Last week I was taking the free dbt Fundamentals course offered by dbt labs, and the only options to take it "easily" are using cloud data warehouses, or starting a Postgres/MySQL/some-db container or application locally. I didn't want the hassle, or more like, I wanted to configure the dbt parts in the easiest way possible.

Luckily there is dbt-sqlite.

Sadly, for a lot of dbt things you can't really use sqlite, ¯\_(ツ)_/¯

All commands are defined in a taskfile to make my life easier.

• Technically, a Poetry project, containing helper tools (including dbt via dbt-sqlite);
• A small (and somewhat dirty) Python script (populate.py) to place some not-very-fancy data in a Sqlite file. It will always delete it on load;
• A dbt project with sources (based on this pre-populated data) and models. Includes some tests, too.

Assuming you have poetry installed, you should only need to run poetry install --no-root to install all dependencies. Then, if you want you can also install taskfile to run the commands easily. The Python version is so high because of issues with Python and Mac M1, this one works well with my setup and all libraries I need work well with it.

The two most interesting commands are:

• task check-total: this runs the whole pipe of populating the database and getting a single number out of the final model. They should be the same (total amount of money in paid invoices);
• task dbt-docs: this runs the whole pipe as well, but ends up opening the docs (with lineage) for the models. It's neat to see;
• task dbt-test: you should only run this after having run task dbt-run, particularly if you are tweaking the models.