My collection of tips and tricks for using jupyter lab.


Jupyterlab is already installed with Anaconda, but if using miniconda install it by:

conda install -c conda-forge jupyterlab ipywidgets

To make plotly work with Jlab you need a bunch of extensions, so putting details here:

conda install -c plotly plotly

magic commands

list all magic commands: %lsmagic

My shortlist:

  • Load a file: %load ./
  • time a cell: %%time


Jupyterlab extensions are built using nodejs, so install that by:

step 1: install nvm, a script to install nodejs:

curl -o- | bash

step 2: install nodejs itself by:

nvm install node

Now install all the Jupyter extensions you want. The below installs all the extensions then builds it in one go by:

jupyter labextension install @jupyter-widgets/jupyterlab-manager --no-build && \
jupyter labextension install @jupyterlab/toc --no-build && \
jupyter labextension install jupyterlab-plotly --no-build && \
jupyter labextension install plotlywidget --no-build && \
jupyter lab build --dev-build=False --minimize=False

Note: the build flags are needed (on my install) cause building was taking too long without this. It needs a better simple way to install extensions without needing node.

blog with jupyter notebooks

Note: just use fastpages to blog with notebooks, but otherwise read on.

So there are tools like Pelican which are setup directly with jupyter notebooks, but I had already set up a hugo powered blog since its free, easy and just works, so my mini shortcut is to put all my notebooks the hugo content directory and run a little script there which converts the jupyter notebooks to markdown files:

import os

# path to jupyter notebooks 
path = 'content/'

# Find all jupyter notebooks in the content folder (and its subfolders)
all_ipynb_files = [os.path.join(root, name)
                   for root, dirs, files in os.walk(path)
                       for name in files
                           if name.endswith((".ipynb"))]

# Remove all notebooks from checkpoint folders
ipynb_files = [x for x in all_ipynb_files if ".ipynb_checkpoints" not in x]

# converting to markdown, one at a time
for notebook in ipynb_files:
    os.system(f'jupyter nbconvert --to markdown {notebook}')

print(f"Converted {len(ipynb_files)} jupter notebooks to markdown.")

The above works just fine, but i had a problem: the web server wasn’t displaying images from the jupyter notebook posts.

One solution is to go through every markdown file and fix the link so it points to the right file, but the easiest solution Google found me was this:

In the nbconvert config file ~/.jupyter/ put the following line:

c.NbConvertApp.output_files_dir = '{notebook_name}'

This saves the output file in a folder which is the same as the notebook name - I’ve found that way then the netlify webserver just serves up the image files without having to rewrite the image path inside the markdown file.

Note: this post is a work in progress…