Open Legal Data

This is our international landing page www.openlegaldata.io, built with Jekyll (Ruby 4.0, Jekyll 4.4). GitHub Actions builds it on every push to master and publishes _site to the gh-pages branch; pull requests are built but not deployed.

Run locally

bundle install
bundle exec jekyll serve          # http://localhost:4000
bundle exec jekyll build          # static site in _site

Without a local Ruby, use Docker:

docker run --rm -it -p 127.0.0.1:4000:4000 -u $(id -u):$(id -g) \
  -e HOME=/tmp -e BUNDLE_PATH=/site/vendor/bundle -v $PWD:/site -w /site \
  ruby:4.0 sh -c 'bundle install && bundle exec jekyll serve --host 0.0.0.0 --force_polling'

Dependencies are pinned in Gemfile.lock. Upgrade with bundle update and check the build.

Content

• _layouts/home.html: homepage (hero, numbers strip, info boxes). The page frame and colours are the oldp theme in _sass/_base.scss; the content elements are styled in _sass/_landing.scss.
• open-data.md: datasets, REST API and MCP server. research.md: research papers. about.md: project and call for contributors (/team/ redirects here).
• _config.yml → links: platform, API, MCP, Hugging Face, Discord URLs used across the site.
• _data/stats.yml: rounded corpus numbers ("+400,000") for the homepage and data page. Raise them only when the corpus passes the next step; exact numbers go in the release blog post.

Research papers

The research page lists papers citing Open Legal Data from oldp-citations:

• _data/papers.yml is generated from the BibTeX files, do not edit it by hand.
• _data/research_topics.yml assigns every paper to a topic. The llm topic is the highlight of language models trained on Open Legal Data: list a paper there only if its full text says our data is part of the training data or corpus.
• scripts/extra_papers.bib holds papers that use our data without citing our paper. List them under not_citing in research_topics.yml so they are shown but not counted as citations.

After adding citations to oldp-citations (checked out next to this repo):

uv run scripts/build_papers.py --bib-dir ../oldp-citations/bib

The script fails and names the citekeys that are not assigned to a topic; add them to research_topics.yml and run it again.

Analytics

Self-hosted, cookieless Matomo at matomo.openlegaldata.io, configured like the tracker on de.openlegaldata.io (_includes/matomo.html). It is only rendered in production builds (JEKYLL_ENV=production, set in CI) and while matomo_site_id in _config.yml is not empty. The privacy page describes it.

License

• Code: MIT
• Content: CC 4.0 Attribution