A New Way to Edit Jupyter Python Notebooks in Emacs

Monday, May 5, 2025

Table of Contents

I created the Jupyter Ascending package to facilitate editing and executing code in a Jupyter Python notebook from Emacs. It provides commands which utilize the Jupytext and Jupyter Ascending command line tools to

Demo

jup-basic.gif

Figure 1: Python usage

jup-markdown.gif

Figure 2: Markdown usage

jup-setup.gif

Figure 3: Starting up a session

Features

  • Synchronization: Editing and saving the Python buffer automatically updates the Jupyter notebook
  • Cell execution commands: Run individual cells or the entire notebook
  • Navigation tools: Jump between cells with simple commands
  • Cell management: Create new cells and toggle between code and markdown types
  • Enhanced markdown editing:
    • Edit markdown cells in dedicated markdown buffers (similar to Org mode's special edit mode)
    • Automatic comment insertion when pressing return in markdown cells
  • Setup utilities: Commands for starting Jupyter notebooks and creating synchronized file pairs

Installation

Jupyter Ascending is available on MELPA.

Prerequisites

The following dependencies must be installed first.

If you haven't already installed Jupyter:

pip install notebook

Then install the Jupyter Ascending command line tool.

pip install jupyter_ascending &&
python3 -m jupyter nbextension    install jupyter_ascending --sys-prefix --py && \
python3 -m jupyter nbextension     enable jupyter_ascending --sys-prefix --py && \
python3 -m jupyter serverextension enable jupyter_ascending --sys-prefix --py

Installing Jupyter Ascending (the command line tool) should install jupytext, but if not, run pip install jupytext.

Configuration

Here's a sample installation which

  • provides keybindings for common commands in jupyter-ascending-mode and
  • uses the minor mode when opening python files with the .sync suffix just before the file extension.
(use-package jupyter-ascending
  :ensure t
  :hook (python-mode . (lambda ()
                         (when (and buffer-file-name
                                    (string-match-p "\\.sync\\.py\\'" buffer-file-name))
                           (jupyter-ascending-mode 1))))
  :bind (:map jupyter-ascending-mode-map
              ("C-c C-k" . jupyter-ascending-execute-line)
              ("C-c C-a" . jupyter-ascending-execute-all)
              ("C-c C-n" . jupyter-ascending-next-cell)
              ("C-c C-p" . jupyter-ascending-previous-cell)
              ("C-c t" . jupyter-ascending-cycle-cell-type)
              ("C-c '" . jupyter-ascending-edit-markdown-cell)))

Usage

Create a notebook pair with M-x jupyter-ascending-create-notebook-pair RET example RET Or, equivalently python3 -m jupyter_ascending.scripts.make_pair --base example This creates synced files: example.sync.py and example.sync.ipynb

If you have an existing Jupyter notebook, create a python file from it: M-x jupyter-ascending-convert-notebook RET example.ipynb RET Or, equivalently, jupytext --to py:percent <file_name> and then add the .sync suffix to both files

Start Jupyter and open the notebook: With example.sync.py open, M-x jupyter-ascending-start-notebook Or, equivalently, python3 -m jupyter notebook example.sync.ipynb

Within your Python file, a line starting with # %% deliminates a new cell.

# %% [markdown]
# This is a markdown cell.
# Below is a code cell.

# %%
def g(x):
  return 3*x

When you edit and save the Python source file, you should see the Jupyter notebook running in your browser update straight away.

I recommend you create keybindings for all the following commands:

Command Description
jupyter-ascending-execute-line Execute cell at cursor
jupyter-ascending-execute-all Execute all cells
jupyter-ascending-next-cell Move to next cell or create new cell
jupyter-ascending-previous-cell Move to previous cell
jupyter-ascending-edit-markdown-cell Edit Markdown cell in dedicated buffer
jupyter-ascending-cycle-cell-type Toggle between Markdown and code cells

Limitations

This package only works with Python notebooks because the Jupyter Ascending command line tool only supports Python notebooks. See here: https://github.com/imbue-ai/jupyter_ascending/issues/25

Alternatives

I think this blog post from 2023 gives a good overview of the other options for editing Jupyter Notebooks from Emacs. It covers emacs-jupyter and ein.

Motivation

I created this package because many of my Computer Science courses force me to use Jupyter notebooks. Of course, I would rather use Org mode, but that's life. And I've since come to appreciate the ability to have auto-completions while editing notebooks, or just treat the notebook as a normal Python file and execute it without Jupyter. I was aware of the other Emacs packages for working with Jupyter, which have the benefit of not being limited to just Python. But this package suits my personal taste very well, so I'm a happy user.

Date: 2025-05-05 Mon 00:00