Dataset Viewer
Auto-converted to Parquet Duplicate
The dataset viewer is not available for this split.
Cannot load the dataset split (in streaming mode) to extract the first rows.
Error code:   StreamingRowsError
Exception:    TypeError
Message:      Couldn't cast array of type large_string to null
Traceback:    Traceback (most recent call last):
                File "/src/services/worker/src/worker/utils.py", line 149, in get_rows_or_raise
                  return get_rows(
                      dataset=dataset,
                  ...<4 lines>...
                      column_names=column_names,
                  )
                File "/src/libs/libcommon/src/libcommon/utils.py", line 272, in decorator
                  return func(*args, **kwargs)
                File "/src/services/worker/src/worker/utils.py", line 129, in get_rows
                  rows_plus_one = list(itertools.islice(safe_iter(ds, dataset=dataset), rows_max_number + 1))
                File "/src/services/worker/src/worker/utils.py", line 544, in safe_iter
                  yield from ds.decode(False) if ds.features else ds
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 3052, in __iter__
                  for key, example in ex_iterable:
                                      ^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2559, in __iter__
                  for key, pa_table in self._iter_arrow():
                                       ~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2584, in _iter_arrow
                  for key, pa_table in self.ex_iterable._iter_arrow():
                                       ~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 558, in _iter_arrow
                  for key, pa_table in iterator:
                                       ^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 440, in _iter_arrow
                  for key, pa_table in self.generate_tables_fn(**gen_kwags):
                                       ~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/parquet/parquet.py", line 220, in _generate_tables
                  yield Key(file_idx, batch_idx), self._cast_table(pa_table)
                                                  ~~~~~~~~~~~~~~~~^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/parquet/parquet.py", line 156, in _cast_table
                  pa_table = table_cast(pa_table, self.info.features.arrow_schema)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2381, in table_cast
                  return cast_table_to_schema(table, schema)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2315, in cast_table_to_schema
                  cast_array_to_feature(
                  ~~~~~~~~~~~~~~~~~~~~~^
                      table[name] if name in table_column_names else pa.array([None] * len(table), type=schema.field(name).type),
                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                      feature,
                      ^^^^^^^^
                  )
                  ^
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 1861, in wrapper
                  return pa.chunked_array([func(chunk, *args, **kwargs) for chunk in array.chunks])
                                           ~~~~^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2155, in cast_array_to_feature
                  return array_cast(
                      array,
                  ...<2 lines>...
                      allow_decimal_to_str=allow_decimal_to_str,
                  )
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 1863, in wrapper
                  return func(array, *args, **kwargs)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2014, in array_cast
                  raise TypeError(f"Couldn't cast array of type {_short_str(array.type)} to {_short_str(pa_type)}")
              TypeError: Couldn't cast array of type large_string to null

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

world_courts

Courthouses, judges, court rules and court forms for every political subdivision, starting with the United States and now covering the United Kingdom (GB), Canada (CA), Australia (AU) and the 27 EU member states plus the EU level. Built by the world_law_pipeline (scrapers/courts/ in endomorphosis/world_law_pipeline).

Configs

  • courthouses: one row per court location. Columns: court_id (osm:{type}/{id} or usdir:{hash}), name, official_name, level (keyword guess: supreme/appellate/trial_general/trial_limited/municipal/probate/family/juvenile/federal_district/other), court_system, country_code, subdivision (ISO 3166-2), county_fips, street/city/postcode, lat/lon, place_ids (ids from the global_municipalities place list: gnis: for US, wd:Q… elsewhere; includes the county id), place_match_method (census_geo / name), source_url, source_kind, site_policy, sources_json (every contributing source with its own policy), needs_review, retrieved_at.
  • judges: public official roster info only (name, official title, court, source page). needs_review = true: the data is machine-extracted and not yet validated.
  • forms: official statewide court forms with pdf_path, pdf_sha256 and source_url.
  • rules: one row per court rule (statewide rules of procedure and local court rules) with full text, rule_set, rule_number, effective_date, source_url, site_policy, source_sha256, and external_ref (the ipfs_cid / identifier of the same rule in justicedao/ipfs_court_rules, where new rules are contributed append-only; see schema/RULES_LINKAGE_PLAN.md).
  • rules_court_crosswalk: which courthouses each rule set governs (rule_set_id -> court_id) with link_basis and link_confidence (high = name match or supreme court, medium = county + court class, low = class only). Links are inferred from the courthouse list, whose court levels are keyword guesses, so treat as approximate.

