refig 0.1.0

Minimally-invasive reproducibility helpers for matplotlib figures.

refig is a Python library for minimally-invasive reproducibility in Jupyter notebooks, with a focus on figures and plots.

Installation

You can install refig using pip:

pip install refig
uv add refig

Usage

To use refig, simply import it in your Jupyter notebook:

from refig import refig

Then use refig.savefig(NAME) instead of matplotlib.pyplot.savefig(PATH) to save your figures with embedded metadata. NAME can end in .png or .svg.

When you save a figure, refig will automatically save the figure along with a steganographic pointer to the code that generated it, including (all optionally),

• The notebook / file that generated the figure
• The cell number (if applicable / available)
• The date and time the figure was generated
• The git commit hash (if applicable / available)
• The installed refig package version

You can view these metadata by loading the figure with refig:

uv run refig meta figure.png
uv run refig meta figure.svg

Figure Saving Paths

By default, refig saves figures in a figures/ directory. There are several subdirectories:

• figures/latest/: Contains the most recently generated version of each figure.
• figures/history/: Contains all historical versions of generated figures, sorted by timestamp and git commit hash.

For example,

figures/
    latest/
        figure1.png
        figure2.png
    history/
        figure1/
            _20250101_abcdef.png
            _20250201_bcdefa.png
        figure2/
            _20250115_cdefab.png