JupyterLab: Extensions, Magic Commands & Best Practices

Magic Commands

# Line magics (single %)
%timeit arr.sum()               # benchmark a single expression
%time result = slow_function()  # time a single execution
%run script.py                  # execute a .py file
%load script.py                 # load file contents into cell
%who                            # list all variables
%whos                           # list variables with type/value
%reset                          # clear all variables
%matplotlib inline              # render plots inline
%matplotlib widget              # interactive plots (ipympl)
%env MY_VAR=value               # set environment variable
%pwd                            # print working directory
%ls                             # list files
%cd /path/to/dir                # change directory
%history                        # show cell history

# Cell magics (%% — apply to entire cell)
%%timeit
result = [x**2 for x in range(10000)]

%%bash
echo "Hello from bash"
ls -la

%%writefile myfile.py
def hello():
    print("Hello")

%%capture output
import subprocess
result = subprocess.run(['ls'], capture_output=True)

# IPython display utilities
from IPython.display import display, HTML, Image, Markdown, JSON
display(HTML('<b>Bold HTML</b>'))
display(Markdown('# Markdown heading'))
display(Image('plot.png'))

Popular Extensions

# Extensions install via pip (JupyterLab 3+)
pip install jupyterlab-git           # Git integration in sidebar
pip install jupyterlab-lsp           # Language Server Protocol (autocomplete, hover)
pip install python-lsp-server        # Python LSP backend for jupyterlab-lsp
pip install jupyterlab_code_formatter # Black/isort formatting
pip install black isort              # formatters for above
pip install ipympl                   # interactive matplotlib (%matplotlib widget)
pip install jupyterlab-drawio        # Diagram editor
pip install elyra                    # AI/ML pipeline builder
pip install nbdime                   # Notebook diffing/merging

# Check installed extensions
jupyter labextension list

nbconvert — Export Notebooks

# Export to other formats
jupyter nbconvert notebook.ipynb --to html
jupyter nbconvert notebook.ipynb --to pdf
jupyter nbconvert notebook.ipynb --to script   # .py file
jupyter nbconvert notebook.ipynb --to markdown
jupyter nbconvert notebook.ipynb --to slides   # reveal.js

# Execute and export (re-runs all cells first)
jupyter nbconvert notebook.ipynb --to html --execute
jupyter nbconvert notebook.ipynb --execute --inplace  # save results back

# Papermill: parameterized execution
pip install papermill
papermill notebook.ipynb output.ipynb -p alpha 0.1 -p n_epochs 10

Best Practices

• Restart kernel and run all cells (Kernel → Restart Kernel and Run All) before sharing — catches hidden state bugs.

• Keep cells short and focused. Long cells with many side effects are hard to re-run selectively.

• Use %store variable_name to persist variables across sessions (stored in IPython profile).

• Name notebooks descriptively with dates: 2024-03-15-eda-user-churn.ipynb.

• Use nbstripout (pre-commit hook) to strip output before git commits — keeps diffs clean.

• For production code, refactor notebook logic into .py modules; import and call from notebook.

• Use papermill for automated notebook execution with parameter injection (CI/CD reports).

• JupyterHub: multi-user server for teams — each user gets an isolated server session.