Skip to main content

Unit tests

note

This functionality is only supported in dbt Core v1.8+ or accounts that have opted for a "Versionless" dbt Cloud experience.

Historically, dbt's test coverage was confined to “data” tests, assessing the quality of input data or resulting datasets' structure. However, these tests could only be executed after building a model.

With dbt Core v1.8 and dbt Cloud environments that have gone versionless by selecting the Versionless option, we have introduced an additional type of test to dbt - unit tests. In software programming, unit tests validate small portions of your functional code, and they work much the same way here. Unit tests allow you to validate your SQL modeling logic on a small set of static inputs before you materialize your full model in production. Unit tests enable test-driven development, benefiting developer efficiency and code reliability.

Before you begin​

  • We currently only support unit testing SQL models.
  • We currently only support adding unit tests to models in your current project.
  • We currently don't support unit testing models that use the materialized view materialization.
  • We currently don't support unit testing models that use recursive SQL.
  • You must specify all fields in a BigQuery STRUCT in a unit test. You cannot use only a subset of fields in a STRUCT.
  • If your model has multiple versions, by default the unit test will run on all versions of your model. Read unit testing versioned models for more information.
  • Unit tests must be defined in a YML file in your models/ directory.
  • Table names must be aliased in order to unit test join logic.
  • Redshift customers need to be aware of a limitation when building unit tests that requires a workaround.

Read the reference doc for more details about formatting your unit tests.

When to add a unit test to your model​

You should unit test a model:

  • When your SQL contains complex logic:
    • Regex
    • Date math
    • Window functions
    • case when statements when there are many whens
    • Truncation
  • When you're writing custom logic to process input data, similar to creating a function.
  • We don't recommend conducting unit testing for functions like min() since these functions are tested extensively by the warehouse. If an unexpected issue arises, it's more likely a result of issues in the underlying data rather than the function itself. Therefore, fixture data in the unit test won't provide valuable information.
  • Logic for which you had bugs reported before.
  • Edge cases not yet seen in your actual data that you want to handle.
  • Prior to refactoring the transformation logic (especially if the refactor is significant).
  • Models with high "criticality" (public, contracted models or models directly upstream of an exposure).

When to run unit tests​

dbt Labs strongly recommends only running unit tests in development or CI environments. Since the inputs of the unit tests are static, there's no need to use additional compute cycles running them in production. Use them in development for a test-driven approach and CI to ensure changes don't break them.

Use the resource type flag --exclude-resource-type or the DBT_EXCLUDE_RESOURCE_TYPES environment variable to exclude unit tests from your production builds and save compute.

Unit testing a model​

This example creates a new dim_customers model with a field is_valid_email_address that calculates whether or not the customer’s email is valid:

with customers as (

    select * from {{ ref('stg_customers') }}

),

accepted_email_domains as (

    select * from {{ ref('top_level_email_domains') }}

),
	
check_valid_emails as (

    select
        customers.customer_id,
        customers.first_name,
        customers.last_name,
        customers.email,
	      coalesce (regexp_like(
            customers.email, '^[A-Za-z0-9._%+-]+@[A-Za-z0-9.-]+\\.[A-Za-z]{2,}$'
        )
        = true
        and accepted_email_domains.tld is not null,
        false) as is_valid_email_address
    from customers
		left join accepted_email_domains
        on customers.email_top_level_domain = lower(accepted_email_domains.tld)

)

select * from check_valid_emails

The logic posed in this example can be challenging to validate. You can add a unit test to this model to ensure the is_valid_email_address logic captures all known edge cases: emails without ., emails without @, and emails from invalid domains.

unit_tests:
  - name: test_is_valid_email_address
    description: "Check my is_valid_email_address logic captures all known edge cases - emails without ., emails without @, and emails from invalid domains."
    model: dim_customers
    given:
      - input: ref('stg_customers')
        rows:
          - {email: cool@example.com,    email_top_level_domain: example.com}
          - {email: cool@unknown.com,    email_top_level_domain: unknown.com}
          - {email: badgmail.com,        email_top_level_domain: gmail.com}
          - {email: missingdot@gmailcom, email_top_level_domain: gmail.com}
      - input: ref('top_level_email_domains')
        rows:
          - {tld: example.com}
          - {tld: gmail.com}
    expect:
      rows:
        - {email: cool@example.com,    is_valid_email_address: true}
        - {email: cool@unknown.com,    is_valid_email_address: false}
        - {email: badgmail.com,        is_valid_email_address: false}
        - {email: missingdot@gmailcom, is_valid_email_address: false}

The previous example defines the mock data using the inline dict format, but you can also use csv or sql either inline or in a separate fixture file. Store your fixture files in a fixtures subdirectory in any of your test paths. For example, tests/fixtures/my_unit_test_fixture.sql.

When using the dict or csv format, you only have to define the mock data for the columns relevant to you. This enables you to write succinct and specific unit tests.

note

The direct parents of the model that you’re unit testing (in this example, stg_customers and top_level_email_domains) need to exist in the warehouse before you can execute the unit test.

Use the --empty flag to build an empty version of the models to save warehouse spend.


dbt run --select "stg_customers top_level_email_domains" --empty

Alternatively, use dbt build to, in lineage order:

  • Run the unit tests on your model.
  • Materialize your model in the warehouse.
  • Run the data tests on your model.

