dbt-coverage

One-stop-shop for docs and test coverage of dbt projects.

Optimized for dbt 1.0, see full support matrix.

Why do I need something like this?

dbt-coverage is to dbt what coverage.py and interrogate are to Python.

It is a single CLI tool which checks your dbt project for missing documentation and tests.

Keeping documentation and tests close to the actual SQL code that generates the final model is one of the best design choices of dbt. It ensures documentation is actually useful and tests are actually used. But how do you make adding those a habit in your dbt project?

That is exactly where dbt-coverage comes in. It will

• Give you a better sense of the level of documentation and test coverage in your project;
• Help your CI/CD pipeline make sure new changes include documentation and tests;
• Let you quickly assess the documentation and tests of a new dbt project you get your hands on.

Still not convinced? Here are some more features:

• ✨ zero-config: just install it and run it, there is nothing to set up
• 🏁 minimal dependences: the only dependencies are click (already installed with dbt) and typer
• 📦 very small: at ~480 SLOC, you can easily validate it works as advertised

Demo

The package was presented during Coalesce, the annual dbt conference, as a part of the talk From 100 spreadsheets to 100 data analysts: the story of dbt at Slido. Watch a demo in the video below.

Installation

pip install dbt-coverage

Usage

dbt-coverage comes with two basic commands: compute and compare. The documentation for the individual commands can be shown by using the --help option.

Compute

Compute coverage from target/catalog.json and target/manifest.json files found in a dbt project, e.g. jaffle_shop.

To choose between documentation and test coverage, pass doc or test as the CLI argument.

$ cd jaffle_shop
$ dbt run  # Materialize models
$ dbt docs generate  # Generate catalog.json and manifest.json
$ dbt-coverage compute doc --cov-report coverage-doc.json  # Compute doc coverage, print it and write it to coverage-doc.json file

Coverage report
=====================================================================
jaffle_shop.customers                                  6/7      85.7%
jaffle_shop.orders                                     9/9     100.0%
jaffle_shop.raw_customers                              0/3       0.0%
jaffle_shop.raw_orders                                 0/4       0.0%
jaffle_shop.raw_payments                               0/4       0.0%
jaffle_shop.stg_customers                              0/3       0.0%
jaffle_shop.stg_orders                                 0/4       0.0%
jaffle_shop.stg_payments                               0/4       0.0%
=====================================================================
Total                                                 15/38     39.5%

$ dbt-coverage compute test --cov-report coverage-test.json  # Compute test coverage, print it and write it to coverage-test.json file

Coverage report
=====================================================================
jaffle_shop.customers                                  1/7      14.3%
jaffle_shop.orders                                     8/9      88.9%
jaffle_shop.raw_customers                              0/3       0.0%
jaffle_shop.raw_orders                                 0/4       0.0%
jaffle_shop.raw_payments                               0/4       0.0%
jaffle_shop.stg_customers                              1/3      33.3%
jaffle_shop.stg_orders                                 2/4      50.0%
jaffle_shop.stg_payments                               2/4      50.0%
=====================================================================
Total                                                 14/38     36.8%

Filtering model paths with --model-path-filter or --model-path-exclusion-filter

You can also choose a subset of tables to compare using one or multiple --model-path-filter and/or --model-path-exclusion-filter options. Here are some examples.

Use the --model-path-filter.

$ cd jaffle_shop
$ dbt run  # Materialize models
$ dbt docs generate  # Generate catalog.json and manifest.json
$ dbt-coverage compute doc --cov-report coverage-doc.json --model-path-filter models/staging/

Coverage report
======================================================
jaffle_shop.stg_customers              0/3       0.0%
jaffle_shop.stg_orders                 0/4       0.0%
jaffle_shop.stg_payments               0/4       0.0%
======================================================
Total                                  0/11      0.0%

Use the --model-path-exclusion-filter.

$ dbt-coverage compute doc --cov-report coverage-doc.json --model-path-exclusion-filter models/staging/

Coverage report (doc)
=====================================================================
dbt_sweco.customers                                    6/7      85.7%
dbt_sweco.orders                                       9/9     100.0%
dbt_sweco.raw_customers                                0/3       0.0%
dbt_sweco.raw_orders                                   0/4       0.0%
dbt_sweco.raw_payments                                 0/4       0.0%
=====================================================================
Total                                                 15/27     55.6%

Use multiple paths. The same can be done with --model-path-exclusion-filter.

