# Tools

Repeatable import/export and processing helpers. Anything that was done twice by hand belongs here.

## Conventions

- One script, one job. No interactive prompts; every input is an argument.
- Each script documents its input format, output format, and the exact tool version it was tested with.
- Blender scripts run headless (`blender --background --python`), so they can be re-run for a regression check.
- No credentials, no absolute machine paths, no writes outside the project by default.
- A script that produces game content logs what it produced, so `Game/Content` can be audited against it.

## Helpers

- `check-lfs.sh` — repository guard: fails if `git-lfs` or the `lfs` filter is missing, or if an LFS-tracked file was committed as a plain blob.
- `blosm_blockout.py` — headless Blosm import of the committed `.osm` extract into a `.blend`, with a per-building scale report. Exact invocation in `Art/blender/README.md`.
- `commons_fetch.py` — Wikimedia Commons reference fetcher. Search or exact `File:` titles, 1600 px thumbnails, provenance `sidecar.json` per folder. Writes only into the gitignored `Art/reference/`; reference images are never committed.

## Planned helpers
- Blender export preset runner (named presets, metric scale assertion)
- Blender→Unreal scale test (export a known-size cube, verify 100× conversion)
- OSM extract fetch and manifest writer (records date, bounds, checksum)
- QGIS processing steps as batch scripts, so the geo cleanup is reproducible rather than GUI-only