phillipdupuis / Dtale Desktop

An interface for saving python scripts as permanent D-Tale launch points.

Getting started

Installation

$ pip install dtaledesktop

Running it from the command line:

$ dtaledesktop

Running it from a python script:

import dtale_desktop

dtale_desktop.run()

Motivation

dtaledesktop simplifies the process of fetching data, cleaning/transforming it, and then performing exploratory data analysis. With dtaledesktop, that entire process is condensed into a single click.

It does this by providing a dashboard GUI, and any python code which returns a pandas DataFrame can be saved to the dashboard as a widget. Users can then execute that code and explore the DataFrame in dtale or pandas-profiling by simply clicking one of the widget buttons. The code associated with that widget can also be edited directly from the dashboard, and upon doing so the dashboard is updated in real-time.

Here's a simple example of using this workflow:

1. launch dtaledesktop from a terminal. The dashboard automatically opens up in your web browser.
2. click the 'Add Data Source' button, fill out the form like so, and click save:
3. and bam! You now have a new 'Stocks' section on your dashboard. It will be there every time you launch dtaledesktop, and the python code can be edited directly from the dashboard:

If at some point you decide you want to watch Apple too, all you need to do is click the "Settings" button and add "AAPL" to the list of stock symbols. It will immediately appear in the dashboard below TSLA.

How it works

The front end is written with react, using a mixture of ant-design and styled-components.

The back end is written in python, and it actually consists of TWO apps which listen on separate ports. The main one is an asynchronous FastAPI application, and it responsible for communicating with the dashboard, interacting with the file system, and executing user-defined code for fetching/transforming data. It is able to do this by saving the submitted code as persistent files and then using importlib.util to build and then import the resulting modules. The second app is for running dtale instances, and it is a synchronous flask application.

Developers/Contributing

First, you'll want to clone the repo and install the python dependencies:

$ git clone https://github.com/phillipdupuis/dtale-desktop.git
$ cd dtale-desktop
$ python setup.py develop

Then you'll need to install the javascript dependencies and build the react app:

$ cd dtale_desktop/frontend
$ npm install
$ npm run build

And now you should be able to launch it like so:

$ python dtale_desktop/app.py

If you want to modify the front-end, do the following:

1. Launch the python app in the normal way
2. Change the "proxy" setting in frontend/package.json to point at the host/port the python app is running on.
3. npm start to launch the react app. It will run on a different port, but will proxy unknown requests to the python app.

Settings

Disabling features:

Routing requests:

Loaders/file storage: