topic Help with cookiecutter Prompts in Databricks Notebooks in Administration & Architecture

Help with cookiecutter Prompts in Databricks Notebooks

vieiralaura — Sun, 17 Nov 2024 16:25:28 GMT

Hi everyone! I’m working on using cookiecutter to help us set up consistent project templates on Databricks. So far, it’s mostly working, but I’m struggling with the prompts – they’re not displaying well in the Databricks environment.

I’ve tried some workarounds, but I’m hoping someone has a simpler solution or knows of any future updates to improve compatibility in notebook environment.

Here’s what I’ve done so far:

def get_user_input():
   project_date = input("What is your project date? YYYYMMDD: ")
   client = input("What is your client name?: ")
   id_project = input("What is your project ID? XXX: ")
   return {
       "project_date ": project_date ,
       "client": client ,
       "id_project ": id_project 
   }

cookiecutter(
    repo_url,
    directory='databricks',
    no_input=True,
    extra_context=get_user_input()
)

I’m looking for tips on making these prompts more user-friendly and avoiding repeated input requests. If anyone has done something similar or has tips for handling cookiecutter in Databricks, I’d really appreciate your advice!

Re: Help with cookiecutter Prompts in Databricks Notebooks

guigasque — Wed, 20 Aug 2025 15:57:15 GMT

Hi! Great initiative using cookiecutter to standardize project templates on Databricks — I’ve been exploring similar workflows.

I ran into the same issue with prompts not displaying well in the notebook environment. One workaround I found effective was running cookiecutter directly in a notebook cell without custom input() functions. Databricks does support interactive prompts to some extent, especially when using %pip install followed by a Python restart (dbutils.library.restartPython()), and then calling cookiecutter() with the template URL.

Here’s a snippet that worked for me:

from cookiecutter.main import cookiecutter

cookiecutter(
    'https://github.com/drivendata/cookiecutter-data-science',
    checkout='v1'
)

This approach triggers the built-in prompts from the template, and they seem to work more reliably than custom input() calls. It also avoids repeated input requests since cookiecutter handles the flow internally.