Countries (per-country parquet files: data/<config>/<cc>.parquet, cc = us | gb | ca | au; EU files below)

  • GB (England & Wales, Scotland, Northern Ireland): courthouses = GOV.UK Find a Court or Tribunal (imported with its court_ids and OGL licence from endomorphosis/ipfs_uk_courts), Scottish Courts and Tribunals Service "find us" pages, Courts and Tribunals NI court offices, UK Supreme Court, merged and de-duplicated (<=150 m) with OSM and Wikidata. Judges: E&W judiciary.uk lists (imported from ipfs_uk_courts, not re-scraped) + UK Supreme Court. Rules: Criminal Procedure Rules 2025, Rules of the Court of Session 1994, Act of Adjournal (Criminal Procedure Rules) 1996, NI Rules of the Court of Judicature 1980, NI Magistrates' Courts Rules 1984, Employment Tribunal Procedure Rules 2024, FTT (SEC) and Upper Tribunal Rules 2008 from legislation.gov.uk CLML XML (Crawl-delay 5 s respected). The CPR/FPR/Practice Directions already published in ipfs_uk_courts are linked, not duplicated, through rules_court_crosswalk rows with linked_rows = "upstream".
  • CA: Ontario MAG courthouse CSV (with coordinates), BC Court Locations (DataBC ArcGIS), Quebec Ministere de la Justice palais de justice / points de service pages, Saskatchewan, Manitoba and Nova Scotia court sites, Ottawa federal courts, plus OSM and Wikidata. Judges: Supreme Court of Canada roster + Federal Court (imported from ipfs_canada_courts, licence gc-non-commercial). Rules: SCC Rules (SOR/2002-156), Ontario Rules of Civil Procedure, BC Supreme Court Civil Rules, Alberta Rules of Court; the Federal Courts Rules already in ipfs_canada_courts are linked via crosswalk. Forms: Federal Court (via Wayback; live site blocked from our network) and Ontario court forms.
  • AU: Geoscience Australia Judicial Courts (federal registries, via archive), Queensland Courts directory, plus OSM and Wikidata. Judges: High Court of Australia. Rules: Qld UCPR 1999 and NSW UCPR 2005. Forms: Federal Circuit and Family Court (via Wayback) and Supreme Court of Victoria.
  • Hierarchy: record_type = court_body rows (no address/geo) describe each court system node (e.g. gb-ew-crown-court, ca-on-superior-court, au-nsw-supreme-court) and form the parent_court_id chain up to the national apex court; record_type = courthouse rows are physical locations whose parent_court_id points to the body that sits there (a location may host several courts, see record_note). place_ids are wd:Q... ids from global_municipalities matched by locality name (and, for rows without a locality, by Nominatim reverse geocode: place_match_method = reverse_geocode_name); for CA/AU the province/state id is appended as a second id. Extra columns: license, geo_source (postcodes.io centroid / nominatim / official), upstream, subdivision_method.
  • Geocoding: postcodes.io (ONS/OS open data) for UK postcodes, Nominatim (<=1 req/s, cached) for CA/AU address-only records. Court forms PDFs are stored on the pipeline box under forms/<cc>/<sha256>.<ext> and referenced by pdf_path / pdf_sha256 in the forms config.
  • site_policy additions: archive_wayback(live_tls_blocked_from_box), robots_disallowed_archive_used, robots_unknown:fetched_via_webfetch, archive:commoncrawl (imported rows).

