Skip to content

nicholasyager/hotdag

Repository files navigation

HotDAG

PyPI - Version GitHub Workflow Status (with branch)


HotDAG is a lightweight tool that allows for testing for dbt node selectors against dbt project manifests on a local filesystem, stored on dbt Cloud, or located remotely. Once evaluated, this selection of nodes can be returned as a list of unique_ids, a JSON list of unique_ids, or as an SVG of the selection.

Table of Contents

But Why Though?

Sometimes, it can be valuable while developing locally to have a means of quickly testing a selection string against the production version of a dbt project without fiddling with the generated docs site or compiling the current production branch. Instead, HotDAG lets you quickly test and visualize your selection using your project's manifest, whether its on your local file system, stored remotely and accessible via URL, or hosted on dbt Cloud.

Installation

pip install hotdag

Getting Started

There are a few primary ways to use HotDAG. The most common way is to load a local manifest file and list the selection. This is roughly equivalent to using dbt ls

hotdag --input file --file targets/manifest.json --select "resource_type:source+"

source.example.example.users
model.example.stg_users
model.example.users
test.example.not_null_stg_users_id
test.example.unique_stg_users_id

HotDAG really shines, however, if you want to see a diagram of the selection.

hotdag --input file --output svg \
    --file targets/manifest.json \
    --select "resource_type:source+"

graph.png

Also of use is the ability to directly reference the latest manifest from your project's dbt Cloud job. For example:

hotdag --input dbt_cloud --output svg \
    --account-id 1234 --job-id 9876 \
    --select "resource_type:source+"

Roadmap

  • Styling the SVG output to better distinguish between node types.
  • Improved error handling for remote resource failures.
  • Add rational defaults (e.g. default input to file and --file to the targets directory)
  • Verify support for Selectors (as defined in yaml)

License

hotdag is distributed under the terms of the MIT license.

About

A lightweight tool for evaluating dbt-core selectors against any dbt project manifest.

Resources

License

Stars

Watchers

Forks

Packages

No packages published

Languages