This library aims to provide a bidirectional bridge between 2 worlds: RDF/Linked Data and EDN/Clojure.
-
Mapping between keyword namespaces and RDF prefixes
In RDF, the IRI is the identity. In EDN, you can use simple or qualified keywords as their internal representations, but only for those RDF prefixes where you have reason to do that. In other words, IRI prefixes are external, but namespaces for keywords are internal. They are not interfering with each other.
In practice, we are not going to convert all the IRIs into keywords, and their RDF prefixes into Clojure namespaces, and vice versa. Instead:
-
we need only some prefixes (probably, very few of them) to map to the namespaces for qualified keywords.
-
prefixes should not determine how namespaces look like and vice versa.
-
moreover, it is not necessary that an input Turtle document will have prefixes at all. But we still can use qualified keywords by "extracting" part of IRI and creating a namespace for that part.
-
-
Handling terse format for collections
A well-constructed RDF/Turtle document implies to reduce triples into trees with nested predicate lists and object lists. But when we work with data, we do not need that. All we need are simple triples: subject, predicate and object. So the parser flattens these trees. On the other hand, the encoder builds the trees for us.
-
Handling trees of Blank Node Property Lists
RDF/Turtle allows Blank Node Property Lists to create nested data structures. The parser transforms everything into a flat list of triples: subject, predicate and object, creating temporary identifiers for anonymous entities, so we can handle them as datalog-like statements. On the other hand, the encoder builds such terse trees from our data.
-
Isomorphic and bidirectional
The same syntax and semantics both on frontend and backend, both for parsing input documents and producing output documents.
This work is in progress.
Currently, the development is concentrated around Turtle standard: https://www.w3.org/TR/turtle and it is already quite accurate.