dbx-sql-runner 0.1.0

A lightweight SQL transformation tool for Databricks SQL

dbx-sql-runner

A lightweight, library-first SQL transformation tool for Databricks SQL, inspired by DBT.

Features

• Simple SQL Models: Just write .sql files. No complex boilerplate.
• Automated Dependency Management: Reference other models using {upstream_model} and let the runner build the DAG for you.
• Environment Aware: Seamlessly switch between Dev and Prod using profiles.yml and Environment Variables.
• Library Design: Import dbx_sql_runner in your Python scripts (great for Airflow/Databricks Jobs) or run it via CLI.

Installation

Local Development

To install the project in editable mode:

pip install -e .

Production

To install the package normally:

pip install .

Configuration (profiles.yml)

Create a profiles.yml file to store your credentials. Do not commit this file to version control.

server_hostname: "dbc-xxxxxxxx-xxxx.cloud.databricks.com"
http_path: "/sql/1.0/warehouses/xxxxxxxxxxxxxxxx"
access_token: "${DBX_ACCESS_TOKEN}"  # Use env vars for security!
catalog: "my_catalog"
schema: "my_schema"
sources:
    my_source: "prod_catalog.schema.table"

Usage

1. CLI (Easiest)

Run your project from the command line. By default, it looks for profiles.yml in the current directory.

# Run with default profile (profiles.yml)
dbx-sql-runner run

# Run with custom profile
dbx-sql-runner run --profile my_config.yml

# Preview execution plan
dbx-sql-runner build

2. Python (Advanced)

For fine-grained control (e.g., inside a Databricks Job):

from dbx_sql_runner.api import run_project

# Run models in the 'models/' directory using the config from 'profiles.yml'
run_project(models_dir="models", config_path="profiles.yml")

Project Structure

.
├── models/                  # SQL files (.sql)
│   └── example.sql
├── dbx_sql_runner/          # Library source code
│   ├── adapters/            # Database Adapters
│   ├── project.py           # Model Loading & DAG
│   └── runner.py            # Execution Orchestrator
├── profiles.yml             # Configuration (gitignored)
├── pyproject.toml           # Project metadata
└── README.md

Defining Models

Create .sql files in your models/ directory.

• Use header comments for metadata.
• Use {upstream_model} syntax for references (automatically infers dependency).

-- name: my_table
-- materialized: table
-- partition_by: date, region

SELECT * FROM {source_view}
WHERE id > 100