EU (phase 6): 27 member states + EU level (cc = at be bg hr cy cz dk ee fi fr de gr hu ie it lv lt lu mt nl pl pt ro sk si es se | eu)

Per-country counts: EU_COUNTS.md. Extra courthouse columns for these files: court_family, tier, level_basis, cdb_id, ecli_code, name_local, geoname_ids, accepted_languages, upstream_json, upstream_court_id (see schema/world_courts_schema.json v0.3).

  • Courthouses: base list = the e-Justice European Court Database open API (webgate.ec.europa.eu/cdb/search, documented on data.europa.eu; bailiffs, notaries, prosecutors, ministries, bar associations and test entries are filtered out), geocoded by (1) matching OSM amenity=courthouse (ODbL) / Wikidata courthouses (CC0) within the CDB municipality (geo_source = osm|wikidata, building precision), else (2) the GeoNames (CC-BY 4.0) centroid of the court's municipality or postcode (geonames_*, municipality-level precision only). OSM/Wikidata courts not in CDB are added after de-duplication (<=330 m). Ireland, Netherlands and Germany are merged with the existing endomorphosis/ipfs_{ireland,netherlands,germany}_courts rows (their court_id and licence kept, provenance in upstream_json; not re-scraped).
  • Hierarchy: court_body rows {cc}-{family}-{tier} (families ordinary/administrative/labour/social/fiscal/commercial/constitutional; tiers limited/general/appellate/supreme) chain upward through parent_court_id; courthouses point at their family/tier body. Family/tier come from multilingual name keywords (level_basis). Geographic appeal-district links (e.g. Amtsgericht -> Landgericht -> OLG) are NOT yet modelled except in the imported IE rows. eu-cjeu -> Court of Justice / General Court (Luxembourg).
  • place_ids: wd:Q ids from global_municipalities via Wikidata P1566 (GeoNames id) of the CDB court municipality (geonames_p1566), the enclosing municipality through P131 (geonames_p1566+p131), else normalised-name matching (name).
  • Judges: Court of Justice and General Court members (curia.europa.eu official pages) and Ireland's 198 judges (imported from ipfs_ireland_courts). needs_review = true.
  • Forms: EU standard forms/annex forms of the main civil-procedure instruments (EUR-Lex PDFs read from the Internet Archive because EUR-Lex answers non-browser clients with a WAF challenge and e-justice.europa.eu returns 403) and 267 rechtspraak.nl forms/regulations (NL). PDFs live under forms/<cc>/<sha256>.<ext>.
  • Rules: DE (ZPO, StPO, GVG, VwGO, ArbGG, SGG, FGO, FamFG, BVerfGG, JGG, OWiG, RPflG, InsO; gesetze-im-internet.de), ES (LEC, LECrim, LRJCA, LRJS, LOPJ, LOTC; BOE open API), PL (KPC, KPK, KPA, PPSA, KPW; Sejm ELI API consolidated texts), NL (Rv, Sv, Awb, Fw; BWB repository), FR (CPC, CPP, CJA, COJ, CPCE from the DILA LEGI open-data dump snapshot 2025-07-13: later amendments not applied), EU (CJEU and General Court Rules of Procedure as originally published, plus 17 procedural regulations/directives). rules_court_crosswalk maps each rule set to the court_body nodes it governs. Legislation text of other codes stays in endomorphosis/ipfs_<country>_laws / ipfs_eu_laws; the laws_dataset_links config lists them.
  • site_policy additions: open_api_documented, upstream_dataset, open_data_dump, derived, robots_unknown:api_documented.

Rest of world, chunk 1 (phase 7): NZ, IN, JP, BR, MX, ZA (cc = nz in jp br mx za)

Same schema as US/EU. Courthouse files add: geonames_city, geonames_admin1, geonames_admin2, upstream, facility_kind, hours, and for ZA court_type_official, also_known_as, places_of_sitting, services, municipality, mag_district, small_claims, equality_court (court-manager names, cell numbers and e-mails in the source list are deliberately NOT stored).

