This dataset compiles official Portuguese legislation text. The underlying content is public and copyright-exempt (see License below); access is gated only so we can track who's using it and for what — briefly tell us your intended use.

leis-pt — consolidated Portuguese legislation corpus

One record per consolidated Portuguese legal act — the law as currently in force, with amendments applied — sourced from Diário da República's official consolidated-legislation records. Compiled by leis-pt, a separate, private project (not published), and released here as a training-ready corpus by teex-pt/pt-amalia.

This is the base corpus for two derived training sets — see teex-pt/amalia-sum-dre (summarization) and teex-pt/amalia-cita-legal (grounded citation / RAG training) — both built directly from this corpus.

Scope

1,727 unique diplomas — 26 base legal codes (Código Civil, Código Penal, Código do Trabalho, ...) plus 1,701 other consolidated acts (decree-laws, portarias, leis) reachable from Diário da República's by-code and by-theme consolidated-legislation indexes.
Consolidated / current-in-force text only — not the full historical Diário da República stream. See Known limitations.
96.2M characters total, median 25.8K chars/diploma, up to 1.38M chars for the largest codes.
Snapshot date: most recently updated through 2026-07.

Files

corpus.jsonl — one record per diploma:
- identity/metadata: diploma_id, diploma_legis_id, tipo, numero, designacao (official descriptive title), titulo, sumario (official DR abstract), emissor, data_publicacao, data_ultima_consolidada, eli (European Legislation Identifier), fonte_url, url_pdf, consolidacao_estado, is_codigo_base, temas (theme tags from DR's own taxonomy)
- texto_integral — the full consolidated text, linearized from the fragment tree (article/chapter headers + body text, document order)
- fragmentos — the raw fragment tree (id, pai_id, ordem, nome, epigrafe, texto, data_entrada_vigor) for consumers that want structure (article-level chunking, amendment timestamps) rather than flat text
- n_fragmentos, n_chars

Known limitations

Consolidated law only, not the full historical stream. Acts that only amend another act without themselves having a consolidated version aren't included as their own record — though the amendments they made are still reflected in amalia-cita-legal.
No historical "atos" PDF stream text extraction yet — a real, currently-open gap upstream, not something this snapshot works around.
Three diplomas (of 1,727) have no theme tag.

License & attribution

Official Portuguese legal texts are exempt from copyright under Portuguese law (Código do Direito de Autor e dos Direitos Conexos, art.º 8.º — official acts of a public nature are excluded from copyright protection). This compilation — the crawl, structuring, and deduplication — is released under CC0 1.0 (public domain dedication). Please retain the fonte_url/eli provenance fields if you redistribute derived data, so downstream users can trace any record back to its official publication.

Citation

@misc{leis-pt-consolidada-2026,
  title = {leis-pt: consolidated Portuguese legislation corpus},
  author = {teex-pt},
  year = {2026},
  howpublished = {\url{https://huggingface.co/datasets/teex-pt/leis-pt-consolidada}},
  note = {Source: Diário da República, legislação consolidada (diariodarepublica.pt)}
}

Downloads last month: 14

Total file size:

216 MB