petersunde/danbooru2026-index

Danbooru 2026 Mirror Index

This dataset contains only the derived byte-range lookup artifact used by the danbooru-mirror Space. It contains no image payloads.

• —tar-index.parquet has one row per image with id, shard, offset_bytes, and length_bytes. The Space loads this Parquet file into PostgreSQL with pg_parquet and uses the primary-key range index for lookups.

The repository may also contain metadata-smoke.parquet and tar-index-smoke.parquet. These are three-row fixtures used only by the local Docker smoke test; they are not the production dataset.

The original fixed-width tar-index.bin is kept only as a local development artifact. It is not copied into the Space image, so the deployed Space does not duplicate the range index in its container layer.

The canonical metadata snapshot is loaded separately from nyanko-devs/danbooru2026/metadata/posts-snapshot.parquet at Space startup. This dataset contains no comments, notes, forums, pools, artists, wiki, user records, or image data.

The image bytes remain in nyanko-devs/danbooru2026 and are fetched by exact HTTP Range requests only when requested by the Space. The source dataset license, original image copyrights, and source-site policies continue to apply.