  • Courthouses: NZ = upstream endomorphosis/ipfs_newzealand_courts (imported by court_id) + OSM + Wikidata; IN = OSM + Wikidata (no machine-readable official courthouse finder: eCourts is captcha-gated); JP = courts.go.jp official location pages (/<region>/about/syozai, 385 locations; coordinates by OSM match or Nominatim geocode, tagged in geo_source) + OSM + Wikidata; BR = upstream CNJ structure (ipfs_brazil_courts, as court_body rows) + OSM fóruns + Wikidata (non-court OSM objects such as Ministério Público/Defensoria/cartórios dropped); MX = OSM + Wikidata (no official open directory located); ZA = Department of Justice lower-courts list (510 operational courts with GPS and addresses) + OSM + Wikidata. Dedup: <=150 m (JP/BR/MX 120 m) staged OSM then Wikidata, then by court_id for upstream rows.
  • Hierarchy: court_body rows {cc}-... (NZ 16, IN 51 = SC + 25 HC + 25 district-court bodies, JP 38 = SC + 8 circuits x (HC, DC, FC, SC) + IP High Court + tier aliases, BR 119 = CNJ list + STF, MX 41 = SCJN, TEPJF, colegiados, unitarios, juzgados de distrito, agrarios, TFJA + 32 state TSJs, ZA 28 = CC, SCA, LAC, LC, LCC, CAC, EC, HC divisions by province, magistrates' bodies by province). parent_court_id chains lower courts to the body, bodies to the apex court.
  • place_ids: wd:Q... ids from global_municipalities by Wikidata P131, GeoNames cities500 nearest city (<=25 km) -> admin2/city name, locality name, or name in court name (place_match_method); prefecture/state place appended where known.
  • Judges (needs_review = true): NZ Supreme Court + Court of Appeal (courtsofnz.govt.nz; the Supreme Court page is behind a CAPTCHA so a Wayback snapshot is used), IN Supreme Court (sci.gov.in), JP Supreme Court (courts.go.jp), BR STF (portal.stf.jus.br, 10 of 11 seats filled), MX SCJN (scjn.gob.mx, 9 ministers), ZA Constitutional Court (concourt.org.za via Wayback snapshot 2026-10-15).
  • Rules: NZ HCR 2016, DCR 2014, SCR 2004, CoA Civil 2005, Criminal Procedure Rules 2012, Family Court Rules 2002 (legislation.govt.nz via Wayback); IN Supreme Court Rules 2013 (OCR, heuristic), CPC 1908, BNSS 2023 (heuristic splits, Wayback PDFs); JP statutes via the open e-Gov Laws API v2 + Supreme Court rules from courts.go.jp PDFs; BR CPC, CPP, Lei 9.099/1995 (planalto.gov.br via Wayback) and the STF Regimento Interno; MX CNPCF, CNPP, Ley de Amparo (diputados.gob.mx via Wayback); ZA Uniform Rules, Magistrates' Courts Rules, SCA, Labour Court and Labour Appeal Court rules (justice.gov.za). One row per rule/article in the original language (language).
  • Forms: official forms only; leaflets/guides/checklists/FAQs were filtered into <cc>_rejected_brochures.jsonl (not published as forms). MX has no standardized official court forms (amparo has no official template) so mx has none.
  • site_policy additions: archive_wayback(live_*_unreachable), state_body_secondary (BR TJSP forms).

Rest of world, chunk 2 (phase 7): KR, TW, PH, AR, CO, CL, NG, KE (cc = kr tw ph ar co cl ng ke)

Same schema and conventions as chunk 1. No upstream ipfs_*_courts dataset exists for these countries, so nothing is imported by court_id; courthouses are OSM (ODbL) + Wikidata (CC0, none had coordinates except AR 1) deduplicated at 120 m, plus compiled court_body hierarchy rows. place_ids from GeoNames nearest city / P131 / name matching; the GeoNames admin1 code is in subdivision (CC-<admin1>).

