.gitignore

# Python
__pycache__/
*.py[cod]
.venv/
build/
dist/
*.egg-info/

# NOTE: `docs/` is deliberately NOT ignored, and it is not documentation-only. It holds THE
# PUBLISHED FEED — index.html, feed.xml, changes.json, sources.json, a feed per jurisdiction,
# and the JSON Schema — alongside the prose docs. It is not a build artifact; it is the
# product, and it is committed on purpose.
#
# Committing it means the site can be served by GitHub Pages, by any static host, or by
# `python -m http.server` out of a clean clone, with **no build step and no CI run**. That
# matters concretely: this account has an account-wide GitHub Actions spending limit, and a
# feed that only exists after somebody else's billing system agrees to run a workflow is a
# feed that does not exist. It also means a reviewer can see, in the diff, exactly what
# changed in the published bytes before it reaches an org that acts on them.
#
# It lives in `docs/` rather than the `dist/` it used to, because branch-based GitHub Pages
# serves exactly two source paths — `/` or `/docs` — and `/` would publish the source tree.
# The Actions-based deploy that could have served `dist/` will never run here, for the billing
# reason above. So `dist/` is now an ordinary Python build directory again, and is ignored.
#
# The one thing that must never appear in the published bytes is an unreviewed record — which
# is not a gitignore's job to enforce, and is enforced by `make no-unreviewed-in-feed`.

# Tooling
.DS_Store
.pytest_cache/
.mypy_cache/
.ruff_cache/
.coverage
coverage.xml
htmlcov/

# The snapshot store: fetched government pages, regenerable, and not something to
# commit (it grows without bound in a git history and adds nothing a reviewer needs).
var/