tgbugs / rkdf

RDF parsers implemented as Racket languages

rkdf README

rkdf

RDF parsers implemented as Racket languages

Warning

This code is extremely broken.

Install

Requirements

#lang rdf/turtle and #lang rdf/nt use brag as a parser generator.
ttl-to-rkt depends on librdf and rapper for fast parsing and conversion.

Scripts

Run these or the equivalent for your setup if you want to use ttl-to-rkt or rkdf-convert-all.

ln -s $(realpath ./bin/ttl-to-rkt) ~/bin/ttl-to-rkt
ln -s $(realpath ./bin/rkdf-convert-all) ~/bin/rkdf-convert-all

Development

raco pkg install rkdf-lib/ rkdf/

Usage

As a #lang

Conveniently turtle comments allow us to include #lang rdf/turtle in the first line of a turtle file. This is also true for nt, which will accept the #lang rdf/nt line. #lang rdf defaults to turtle syntax.

Converting a single file

To convert a single file run the following

ttl-to-rkt my-ontology-file.ttl my-ontology-file.rkt

ttl-to-rkt runs raco make automatically so if you are converting a large file it may take some time.

rapper supports many formats besides turtle, an they could be included with minimal changes to ttl-to-rkt.

Converting files in git

The fastest way to convert RDF representations store in git into Racket is to use rkdf-convert-all. The process is in no way robust and may fail silently. Once the conversion to rkt is complete raco make is used to compile the outputs to speed up loading by a few orders of magnitude. Note that convert-all will exactly duplicate the file hierarchy of the original repository inside the rkt/ folder.

To make files available as racket modules do the following

raco pkg install path/to/my/ontology/
cd path/to/my/ontology
rkdf-convert-all

It should then be possible require a single file using normal Racket require syntax.

#lang racket/base
(require ontology/rkt/my-ontology-file)

Parsing is slow!

Running raco make my-file.ttl will make all future loads stupidly fast. A faster parser would be preferred, but no time to implement that at the moment. The BNF used for the turtle parser was a copy/paste fixup job from the turtle spec and is not well written or optimized in the slightest.

The real solution

https://bitbucket.org/nxg/racket-librdf is probably the correct solution, but is more complicated to install.