  • Courthouses / bodies: KR 150 + 8, TW 83 + 9, PH 342 + 7, AR 641 + 9, CO 202 + 8, CL 680 + 10, NG 64 + 9, KE 64 + 8. Rows that are not courts (prosecutors, police stations, registries, notaries, prisons) are dropped by name filter; Chilean Juzgados de Policía Local are kept (they are courts). Unclassifiable tiers have needs_review = true. No official courthouse finder was reachable/machine-readable for any of the eight (see blockers in the run report), so these are OSM/Wikidata only.
  • Judges (needs_review = true, top court): TW Constitutional Court 8 (partial), PH SC 15, KR SC 13 (partial), AR CSJN 3 (official page), CO Constitutional Court 9, CL Corte Suprema 18 (partial), NG SC 21, KE SC 6 (official page). Where the official host is unreachable the name list comes from official-site excerpts returned by web search and site_policy/extraction say so.
  • Rules (one row per article/section, original language): TW 3,421 articles in 15 statute sets (MOJ open-data dump), PH 1,087 sections (2019 Amendments, 1997 Rules of Civil Procedure, 2000 Criminal Procedure; lawphil.net secondary copy), AR 1,307 (CPCCN, CPPN; InfoLEG), CO 1,084 (CGP Ley 1564/2012 - 538 of ~626 articles because Wayback lacks one page, CPP Ley 906/2004; Senate basedoc via Wayback), CL 2,170 (CPC, CPP, COT; BCN LeyChile XML), NG 491 (ACJA 2015; text layer of a secondary copy because the official PDF is an image scan), KE 277 (Civil Procedure Act Cap. 21 complete; Criminal Procedure Code Cap. 75 truncated by Wayback, flagged in record_note). KR has none (law.go.kr unreachable).
  • Forms: AR 1 (CSJN demanda originaria), CO 6 (Rama Judicial formatos 1-6), KE 2 (Small Claims, probate registry); guides/feedback/HR forms moved to <cc>_rejected_brochures.jsonl (CO 5, KE 4). No reachable official forms index for KR, TW, PH, CL, NG.
  • site_policy additions: robots_unknown:official_site_unreachable_from_box|search_excerpt, robots_unknown:official_site_behind_captcha|search_excerpt (CL), secondary:lawphil.net ... (PH).

Rest of world, chunk 3 (phase 7, in progress): SG MY IL PK BD TH VN ID TR EG SA AE PE VE EC GH ET MA RU DZ UA KZ IR CN (cc = lowercase)

Same schema as chunks 1-2. Courthouses = OSM amenity=courthouse (ODbL; CN queried per province, ZJ added separately; SG has 3 OSM building-tagged court buildings because no amenity=courthouse objects exist) + Wikidata (CC0; only the direct-class query P31 in {courthouse, court} with coordinates worked while WDQS was degraded) + compiled court_body hierarchy; non-court offices (prosecutors, police, prisons, notaries, registries, Iranian dadsara) are dropped by a multilingual name filter. place_ids use GeoNames nearest city / P131 / name; many countries have low place coverage because global_municipalities has few or non-Latin-named places there (see place_match_method).

  • Courthouse / body counts: SG 3+4, MY 161+6, IL 40+5, PK 111+4, BD 67+2, TH 146+5, VN 213+4, ID 258+7, TR 279+9, EG 51+5, SA 70+5, AE 41+4, PE 227+5, VE 35+3, EC 87+4, GH 66+5, ET 43+4, MA 83+5, RU 2,816+6, DZ 309+4, UA 593+4, KZ 163+4, IR 713+4, CN 1,540+5.
  • Judges (needs_review, from official-site excerpts via web search): SG 10, MY 12 (Federal Court), IL 11, PK 17, GH 21.
  • Rules (one row per article/rule, original language): TR 1,014 (HMK 6100, CMK 5271, IYUK 2577, Danıştay Kanunu 2575; mevzuat.gov.tr), EC 444 (COGEP), MY 1,041 (Rules of Court 2012, per order/rule).
  • Fix applied to all earlier countries: place matching of non-Latin names previously collapsed to an empty key and attached a bogus municipality (JP 79, RU/TW/SA/EG/IL rows); rcore now keeps Unicode names; TW, KR, IL, EG, SA, AE, RU, TH, MA, ET and JP were rebuilt.
  • Not yet done (chunk 3): judges for most countries and forms for all chunk-3 countries.

