GitHub - Ben-Epstein/sqlmesh: Efficient data transformation and modeling framework that is backwards compatible with dbt.

SQLMesh is a next-generation data transformation and modeling framework that is backwards compatible with dbt. It aims to be easy to use, correct, and efficient.

SQLMesh enables data teams to efficiently run and deploy data transformations written in SQL or Python.

It is more than just a dbt alternative.

Core Features

Get instant SQL impact analysis of your changes, whether in the CLI or in SQLMesh Plan Mode

Virtual Data Environments

See a full diagram of how Virtual Data Environments work
Watch this video to learn more

Plan / Apply workflow like Terraform to understand potential impact of changes
Automatic column level lineage and data contracts
Easy to use CI/CD bot

Efficiency and Testing

Running this command will generate a unit test file in the tests/ folder: test_stg_payments.yaml

Runs a live query to generate the expected output of the model

sqlmesh create_test tcloud_demo.stg_payments --query tcloud_demo.seed_raw_payments "select * from tcloud_demo.seed_raw_payments limit 5"

# run the unit test
sqlmesh test

MODEL (
  name tcloud_demo.stg_payments,
  cron '@daily',
  grain payment_id,
  audits (UNIQUE_VALUES(columns = (
      payment_id
  )), NOT_NULL(columns = (
      payment_id
  )))
);

SELECT
    id AS payment_id,
    order_id,
    payment_method,
    amount / 100 AS amount, /* `amount` is currently stored in cents, so we convert it to dollars */
    'new_column' AS new_column, /* non-breaking change example  */
FROM tcloud_demo.seed_raw_payments

test_stg_payments:
model: tcloud_demo.stg_payments
inputs:
    tcloud_demo.seed_raw_payments:
    - id: 66
    order_id: 58
    payment_method: coupon
    amount: 1800
    - id: 27
    order_id: 24
    payment_method: coupon
    amount: 2600
    - id: 30
    order_id: 25
    payment_method: coupon
    amount: 1600
    - id: 109
    order_id: 95
    payment_method: coupon
    amount: 2400
    - id: 3
    order_id: 3
    payment_method: coupon
    amount: 100
outputs:
    query:
    - payment_id: 66
    order_id: 58
    payment_method: coupon
    amount: 18.0
    new_column: new_column
    - payment_id: 27
    order_id: 24
    payment_method: coupon
    amount: 26.0
    new_column: new_column
    - payment_id: 30
    order_id: 25
    payment_method: coupon
    amount: 16.0
    new_column: new_column
    - payment_id: 109
    order_id: 95
    payment_method: coupon
    amount: 24.0
    new_column: new_column
    - payment_id: 3
    order_id: 3
    payment_method: coupon
    amount: 1.0
    new_column: new_column

Never builds a table more than once
Partition-based incremental models
Unit tests and audits

Take SQL Anywhere

Write SQL in any dialect and SQLMesh will transpile it to your target SQL dialect on the fly before sending it to the warehouse.

Compile time error checking and can transpile 10+ different SQL dialects
Definitions using simply SQL (no need for redundant and confusing Jinja + YAML)
Self documenting queries using native SQL Comments

For more information, check out the website and documentation.

Getting Started

Install SQLMesh through pypi by running:

mkdir sqlmesh-example
cd sqlmesh-example
python -m venv .env
source .env/bin/activate
pip install sqlmesh
sqlmesh init duckdb # get started right away with a local duckdb instance

Follow the quickstart guide to learn how to use SQLMesh. You already have a head start!

Join Our Community

We want to ship better data with you. Connect with us in the following ways:

Join the Tobiko Slack Community to ask questions, or just to say hi!
File an issue on our GitHub
Send us an email at hello@tobikodata.com with your questions or feedback
Read our blog

Contribution

Contributions in the form of issues or pull requests are greatly appreciated. Read more on how to contribute to SQLMesh open source.