Now you’re ready to run this unit test. You have a couple of options for commands depending on how specific you want to be:

  • dbt test --select dim_customers runs all of the tests on dim_customers.
  • dbt test --select "dim_customers,test_type:unit" runs all of the unit tests on dim_customers.
  • dbt test --select test_is_valid_email_address runs the test named test_is_valid_email_address.

dbt test --select test_is_valid_email_address
16:03:49  Running with dbt=1.8.0-a1
16:03:49  Registered adapter: postgres=1.8.0-a1
16:03:50  Found 6 models, 5 seeds, 4 data tests, 0 sources, 0 exposures, 0 metrics, 410 macros, 0 groups, 0 semantic models, 1 unit test
16:03:50  
16:03:50  Concurrency: 5 threads (target='postgres')
16:03:50  
16:03:50  1 of 1 START unit_test dim_customers::test_is_valid_email_address ................... [RUN]
16:03:51  1 of 1 FAIL 1 dim_customers::test_is_valid_email_address ............................ [FAIL 1 in 0.26s]
16:03:51  
16:03:51  Finished running 1 unit_test in 0 hours 0 minutes and 0.67 seconds (0.67s).
16:03:51  
16:03:51  Completed with 1 error and 0 warnings:
16:03:51  
16:03:51  Failure in unit_test test_is_valid_email_address (models/marts/unit_tests.yml)
16:03:51    

actual differs from expected:

@@ ,email           ,is_valid_email_address
→  ,cool@example.com,True→False
   ,cool@unknown.com,False
...,...             ,...


16:03:51  
16:03:51    compiled Code at models/marts/unit_tests.yml
16:03:51  
16:03:51  Done. PASS=0 WARN=0 ERROR=1 SKIP=0 TOTAL=1

The clever regex statement wasn’t as clever as initially thought, as the model incorrectly flagged cool@example.com as an invalid email address.

Updating the regex logic to '^[A-Za-z0-9._%+-]+@[A-Za-z0-9.-]+\.[A-Za-z]{2,}$' (those pesky escape characters) and rerunning the unit test solves the problem:


dbt test --select test_is_valid_email_address
16:09:11  Running with dbt=1.8.0-a1
16:09:12  Registered adapter: postgres=1.8.0-a1
16:09:12  Found 6 models, 5 seeds, 4 data tests, 0 sources, 0 exposures, 0 metrics, 410 macros, 0 groups, 0 semantic models, 1 unit test
16:09:12  
16:09:13  Concurrency: 5 threads (target='postgres')
16:09:13  
16:09:13  1 of 1 START unit_test dim_customers::test_is_valid_email_address ................... [RUN]
16:09:13  1 of 1 PASS dim_customers::test_is_valid_email_address .............................. [PASS in 0.26s]
16:09:13  
16:09:13  Finished running 1 unit_test in 0 hours 0 minutes and 0.75 seconds (0.75s).
16:09:13  
16:09:13  Completed successfully
16:09:13  
16:09:13  Done. PASS=1 WARN=0 ERROR=0 SKIP=0 TOTAL=1

Your model is now ready for production! Adding this unit test helped catch an issue with the SQL logic before you materialized dim_customers in your warehouse and will better ensure the reliability of this model in the future.

Unit testing incremental models​

When configuring your unit test, you can override the output of macros, vars, or environment variables. This enables you to unit test your incremental models in "full refresh" and "incremental" modes.

When testing an incremental model, the expected output is the result of the materialization (what will be merged/inserted), not the resulting model itself (what the final table will look like after the merge/insert).

For example, say you have an incremental model in your project:

my_incremental_model.sql

{{
    config(
        materialized='incremental'
    )
}}

select * from {{ ref('events') }}
{% if is_incremental() %}
where event_time > (select max(event_time) from {{ this }})
{% endif %}

You can define unit tests on my_incremental_model to ensure your incremental logic is working as expected:


unit_tests:
  - name: my_incremental_model_full_refresh_mode
    model: my_incremental_model
    overrides:
      macros:
        # unit test this model in "full refresh" mode
        is_incremental: false 
    given:
      - input: ref('events')
        rows:
          - {event_id: 1, event_time: 2020-01-01}
    expect:
      rows:
        - {event_id: 1, event_time: 2020-01-01}

  - name: my_incremental_model_incremental_mode
    model: my_incremental_model
    overrides:
      macros:
        # unit test this model in "incremental" mode
        is_incremental: true 
    given:
      - input: ref('events')
        rows:
          - {event_id: 1, event_time: 2020-01-01}
          - {event_id: 2, event_time: 2020-01-02}
          - {event_id: 3, event_time: 2020-01-03}
      - input: this 
        # contents of current my_incremental_model
        rows:
          - {event_id: 1, event_time: 2020-01-01}
    expect:
      # what will be inserted/merged into my_incremental_model
      rows:
        - {event_id: 2, event_time: 2020-01-02}
        - {event_id: 3, event_time: 2020-01-03}

There is currently no way to unit test whether the dbt framework inserted/merged the records into your existing model correctly, but we're investigating support for this in the future.

Unit testing a model that depend on ephemeral model(s)​

If you want to unit test a model that depends on an ephemeral model, you must use format: sql for that input.

unit_tests:
  - name: my_unit_test
    model: dim_customers
    given:
      - input: ref('ephemeral_model')
        format: sql
        rows: |
          select 1 as id, 'emily' as name
    expect:
      rows:
        - {id: 1, first_name: emily}

Unit test exit codes​

Unit test successes and failures are represented by two exit codes:

  • Pass (0)
  • Fail (1)

Exit codes differ from data test success and failure outputs because they don't directly reflect failing data tests. Data tests are queries designed to check specific conditions in your data, and they return one row per failed test case (for example, the number of values with duplicates for the unique test). dbt reports the number of failing records as failures. Whereas, each unit test represents one 'test case', so results are always 0 (pass) or 1 (fail) regardless of how many records failed within that test case.

Learn about exit codes for more information.

Additional resources​

100
Edit this page