Rest of world, chunk 4 (phase 7): IQ TZ UG SN CI CM JO LB TN LK NP MM (cc = lowercase)

  • Same layout and schema; courthouses = OSM amenity=courthouse (ODbL) + Wikidata court instances, four-tier body hierarchy (<cc>-supreme-court ... <cc>-lower-court), place_ids via Wikidata P131 / GeoNames / name matching. Generic Supreme Court body ids are used for the apex court even where it is a Court of Appeal (TZ), a Cour de cassation (CI) or a Federal Supreme Court (IQ); this is stated in the judge extraction note.
  • Noise filtering: OSM rows that are municipalities, cafés, ministries, personal names, shipping companies (MM) and prosecutors' offices are dropped (NOT_COURT, plus a positive court-token filter for MM). Remaining unnamed Courthouse rows carry needs_review.
  • Judges (top-court rosters, needs_review=true): UG 11 (judiciary.go.ug), TZ 37 (Judiciary of Tanzania report annex), IQ 8 (Federal Supreme Court), CM 75 (coursupreme.cm member page, includes parquet rows with position_type=prosecutor), CI 1 (Cour de cassation president), LK 13 (secondary roster cross-checked with official bench lists), NP 16 (supremecourt.gov.np).
  • Rules and forms: none. TanzLII/ULII (content signals ai-train=no, ai-input=no) and justice.lk (unofficial consolidation) were not used; supremecourt.lk is behind a Cloudflare challenge; lawnet.gov.lk, cassation.tn, legislation.tn, jo.gouv.sn and parliament.go.ug were unreachable from the box; no downloadable official text for Nepal's Civil Procedure Code was found.
  • Place-matching update: generic affix stripping (governorate/province/region/district/municipality) and Arabic prefixes (محافظة، بلدية، لواء، قضاء...) for JO LB TN IQ TZ LK NP MM UG SN CI CM raised IQ place links from 199 to 279 and TZ from 37 to 106.

Rest of world, chunk 5 (phase 7 completion): every remaining country in global_municipalities

  • 137 country/territory codes (KH BO PY UY CR PA GT DO ZM ZW UZ RS CH ... through the small territories) now have data/<cc>/<cc>_courthouses.jsonl with four generic court bodies and courthouse rows (about 3,200 courthouses). See WORLD_COUNTS.md for per-country coverage and gaps.
  • Sources: OSM amenity=courthouse (ODbL) via batched Overpass, Wikidata court items with coordinates (CC0), and for ten territories with neither (CK AI NR SM TC TV NU IO KY MS) official court contact pages plus Nominatim positions (needs_review=true).
  • No judges, rules or forms were collected for these countries.

Sources and licensing

  • Base courthouse locations come from OpenStreetMap amenity=courthouse (© OpenStreetMap contributors, ODbL 1.0), so this dataset is released under ODbL. Rows from OSM have site_policy = open_license:ODbL.
  • Official state judiciary directory pages are merged in and deduplicated by distance (≤150 m, or ≤600 m with a matching name).
  • site_policy values: robots_allowed, robots_absent_allowed (robots.txt returned 4xx, so allowed per RFC 9309), robots_unreachable, robots_disallowed|archive_wayback (the page was read from an Internet Archive snapshot instead of the live site), …+live_blocked|archive_wayback, and terms_restricted.
  • Geocoding and place matching use the US Census Geocoder (public domain).
  • No captcha or Cloudflare bypass, proxies, or extracted API keys were used.

Schema: schema/world_courts_schema.json.

Downloads last month
1,665