schematized config

Install

pip install python_schematized_config

usage

python

from schematized_config.core import ConfigValidator

import os
import os.path as _p

if _p.exists('.env') and 'CONFIG_VALIDATOR_JSON_SCHEMA' in os.environ:
    ConfigValidator.load_dotenv()  # use defaults of .env and CONFIG_VALIDATOR_JSON_SCHEMA

CLI

generate skeleton code for working with a validated .env

this will generate stub files in these conventional locations if they do not yet exist:

• Makefile
• generators/env.schema.jsonnet
• src/schemas/autogen/EnvSchema.py
• python stub code for loading the validated EnvSchema class against .env, as a variable called Env (this is output to stdout)

schematized-config --codegen all > src/env.py

• edit generators/env.schema.jsonnet to add environment variable names and validation
• make schemas/env.schema.json
• make src/schemas/autogen/EnvSchema.py
• optionally update the make rule
• use the src/env.py (or whereever you put it)

development

entering the development environment

assuming you have nix installed and ready, make sure nix-command and flake are enabled (oneliner: run export NIX_CONFIG="experimental-features = nix-command flakes" in the terminal), then enter the dev shell using nix develop

start the jupyter notebook using the provided alias or just jupyter notebook, and hack away

nbdev

this package is developed using nbdev, so we use an nbdev-centric development flow. For a quick guide, we recommend checking out the end-to-end walkthrough. But in short: edit notebooks, then run the nbdev_* management commands. The most essential flow is as follows:

1. edit the notebook files (core fore core, and cli for the command line interface)
2. python setup.py install # note we don’t use nbdev_install because we manage quarto using nix
3. nbdev_prepare
4. run code using the package
5. nbdev_release
6. nbdev_pypi

running tests

if you share variables between cells in your test blocks, this causes trouble during nbdev_prepare as it tries to run cells in isolation, leading to e.g.

NameError: name 'example_properties_schema' is not defined

to deal with this, you can merge cells that use a common variable

updating package dependencies

note that package dependencies are specified in settings.ini; you shouldn’t be editing setup.py by hand. To add a requirement, add it to the requuirements entry in settings.ini, then run python setup.py install