Using dbt with Dagster, part one: Set up the dbt project#

First, you'll download Dagster's tutorial_dbt_dagster example and configure the dbt project it contains. This example uses dbt's jaffle shop project, but you can follow along with a different project as well.

In this part, you'll:

• Download the tutorial_dbt_dagster example
• Create a DuckDB connection profile
• Configure dbt model data sources

Step 1: Download the tutorial_dbt_dagster example#

Let's get started by downloading the tutorial_dbt_dagster example.

1. In the terminal, run the following to download the example:

   dagster project from-example --name tutorial_dbt_dagster --example tutorial_dbt_dagster
   

2. Navigate into the tutorial_dbt_dagster folder:

   cd tutorial_dbt_dagster
   

3. Run the following to install the project's dependencies:

   pip install -e ".[dev]"
   

Step 2: Create a DuckDB profile#

In this step, you'll add a DuckDB profile to the dbt project. Profiles in dbt allow you to connect dbt to a database - in this case, that's DuckDB.

1. Open the profiles.yml file, located in /tutorial_template/jaffle_shop/config.

2. Add the following code to the file:

   # /tutorial_template/jaffle_shop/config/profiles.yml
   
   jaffle_shop:
     target: local
     outputs:
       local:
         type: duckdb
         path: tutorial.duckdb
         schema: jaffle_shop
   

Step 3: Configure dbt model data sources#

Next, you'll tell the dbt models where their source data will be located. There isn't any data yet - we'll create Dagster assets that fetch and provide data to the dbt models later in the tutorial.

In dbt, sources are declared using sources.yml files. We've provided this file for you in /tutorial_template/jaffle_shop/models, which defines the location of tables containing source data:

# /tutorial_template/jaffle_shop/models/sources.yml

version: 2

sources:
  - name: jaffle_shop
    tables:
      - name: orders_raw
      - name: customers_raw

The Dagster assets you build in part three of this tutorial will create orders_raw and customers_raw tables when materialized. To have the models select data from these tables, you'll need to update the models.

In /tutorial_template/jaffle_shop/models/staging, update the stg_customers.sql and stg_orders.sql models to use the {{ source()}} function. This function tells the dbt model to select data from the defined source. In this example, that's the customers_raw and orders_raw tables in the DuckDB database:.

1. In stg_customers.sql, replace its contents with the following:

   # /tutorial_template/jaffle_shop/models/staging/stg_customers.sql
   
   select
       id as customer_id,
       first_name,
       last_name
   from {{ source('jaffle_shop', 'customers_raw') }}
   

2. In stg_orders.sql, replace its contents with the following:

   # /tutorial_template/jaffle_shop/models/staging/stg_orders.sql
   
   select
       id as order_id,
       user_id as customer_id,
       order_date,
       status
   from {{ source('jaffle_shop', 'orders_raw') }}
   

What's next?#

At this point, you should have a fully-configured dbt project that's ready to work with Dagster. The next step is to load the dbt models into Dagster as assets.