endomorphosis/world_court_forms

World Court Forms

Archive of blank court / tribunal forms collected for endomorphosis/world_courts (US state judiciaries, Canada, Australia, England & Wales, ...). The structured rows (title, form number, court, source URL, per-row license and robots policy) live in world_courts under data/forms/<cc>.parquet; this dataset only stores the original files, packed into plain (uncompressed) tar batches of roughly 0.5-1 GB so the repo stays small in file count.

Layout

pdfs/<cc>/batch-NNNN.tar      # cc = us, gb, ca, au, ...; members are "<ST>/<sha256>.<ext>" (us) or "<sha256>.<ext>"
index/forms_index.parquet     # one row per stored file

index/forms_index.parquet columns: sha256 (content hash = file name), cc, member (path inside the tar), pdf_path (path used in the world_courts rows, forms/<cc>/...), bytes, ext, batch (repo path of the tar), tar_sha256, tar_bytes, verified_at. Join a world_courts forms row to its file with pdf_path (or pdf_sha256); rows also carry hf_tar / hf_member once offloaded.

Fetch one file

import tarfile, pandas as pd
from huggingface_hub import hf_hub_download
idx = pd.read_parquet(hf_hub_download("endomorphosis/world_court_forms", "index/forms_index.parquet", repo_type="dataset"))
r = idx[idx.sha256 == "<sha256>"].iloc[0]
tar = hf_hub_download("endomorphosis/world_court_forms", r.batch, repo_type="dataset")
data = tarfile.open(tar).extractfile(r.member).read()

Licensing / provenance

Forms are published by courts and governments. Terms differ by source (e.g. US court forms are generally public documents, UK forms are OGL / Crown copyright, Canadian federal forms carry a non-commercial government licence). The authoritative per-form license, source_url and site_policy are in the world_courts rows - check them before reuse.

Producer code: scrapers/courts/forms_offload.py in endomorphosis/world_law_pipeline.