Datasets

The canonical records every page on this site is rendered from, shown as they are written in data/*.json: one file and one record at a time, as a graph, a tree or the raw text. Read-only. Follow a reference to move from a record to the record it names.

Loading the explorer. Until it arrives — or without JavaScript — these are the 10 files it reads, each of which opens as it is written:

How to read this view

Nesting is not a relationship. In the Structure view a solid line joins a card to an object or list written inside the same record. That is the shape of the file, and nothing more: it does not say that one legal instrument contains, amends or depends on another.

A reference is an id, and only an id. Records point at each other by holding another record’s id as a string. A value is shown as a reference only when it is exactly the id of a record in one of these files, and the field that holds it is named. The file records that the id is there; what the connection means is stated, where it is stated at all, in the records themselves — for instance in the summary of an interaction between instruments.

Derived facts are not in the files. A claim’s evidence grade, an enforcement record’s pipeline, an instrument’s competent authority and its key dates are computed when a page loads and are never stored. This view shows the files, so it shows none of them — except a claim’s grade, which is computed here by the same function the rest of the site uses and labelled as computed.

What a file does not say is not shown as a value. A field written as null is shown as null; a field a record does not have is not drawn. Neither is a finding that something does not exist. Each file’s own notes, shown beside its records, state what its fields mean.

Reuse. The datasets are licensed under CC BY 4.0, as set out in Licence and reuse. Every file can be opened and downloaded as it is.