lusca 0.1.3

A Jupyter magic command for creating reproducible matplotlib figures.

lusca

pip install lusca

lusca is a Python library for creating reproducible matplotlib figures using Jupyter magic commands. You often want to use Jupyter for experiments without rerunning the entire notebook to recreate a plot, and saving data alongside figures is essential for artifact generation and reproducibility.

---

%%mplfreeze

%%mplfreeze <name> [vars ...] [--outdir DIR]

• <name>: Base name for outputs (folder + files).
• [vars ...]: Variable names to save into the NPZ file.
• [--outdir DIR]: Parent output directory (default: docs/figs).

The magic command:

• Captures the data used in your plots and saves it in a compressed NPZ file.
• Exports your figures in PDF, PNG, and SVG.
• Generates a minimal standalone script that reproduces the figure.
• Snapshots Python/package versions and the git commit into <name>.meta.json.
• Statically checks the cell for unsaved free names before writing anything, then runs the generated replot in a subprocess to confirm the bundle actually reproduces the figure - if %%mplfreeze succeeds, the replot is guaranteed to work.
• Applies lusca's built-in stylesheet.

---

Example

import matplotlib.pyplot as plt
import numpy as np
import lusca
%load_ext lusca

x_data = np.linspace(-10, 10, 100)
sine = np.sin(x_data)
cosine = np.cos(x_data)

%%mplfreeze trig_demo x_data sine cosine
with plt.style.context("lusca"):
    fig, ax = plt.subplots(1, 1, figsize=(3.5, 2.6), sharey=True)
    ax.plot(x_data, sine, label="Sine")
    ax.plot(x_data, cosine, label="Cosine")
    plt.show()

An example notebook is available in src/demo.ipynb. Generated plots are saved under docs/figs/:

name_stamp/
    name.npz             # saved variables
    name.pdf             # exported figure
    name.png
    name.svg
    name.meta.json       # python/package versions + git commit
    replot_name.py       # standalone replot script

---

[!NOTE] If you are using VS Code, set the workspace root as the default directory for saving figures by adding the following to your settings.json. Otherwise, output paths will be relative to the notebook location.

"jupyter.notebookFileRoot": "${workspaceFolder}"