$ dbt-coverage compute doc --cov-report coverage-doc.json --model-path-filter models/orders.sql --model-path-filter models/staging/

Coverage report
======================================================
jaffle_shop.orders                     0/9       0.0%
jaffle_shop.stg_customers              0/3       0.0%
jaffle_shop.stg_orders                 0/4       0.0%
jaffle_shop.stg_payments               0/4       0.0%
======================================================
Total                                  0/20      0.0%

Use both --model-path-filter and --model-path-exclusion-filter.

$ dbt-coverage compute doc --cov-report coverage-doc.json --model-path-filter models/staging --model-path-exclusion-filter models/staging/stg_customers

Coverage report (doc)
=====================================================================
dbt_sweco.stg_orders                                   0/4       0.0%
dbt_sweco.stg_payments                                 0/4       0.0%
=====================================================================
Total                                                  0/8       0.0%

Markdown output with --cov-format

You can also choose to print the output in the Markdown table format by specifying the --cov-format option. This can be especially useful when using dbt-coverage in CI/CD pipelines.

$ cd jaffle_shop
$ dbt run  # Materialize models
$ dbt docs generate  # Generate catalog.json and manifest.json
$ dbt-coverage compute doc --model-path-filter models/staging/ --cov-format markdown

# Coverage report
| Model | Columns Covered | % |
|:------|----------------:|:-:|
| jaffle_shop.stg_customers                         |     0/3     |   0.0% |
| jaffle_shop.stg_orders                            |     0/4     |   0.0% |
| jaffle_shop.stg_payments                          |     0/4     |   0.0% |
| Total                                             |     0/11    |   0.0% |

Custom run artifacts path with --run-artifacts-dir

To compute the coverages, dbt-coverage looks up the artefacts from the dbt run execution in the ./target/ folder in the current directory. You can specify a custom path via the --run-artifacts-dir option.

$ dbt-coverage compute doc --run-artifacts-dir jaffle_shop/target --cov-report coverage-doc.json  # Compute doc coverage from the artefacts located in jaffle_shop/target, print it and write it to coverage-doc.json file

Coverage report
================================================
jaffle_shop.customers             0/7       0.0%
jaffle_shop.orders                0/9       0.0%
jaffle_shop.raw_customers         0/3       0.0%
jaffle_shop.raw_orders            0/4       0.0%
jaffle_shop.raw_payments          0/4       0.0%
jaffle_shop.stg_customers         0/3       0.0%
jaffle_shop.stg_orders            0/4       0.0%
jaffle_shop.stg_payments          0/4       0.0%
================================================
Total                             0/38      0.0%

Compare

Compare two coverage.json files generated by the compute command. This is useful to ensure that the coverage does not drop while making changes to the project.

$ dbt-coverage compare coverage-after.json coverage-before.json

# Coverage delta summary
              before     after            +/-
=============================================
Coverage      39.47%    38.46%         -1.01%
=============================================
Tables             8         8          +0/+0
Columns           38        39          +1/+0
=============================================
Hits              15        15          +0/+0
Misses            23        24          +1/+0
=============================================

# New misses
=========================================================================
Catalog                         15/38   (39.47%)  ->    15/39   (38.46%)
=========================================================================
- jaffle_shop.customers          6/7    (85.71%)  ->     6/8    (75.00%)
-- new_col                       -/-       (-)    ->     0/1     (0.00%)
=========================================================================

Combined use-case

$ cd my-dbt-project

$ dbt run  # Materialize models
$ dbt docs generate  # Generate catalog.json and manifest.json
$ dbt-coverage compute doc --cov-report before.json --cov-fail-under 0.5  # Fail if coverage is lower than 50%

# Make changes to the dbt project, e.g. add some columns to the DWH, document some columns, etc.

$ dbt run  # Materialize the changed models
$ dbt docs generate  # Generate catalog.json and manifest.json
$ dbt-coverage compute doc --cov-report after.json --cov-fail-compare before.json  # Fail if the current coverage is lower than coverage in before.json
$ dbt-coverage compare after.json before.json  # Generate a detailed coverage delta report

Supported dbt versions

Different versions of dbt-coverage support different versions of dbt. Here is the support matrix.

dbt versions older than 0.20 are not tested.

Related packages

• https://github.com/mikaelene/dbt-test-coverage
• interrogate (docs coverage for Python)
• coverage.py (execution coverage for Python)

Contributing

Clone this repo including submodules, create a virtual environment and install dependencies:

git clone --recurse-submodules git@github.com:slidoapp/dbt-coverage.git
cd dbt-coverage
pip install poetry
poetry shell
poetry install
pre-commit install

To run all integration tests locally, run:

tox

License

Licensed under the MIT license (see LICENSE.md file for more details).