Wilbur vs rdflib

rdf-to-json.py

import sys
import jsonlines

import rdflib
from rdflib.namespace import RDFS, SKOS, RDF
from rdflib import Namespace
import json


ont = sys.argv[1]
g=rdflib.Graph()
g.parse(ont, format="nt")

DU = Namespace("http://br.ibm.com/document-understanding/")

def get_relations(sent):
    return [ dict(origin    = g.value(rel, DU.origin),
                  subject   = g.value(rel, DU.subject),
                  predicate = g.value(rel, DU.predicate),
                  object    = g.value(rel, DU.object))
             for rel in g.subjects(sent,DU.hasRelation)]


def get_tokens(sent):
    tks = g.subjects(DU.sentence, sent)
    objs = []
    for tk in tks:
        type = g.value(tk, RDF.type)
        lbl  = g.value(tk, RDFS.label)
        if type == DU.Token:
            wn  = g.value(tk, DU.wn30_sense)
            geo = g.value(tk, DU.geoname_sense)
            if wn != None or geo != None:
                obj = dict(txt = lbl,
                           type = type,
                           wn = wn,
                           geo = geo)
                objs = objs + [obj]
        else:
            obj = dict(txt = lbl, type = type)
            objs = objs + [obj]
    return objs


def sentence(sent):
    return dict(text = g.value(sent, RDFS.label),
                status = g.value(sent, DU.complete),
                relations = get_relations(sent),
                tokens = get_tokens(sent))


with jsonlines.open(sys.argv[2], mode='w') as writer:
    for a,b,c in g.triples( (None, RDF.type, DU.Sentence) ):
        writer.write(sentence(a))