How do I collect Swedish legal documents programmatically?

Install juris with `pip install juris`, then run `juris collect-all` to fetch documents from Riksdagen, Regeringen, Domstolsverket, JO/JK, Lagrummet, EUR-Lex, CJEU, and HUDOC into JSON + Markdown files.

Which Swedish legal document types does juris support?

21 document types: propositioner (prop), motioner (mot), betänkanden (bet), riksdagsskrivelser (skr), SOU, Ds, kommittédirektiv (dir), lagrådsremisser (lagr), SFS, NJA, AD, HFD, mark- och miljödomstolen (mod, pmod), JO and JK beslut, myndighetsföreskrifter (foreskrift, AFS, HSLF-FS), EU regulations (eu_reg), EU directives (eu_dir), CJEU rulings (cjeu), and ECHR rulings (echr).

Is the data stored in a machine-readable format?

Yes. Every document is stored both as structured JSON (for machines and pipelines) and as Markdown with YAML frontmatter (human-readable, browsable on GitHub, ideal for LLM context).

Does juris extract text from PDF attachments?

Yes — PDF attachments are downloaded and full-text extracted using pymupdf, with graceful fallback when extraction fails.

juris — A git-native open database for Swedish law

juris is an open-source Python CLI that collects and normalizes Swedish and EU legal documents into a unified, version-controllable database of JSON + Markdown files. It is designed for legal researchers, journalists, civic-tech developers, and AI/LLM pipelines that need a clean, reproducible corpus of Swedish and European law.

GitHub repository · PyPI package · Documentation · CLI manual · llms.txt

Quick start

pip install juris
juris collect-all
juris status

Documents are written to data/{doc_type}/{session}/{id}.{json,md}. JSON for machines, Markdown with YAML frontmatter for humans and LLMs.

Why juris?

One CLI, every Swedish source. Riksdagen, Regeringen, Domstolsverket, JO, JK, Lagrummet, EUR-Lex, CJEU (Curia), and HUDOC (ECHR) — all behind one async, rate-limited Python tool.
Git-native storage. Plain JSON + Markdown means you can git diff changes to law over time and review them in a pull request.
Incremental. State tracking under .state/ resumes where you left off, skipping already-collected documents.
PDF text extraction. Attachments are downloaded and full-text extracted via pymupdf, with graceful fallback.
Agent-friendly. Ships --help-agent, --debug-agent, a commands index, and full man pages so AI coding agents can use juris without prior knowledge.

Document types (21)

Bills (prop), motions (mot), committee reports (bet), parliamentary writs (skr), official inquiries (SOU), ministry memoranda (Ds), committee directives (dir), council-of-legislation referrals (lagr), statutes (SFS), Supreme Court rulings (NJA), Labour Court (AD), Supreme Administrative Court (HFD), land & environment courts (mod, pmod), Parliamentary Ombudsman decisions (JO), Chancellor of Justice decisions (JK), agency regulations (foreskrift, AFS, HSLF-FS), EU regulations (eu_reg), EU directives (eu_dir), CJEU rulings (cjeu), and ECHR rulings (echr).

Data sources (8)

Riksdagen — Swedish Parliament open data (JSON API, data.riksdagen.se)
Regeringen — Swedish Government publications (web scraping, regeringen.se)
Domstolsverket — Swedish courts (REST API, domstol.se)
JO & JK — Parliamentary Ombudsman + Chancellor of Justice (jo.se / jk.se)
Lagrummet — agency regulations and guidelines (lagrummet.se)
EUR-Lex — EU regulations and directives (SPARQL, eur-lex.europa.eu)
Curia — Court of Justice of the EU (SPARQL, curia.europa.eu)
HUDOC — European Court of Human Rights (JSON API, hudoc.echr.coe.int)

Documentation

Overview — what juris does and who it's for
Architecture — modules, data flow, async pipeline
Collectors — auto-discovery framework and adding new sources
Document model — Pydantic schema, enums, validation
Storage format — directory layout, JSON + Markdown shapes
Data sources — every source's API, quirks, and licensing
Parsing rules — text normalization, ID building, attachments

CLI commands

juris — top-level entry point
juris collect — collect from one source
juris collect-type — collect a single document type
juris collect-all — collect every source × type combination
juris status — incremental-collection state report
juris stats — corpus statistics
juris man — built-in man-page viewer

Tech stack

Python 3.11+, async/await, httpx (HTTP), pydantic (validation), beautifulsoup4 + lxml (HTML/XML), click (CLI), pymupdf (PDF text extraction), and pyyaml (Markdown frontmatter). Strict mypy, ruff linting, pytest with asyncio_mode = auto.

License

MIT — see LICENSE. Source on GitHub, package on PyPI.