A tool for making excalidraw figures procedurally from python. Its very easy to visualize excalidraw files either from web (just load the file at excalidraw.com) or with a vs-code plugin (simply open a .excalidraw file in vscode).

Installation

Simpy run python setup.py install to install procXD as a package in your python environment. Then you can create excalidraw files as follows:

from procXD import SketchBuilder, Text
sketch_builder = SketchBuilder()
text_elem = Text(text="Hello, World!")
# Helper example: create bounding boxes
bbox_elem = sketch_builder.create_bounding_element(text_elem, return_group=False)
sketch_builder.draw_objs.extend([bbox_elem, text_elem])
save_path = "temp.excalidraw"
sketch_builder.export_to_file(save_path=save_path)

Usecases

Visualizing large configurations

If you have large configurations, it might be useful to visualize the configuration tree. Helpful for debugging and to confirm variations between different ablations. Detailed example in examples/config_vis.py.

We can visualize configuration (YACS objects) as below using SketchBuilder.render_stack_sketch.

Alternatively, we can also compare different configurations with SketchBuilder.render_comparitive_stack_sketch

Procedural graph objects

You may want to create a excalidraw graph procedurally for precision. Some examples in examples/proc_figures.py.

Create Trees

Create Graphs

NetworkX graph visualization

It might also be useful for converting arbitrary graphs from networkx to excalidraw figures. The benefits are two-folds: SVG format helps manage resolution, and the arrows are bounded between elements which means you can move the graph around!

Todo

1. Support Images.

2. Converter for SVG: Will be nice to convert datasets like FIGR-8 into excalidraw format. That way, output of SVG optimization works (such as Vector-Fusion) can be viewed in excalidraw as well!

What ProcXD does not support

1. Versioning
2. Exact text bounding box size - as a result the text bounding box sizes are adjusted in the file when the file is opened with excalidraw.

Acknowledgements

I used ChatGPT while creating this tool. It was helpful in many ways, but also misleading at times (especially for visual math functions).