_id': 2 , "order_size": 'large'},
    {'order_id': 3 , "order_size": 'medium'},
    {'order_id': 4 , "order_size": 'large'},
    {'order_id': 5 , "order_size": 'large'},
]

tables = {
    'orders': orders,
}

def test_customers_no_order():
    result = run_model('order_size', tables)
    assert assert_frame_equal(result, pd.DataFrame(customer_no_order)) == None

Incorporating User-Defined Functions (UDFs)

If your databases have defined functions, you can incorporate them into the execution. For instance, if there's a model customer_details which employs a UDF full_name, you can design a test file test_customer_details.py and append the following test:

def full_name(fname, lastname):
    return fname + " " + lastname

udfs = {"full_name": lambda name, last_name: f"{name} {last_name}"}

def test_customer_details():
    result = run_model("customer_details", tables=tables, udfs=udfs)
    assert assert_frame_equal(result, pd.DataFrame(customer_details)) == None

5. Finally, execute the tests with the command pytest:

➜  jaffle_shop git$ ✗ pytest
============================================================= test session starts =============================================================
platform darwin -- Python 3.8.6, pytest-7.3.1, pluggy-1.0.0
rootdir: /Users/anupkalburgi/projects/pytest-dbt/jaffle_shop
configfile: pytest.ini
plugins: dbt-0.1.1
collected 5 items

tests/test_customers.py .                                                                                                               [ 20%]
tests/test_customers_no_order.py .                                                                                                      [ 40%]
tests/test_order_size.py .                                                                                                              [ 60%]
tests/test_orders.py .                                                                                                                  [ 80%]
tests/test_udf_customer_details.py .                                                                                                    [100%]

============================================================== 5 passed in 0.15s ==============================================================

Why Not Utilize DBT Test?

While DBT Test excels in conducting integration tests, it falls short when applied to unit testing.

The Advantage of Pytest

Pytest is a robust testing framework with a wide variety of plugins, backed by strong community support.

Anticipated Updates

The following enhancements are being planned for the plugin:

• Inclusion of support for User-Defined Functions (UDFs).
• Incorporation of a cyclic redundancy check (CRC) for tracking model changes.
• Integration of a coverage report for SQL files.
• Addition of a command-line option for test configuration input, which will be particularly beneficial for Continuous Integration (CI) scenarios.
• Rectification of issues concerning missing files in test_dbt_manifest_file_path = dbt_manifest.json.
• Resolution of the AssertionError: File does not exist at path: dbt_manifest.json error.