Arnab aims to offer a simple, cross-platform solution for organizing and executing data models, focusing on core functionalities essential for data workflows. Just like DBT, but is with subset of its features, jinja-less (except for macros), and (theoretically) more portable.

Its main purpose is simple: give it a directory containing SQL files, it will scan the directory recursively and determine the order of execution of all queries.

Simple transformation demo

Create a project directory where you want to put all of your SQL files. Create config.yaml at the root of the directory. At the very least, it should contain models_dir and db_path information:

models_dir: .
db_path: data.duckdb

Optionally, you can create another directories (that may contain subdirectories) and use its name for models_dir.

Check examples directory for more.

Set the working directory to the root of your project, then run arnab run.

We can get the visualization of the pipeline in a SVG file format for an additional way to debug the pipeline. Set the working directory to the root of your project, then run arnab viz outout_name.svg.

• Single executable file
• Cross-platform
• Automatically determine the order of model execution
• Model materialization types:
  • Table
  • View
  • Incremental (probably not necessary for now)
• Macro