Skip to main content

Jupyter nbextension for JavaScript execution and managing linked libraries and CSS stylesheets in Jupyter notebooks.

Project description

Jupyter nbextension for JavaScript execution and managing linked libraries and CSS stylesheets in Jupyter notebooks.


The jupyter-require library is intended to be used in Jupyter notebooks.

Jupyter-require allows to execute and manage custom JavaScript and CSS files and even create and load your own styles and scripts directly from Jupyter notebook.

Jupyter-require provides a unique opportunity to customize Jupyter notebooks and enables users to handcraft their own JavaScript-augmented workflows while keeping in mind synchronicity demands and security implications of such approaches.

What is this for?

Let’s demonstrate the usage on an example. Note that the usage is limited only by your imagination, this is just a demonstration of a single use case.

If you are into data visualization like me, you’ve most likely already heard of d3.js JavaScript ecosystem. It’s an incredibly powerful tool which can be used to create advanced interactive visualizations.

However, it is not very comfortable to use in Jupyter notebooks let alone integrate with Python.

See the article about Custom D3.js Visualization in a Jupyter Notebook.

That’s where jupyter-require and related jupyter-d3 come into play. jupyter-require allows you to source custom scripts (like d3) and styles and use them within the notebook with ease.

Check out also jupyter-d3 which takes the d3 workflow in Jupyter notebooks to another level.


To install jupyter-require Python package:

pip install jupyter-require

And to install the nbextension itself and enable it, we have supplied a helper functions in the jupyter-tools utils module.

# this is a very young project and there hasn't been time to turn it into a package yet
pip install -e "git://"
from jupyter_tools import utils

# install jupyter-require extension
utils.install_nbextension('jupyter_require', overwrite=True)  # note there is an underscore, it's Python module name

# load and enable the extension
utils.load_nbextension('jupyter-require', enable=True)

All of that above can be done from command line, so if you’re used to installing nbextensions the regular way, feel free to do so. In fact, you are recommended to, this approach is just for lazy people like myself.

NOTE: You may need to reload the page (just hit F5) after these steps for the jupyter-require nbextension to initialize properly.

Example usage

In Jupyter notebooks:

%load_ext jupyter_require

Loading libraries

Loading required libraries is now as simple as:

%require d3
%require d3-hierarchy

NOTE: Note that the path does **NOT** contain the `.js` file extension. This is `requireJS`_ standard.

The %require is jupyter magic command and the rest are the parameters. The command takes a lib name and path.

Creating custom style elements

If you’re not a fan of magic commands, you can make use of equivalent API calls.

from jupyter_require import require
from jupyter_require import load_css

  'd3': ''
  'd3-hierarchy': ''

load_css(...)  # stylesheet goes here

Executing custom script

Now we can actually make use of the bidirectional communication between Python and JS

Let’s say we have the following d3 script:

NOTE: I assume that we work in Jupyter notebook environment and hence we have the context cell at our disposal.
 * @module
 * @description  Print coloured circles into the cell output
 * @file  d3-simple-example.js

 // Inspired by:

// create SVG element in the output area
// the ``element`` is a contextual binding to the output of the current cell
let svg =

// create group
let g = svg.append('g');

  .attr("cx", 2).attr("cy", 2).attr("r", 40).style("fill", "blue");
  .attr("cx", 140).attr("cy", 70).attr("r", 40).style("fill", "red");
  .attr("cx", 300).attr("cy", 100).attr("r", 40).style("fill", "green");

Now in order to execute the script in a cell, we will have to tell it to use the d3. The execute_with_requirements is made exactly for that purpose.

from pathlib import Path
from jupyter_require import execute_with_requirements

script = Path('d3-simple-example.js').read_text()

execute_with_requirements(script, required=['d3'])
SVG Example generated by d3

And you should see those three pretty circles :point_up: .

⚠️ It is possible that the current markdown renderer does not render the raw </svg> element above, all the more reason to try it yourself! :smirk:

There is certainly more to it, but I am gonna leave it to your adventurous desires.


