Turn your Python CLI program into a modern web-based UI
Based on Gooey and created using nicegui, NiceGooey transforms your CLI tool written with Python's argparse into a modern web UI with just a single line of code.
(NiceGooey is currently released as a Beta. It is functional and no major refactorings are planned, but some breaking changes and bug might happen.)
This project is published at Codeberg and mirrored to GitHub. It is available on PyPi as nicegooey.
Just importing and adding the annotation @nice_gooey_argparse_main(patch_argparse=True) to your main function is
enough to start using NiceGooey.
from nicegooey.argparse import nice_gooey_argparse_main
import argparse
@nice_gooey_argparse_main(patch_argparse=True)
def main():
parser = argparse.ArgumentParser()
parser.add_argument("--my-name", type=str, required=True)
namespace = parser.parse_args()
print(f"Hi, {namespace.my_name}")Without using NiceGooey, this is a simple command line tool for which the call tool --my-name=Max would print Hi, Max to the terminal.
With NiceGooey, this renders a web page with a text input and a submit button that shows "Hi, Max" in your browser.
If the simple one-liner patch is not enough for you, a more type-safe and configurable version of NiceGooey
can be used by using the NgArgumentParser class instead of argparse.ArgumentParser.
See below for more info about nicegooey_config.
from nicegooey.argparse import nice_gooey_argparse_main, NgArgumentParser, NiceGooeyConfig
from nicegooey.argparse.ui_classes.actions.action_alternatives import store_action_slider
@nice_gooey_argparse_main(patch_argparse=False)
def main1(*args, **kwargs):
parser = NgArgumentParser()
my_name_action = parser.add_argument("--my-name", type=str, required=True)
my_age_action = parser.add_argument("--my-age", type=int, required=True)
parser.nicegooey_config = NiceGooeyConfig(
root_card_class="max-w-4xl", # don't take up the entire screen on wide desktops
action_element_overrides={
# Display the age as a slider instead of a number field.
my_age_action: store_action_slider(min=0, max=100, step=1)
}
)
namespace = parser.parse_args()
print(f"Hi, {namespace.my_name}")
By setting the nicegooey_config field of an NgArgumentParser instance, you can configure the style and behavior of your page.
root_card_class
A space-separated list of Tailwind classes to be applied to the root of the argument UI.
Especially useful to pass something like 'w-4xl' to limit the width on large screens.
action_config
A dict that maps parser action objects to ActionConfig classes.
Use this to set configurations affecting individual actions.
element_override
If set, this class is instantiated to display the widget for this element in the UI instead of the default.
For example, use this to render a slider with min and max limits, instead of a numeric input field.
Look at nicegooey.argparse.ui_classes.actions.action_alternatives for predefined options.
validation
A callable that is run when "Submit" is pressed. The value of the action is passed as the argument.
If the function returns a string, the process is stopped and the string is shown as an error.
NiceGooey is published under LGPL. You are almost certainly allowed to use it for whatever your use case is, but are required to publish a copy of LGPL with it. This is covered by NiceGooey already, which shows a "License" button at the bottom of the page.
Generative AI and coding agents being a hot topic at the moment, especially in Open Source projects, this section is about disclosing any use of AI within this project, for anyone who is intersted:
- There is no AI of any kind being called from the NiceGooey code itself.
- Anthropic's Claude Code was used in some occasions of:
- writing unit tests
- creating showcase examples
- refactorings on the implementation