Jan 14 2022 · Piotr Płoński

Share Jupyter Notebook as web app

Mercury convert Jupyter notebook to web appHave you ever wanted to share your Jupyter notebook with non-programming friends? Would you like to hide the code, to not scare them? Give them the option to change the input with interactive widgets. In this post, I will show you how to easily share a notebook as an interactive web app with the Mercury framework.

Mercury

Mercury is an open-source framework that is perfect for sharing multiple notebooks as web applications. You can turn a notebook into a web app without changing any code in the notebook. You just need to specify a YAML header as a first raw cell.

You can read more about Mercury at GitHub: https://github.com/mljar/mercury.

In this post, I will show you step-by-step how to create a very simple notebook and deploy it with Mercury to Heroku (using free dyno).

Setup GitHub repo

Please create a new repo for our project. You can do this by going to https://github.com/new. Below is my filled form to create the repo.

Create GitHub repository

After repository creation, you should see a view similar to mine. Please click on a green Code button and copy the address to clone the repository.

Create GitHub repository

Clone the repository with the command:

git clone git@github.com:pplonski/mercury-simple-demo.git

Set virtual environment and install requirements

Let's change the directory to the project folder. We will use virtualenv to create a virtual environment for our python packages.

virtualenv simple_venv
source simple_venv/bin/activate

Let's create a requirements.txt file with the Mercury package:

mljar-mercury

Please install the needed package by running:

pip install -r requirements.txt

This should install the required packages and Jupyter Notebook. To use the new virtual environment in the Jupyter run:

python -m ipykernel install --user --name=simple_venv

Create demo Notebook

Please start Jupyter Notebook with the following command:

jupyter notebook

Let's create a new notebook. Remember to select a kernel with the name of the virtual environment.

Create new Jupyter notebook

Change the name of the notebook by clicking the Untitled at the top. I set greetings as a new notebook name. It will be a very simple notebook. Please add a code:

print("Cześć 2022")

Run the cell by clicking the Run button in the top menu or by Ctrl+Enter. You should see

Simple notebook in Jupyter

OK. Let's say our very simple notebook is ready. You would like to share it with friends. Your friends don't speak Polish, so they would like to see Hi instead of Cześć. What if your friends would like to use an application in other than the 2022 year?

Add parameters to notebook

We need to parametrize our notebook. Let's add one more cell at the top (click Insert at the top menu). In the first cell we will define two variables:

greetings = "Cześć"
year = 2022

and our greetings message will be in the second cell:

print(f"{greetings} {year}")

It should look like below:

Notebook with parameters in Jupyter

OK, so our notebook has two variables that can be changed, and the notebook can be executed. Is it ok to share it with friends? If they are comfortable with Python and Jupyter then yes. If they can't code and don't know Python, then Mercury is to the rescue.

Define widgets

We will use Mercury to convert our simple notebook to a web application with two interactive widgets. Let's start the Mercury in watch mode to see our app development. Please open a separate terminal in the project's folder and activate the virtual environment:

source simple_venv/bin/activate

Then start watch mode in Mercury:

mercury watch greetings.ipynb 

It will start the server at http://127.0.0.1:8000/. Please copy the address and paste it to your web browser. You should see your notebook.

Mercury gallery watch mode

Please click Open button to see your notebook running as a web app. You should see view similar to the image below:

Mercury notebook watch mode

As you can see from the warning message, we need to define the YAML config in the notebook. Please go to the Jupyter tab and add a new cell at the top of the notebook.

Notebook with raw cell

Please set the raw cell content to:

---
title: My app
description: My first notebook shared on Mercury
params:
    greetings:
        input: select
        label: Select greetings
        value: Cześć
        choices: [Cześć, Hello, Hi, Salut, Ciao]
    year:
        input: slider
        label: Select year
        value: 2022
        min: 2021
        max: 2030
---

Your notebook should look like:

Greetings Notebook

Your web app in Mercury should auto-refresh after you save the Notebook.

Greetings app

You can now change widgets' values and click Run to see recomputed values.

If you would like to hide the code, you can add the parameter show-code: False before params definition in YAML config. To read more about available widgets please check our readme in the Mercury repo.

Deploy to Heroku

I assume that you have Heroku account and a Heroku CLI installed. Please login to your Heroku account in the command line:

heroku login

Let's create a free dyno by running:

heroku create mercury-simple-demo

The string after the create command will be the URL address of your app. It should be unique across Heroku. If you just run heroku create then some random name will be generated.

We need to add a Procfile to the project. It will tell Heroku how to run our app. The Procfile will have only one line:

web: mercury run 0.0.0.0:$PORT

OK, let's add our files to GitHub:

git add greetings.ipynb
git add requirements.txt
git add Procfile
git commit -am "initial commit with simple notebook"
git push

Now we are ready to deploy to Heroku:

git push heroku main

Please wait for a little till all packages are installed. And that's all! Our notebook should be running as a web app available to others.

My is running at http://mercury-simple-demo.herokuapp.com/.

Mercury on Heroku

Thank you!

My repository with a simple notebook: https://github.com/pplonski/mercury-simple-demo

The Mercury GitHub repository: https://github.com/mljar/mercury

Become a Data Science wizard, today!

Forget about Python problems, just do your work.

MLJAR Studio