JavaScript execution is by default asynchronous. All the more in Jupyter notebooks. Executing custom JavaScript script will happen asynchronously and the rest of the notebook won’t wait for the execution to complete.

This is very often not the desired behaviour, since we might to work with the results of the execution in the next cell.

Jupyter-require solves this issue by converting every executed script into Promise and awaiting it while pausing the execution of Python kernel.

Execution & Security – safe scripts and finalization

In Jupyter notebooks, it might be sometimes unfortunate how the JavaScript is stored (and treated) in general in the notebook environment. jupyter-require introduces the notion of safe scripts and finalization. Let’s look at the latter first.


When a user executes a script via native Jupyter API, that is typically something like display(Javascript("""...""")), what happens behind the scenes is actually quite complicated. The one important thing to now, however, is that the whole script is embedded into the cell output and the resulting *.ipynb file. Then, every time a cell is copied or re-created (i.e., on notebook reload), the script is executed. Since this execution is not sandboxed. In fact, it is executed in window context using eval function.

See: The section ‘Do not ever use eval!’ from the official MDN web docs).

This can potentially be a security threat! Also, if you don’t want to share the script that produced the output, but you want the output to be present, this does not happen either.

We try to combat that issue step by step, our approach is not optimal either, but it does yield some improvements and we believe that over time, it will get even better. When executing script with jupyter-require execute_with_requirements function, it is not the script which is embedded, it is the Function object itself which the cells carry with. This allows the script to be re-executed when we copy/paste a cell or stored in a clipboard when cutting the cell.

Also, we do not evaluate the script in window context using the eval function, as Jupyter by default does. Instead, at the current development state, we wrap it in its own Function scope and set its this and element context manually.

Aight, still not a word about finalization, right? What finalization means in this context, is discarding the JavaScript code which produced the output, cleaning the metadata and saving the output displayed in the cell output area into a static state. Going back to the d3 example, finalizing the cells would make the plot that we produced persistent and JSON serializable. The output would then be visible in tools like nbviewer or GitHub ipynb preview.

⚠️ SVG poses another security issue, however, hence GitHub might not display them to prevent that, see for example this conversation. We will try to act on this issue in the future.

We are thinking about the ways we could sandbox the execution and the output even more, but bare in mind that this project is very young, so let’s put one foot in front of the other.

To finalize your outputs, use the Save and Finalize action button which should be present on the right of the regular Save and Checkpoint button. The finalization also happens automatically when you properly close the notebook. We cannot handle SIGTERMs at the moment, so be aware that in that case the scripts will be discarded and the output lost.

Safe scripts

⚠️ The notion of safe scripts is something which has been added pretty recently and is under heavy observation.

By the word safe we don’t refer to an execution which reduces security threats, no, nothing like that. It is YOU who guarantee that the script is safe and can be treated as such. The mechanism which we treat safe scripts by is very similar to the one described above, with one important change: safe scripts are similar to the default Jupyter notebook behaviour in a sense that they are also executed on the notebook reload and are also stored in the resulting `*.ipynb` notebook file.

Hence you can enjoy the benefits of a sandbox(ish) synchronous execution while still having the scripts stored in the output. The one limitation is that they do not allow to specify requirements as the execute_with_requirements function does by its required parameter. This is because those scripts can be executed before extensions are actually loaded and we can not guarantee (at least we don’t know how right now) that the functionality of jupyter-require will be present at that time.

To treat your script as safe script, execute it with safe_execute function.


resource link
MDN web docs

Author: Marek Cermak <>

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Filename, size & hash SHA256 hash help File type Python version Upload date
jupyter_require-0.2.1-py2.py3-none-any.whl (37.1 kB) Copy SHA256 hash SHA256 Wheel py2.py3
jupyter-require-0.2.1.tar.gz (673.3 kB) Copy SHA256 hash SHA256 Source None

Supported by

Elastic Elastic Search Pingdom Pingdom Monitoring Google Google BigQuery Sentry Sentry Error logging AWS AWS Cloud computing DataDog DataDog Monitoring Fastly Fastly CDN SignalFx SignalFx Supporter DigiCert DigiCert EV certificate StatusPage StatusPage Status page