// the find
holtzy/The-Python-Graph-Gallery
A website displaying hundreds of charts made with Python
This is the source for python-graph-gallery.com — a Gatsby site with 600+ Jupyter notebooks, each one a self-contained recipe for a specific chart type across matplotlib, seaborn, plotly, and a dozen smaller libraries (plotnine, folium, bumplot, plottable, flexitext). It's for developers who want copy-pasteable code for a specific chart, not a library to import.
The notebook-as-source-of-truth approach means every example is actually executable, not a static code block that's gone stale — you can open the .ipynb and run it. Coverage is genuinely wide: it tracks new small libraries (great-tables, pypalettes, arcplot, morethemes) as they show up, and the family/chartType metadata taxonomy in sectionDescriptions.js gives a real classification system instead of a flat list. Still actively pushed to as of last month, which for a reference site matters more than for most repos.
The whole thing is coupled to a specific Gatsby site build — there's no way to consume this as a library or pull individual examples without cloning the full frontend (gatsby-node, custom ipynb transform plugin, ImageMagick image pipeline). Adding or updating a post is a manual, multi-step process (rename file, hand-edit slug metadata, screenshot at exact pixel dimensions, wire up a JS component) with zero automated checks — nothing verifies a notebook still runs before merge. Notebooks aren't pinned to library versions, so a seaborn or plotly API change can quietly break an example without anyone noticing until a reader reports it.