Bruin CLI

Appearance

Quickstart ​

Make sure you have installed Bruin CLI already in your device.

INFO

The steps here assume you have DuckDB installed. You can replace to any reference to duckdb.sql with your data warehouse if you'd like to run the examples elsewhere.

Create your first pipeline ​

Bruin includes a handy command called init, you can simply run that command to create a new Bruin project.

bash
bruin init default my-pipeline

This command will:

  • Create a project named my-pipeline.
  • Generate a folder called my-pipeline containing the following:
    • An assets folder
    • .bruin.yml file (where you will add connection and credentials )
    • pipeline.yml file to manage your pipeline.

Once you have the project structure, you can run the whole pipeline:

bruin run

Adding a new asset ​

In a nutshell, an asset is anything that generates value with data. In practice, an asset can be a table in your data warehouse, a Python script, or a file in S3. Bruin represents assets in code, put under the assets folder.

In order to create a new asset, you can simply create a new file under the assets folder. Let's start with a SQL asset.

Creating a ingestr asset ​

Let's start by ingesting some data from an external source.

ingestr is an open-source CLI tool that allows ingesting data from any source into any destination.

You can create ingestr assets with a file assets/players.asset.yml

yaml
name: dataset.players
type: ingestr
connection: duckdb
parameters:
  source_connection: chess
  source_table: players

The configuration in the YAML file has a few bits:

  • name: the name of the asset, also the table that will store the ingested data
  • connection: the destination that will be used to load the data
  • parameters: the configuration for ingestr
    • source_connection: the connection that will be used for the source data
    • source_table: the table that will be ingested

That's it, this asset will load data from the chess source and load it into your DuckDB database.

You can run this asset either via the Bruin VS Code extension, or in the terminal:

bash
bruin run assets/players.asset.yml

Creating a SQL asset ​

Let's create a new SQL asset with a file assets/player_stats.sql:

bruinsql
/* @bruin

name: dataset.player_stats
type: duckdb.sql
materialization:
  type: table
   
depends:
   - dataset.players

@bruin */

SELECT name, count(*)
FROM dataset.players
GROUP BY 1

This asset have a few lines of configuration at the top:

  • name: the name of the asset, needs to be unique within a pipeline
  • type: duckdb.sql means BigQuery SQL, Bruin supports many other types of assets.
  • materialization: take the query result and materialize it as a table

Bruin will take the result of the given query, and will create a dataset.player_stats table on DuckDB with it. You can also use view materialization type instead of table to create a view instead.

INFO

Bruin supports many asset types, including BigQuery, Snowflake, Python, Redshift, Databricks, and more.

You can run this asset either via the Bruin VS Code extension, or in the terminal:

bash
bruin run assets/player_stats.sql

Creating a Python asset ​

Similar to SQL, Bruin supports running Python natively as well.

You can create a Python asset with a file assets/my_python_asset.py:

bruin-python
"""@bruin
name: my_python_asset
@bruin"""

print('hello world')
  • The sections between """@bruin and @bruin""" are comment markers, they are the configuration for the asset.
  • The rest of the body is a regular Python script.

You can run this asset either via the Bruin VS Code extension, or in the terminal:

bash
bruin run assets/my_python_asset.sql

At this point, you can also run the whole pipeline:

bash
bruin run

Data Quality Checks ​

Bruin supports data quality checks natively, as part of the asset definition. It includes a handful of data quality checks, and it also supports custom checks.

Let's add a few data quality checks to our table:

bruinsql
/* @bruin

name: dataset.player_stats
type: duckdb.sql
materialization:
  type: table
   
depends:
   - dataset.players

# you can define column metadata and quality checks
columns: 
  - name: name
    type: string
    description: this column contains the player names
    checks: 
      - name: not_null
      - name: unique
  - name: player_count
    type: int
    description: the number of players with the given name
    checks: 
      - name: not_null
      - name: positive

# you can also define custom checks 
custom_checks:   
  - name: row count is greater than zero  
    description: this check ensures that the table is not empty  
    query: SELECT count(*) > 1 FROM dataset.player_count  
   
@bruin */

SELECT name, count(*)
FROM dataset.players
GROUP BY 1

We have added a new columns section in the asset, it contains a list of all the columns and the checks applied to them.

Under the checks section, each column defines some quality checks:

  • name column is marked to be not null and unique
  • player_count column is marked to be not null and consisting of positive numbers

Under the custom_checks section, we have added a new custom quality check that checks if the table is empty or not.

You can refresh the asset & run the quality checks via simply running the asset:

bash
bruin run assets/player_stats.sql

If you'd like to run only the checks, you can run:

bash
bruin run --only checks assets/player_stats.sql

Next steps ​

You have created your first pipeline, ingested data from a new source, added a bunch of assets, and ran quality checks on that, you are ready to dig deeper. Jump into the Concepts to learn more about the underlying concepts Bruin use for your data pipelines.