1951 to today
From the founding treaties onward. Regulations, directives, decisions, Court of Justice judgments, consolidated texts, and the national measures that transpose them.
EU legal corpus · article level
Every EU act,
broken down to the article.
EUR-Lex publishes the law. It does not publish it in a shape you can work with. We ingest the whole corpus from the Publications Office, parse it down to the article and the paragraph, resolve every act to its canonical identifier, and map how the acts point at one another.
Search by title, or jump straight to an identifier:
32016R0679, 02016R0679-20160504, 62013CJ0067
- 859 411 acts
- 4 languages
- 2 390 267 paragraphs parsed
- 3 496 276 relations between acts
What the corpus contains
Described as it is, not as it is usually advertised
24 languages
Each act is a distinct expression per language. A work need not exist in all of them, and where it does not, we say so rather than quietly serving you English instead.
Three tiers of structure
Post-2008 sources carry real article markup and parse cleanly. Pre-2008 HTML carries none, so it yields text only. Some major acts exist as PDF alone. Every document states which tier it belongs to.
Consolidations are documents
A consolidated version is not an edit of an act. It is a separate work with its own identifier and its own date, which is why point-in-time reading works here at all.
What we add
The text is public. The structure is the work.
Article-level parsing
Formex and XHTML manifestations are reduced to articles and numbered
paragraphs, with the printed numbering preserved exactly as printed.
4a stays 4a, never coerced into an integer.
Canonical identifiers
Every document is addressed by its CELEX, including the awkward ones: corrigenda carrying parentheses, treaty protocols carrying slashes. No internal identifier is ever exposed.
The relation graph
Sixty-plus relation types, grouped into families that read as sentences: what an act amends, repeals, cites and is based on, and in the other direction, what amends and interprets it.
Point-in-time versions
The full consolidation chain of an act, reachable from any link in it, so "what did this say in 2019" takes one click rather than an afternoon.
Start here
Four ways into the corpus
Honest limits
What this is not
Only the Official Journal is authentic
What you read here is a structured reproduction of text published by the Publications Office of the European Union. For anything with legal consequence, cite the Journal.
The backfill is continuous
A document that exists on EUR-Lex may not be here yet, and a relation may point at an act we have not ingested. Where that happens the explorer says so instead of hiding the gap.
Structure depends on the source
Article-level breakdown exists only where the manifestation carries article markup. Older acts yield continuous text, and a few exist as PDF alone. Each document states which case it is.
Need this programmatically?
This explorer is built to be read. Bulk and machine access run through an authenticated JSON API, granted per integration rather than opened to the public. The source data is free; the structuring is ours to steward.