DBpedia Ontology Extraction

ontology.py

from rdflib import Graph

def ontology(file_name):
    g = Graph()
    index = file_name.rfind('.')
    if index < 0:
        return None
    type = file_name[index + 1:]
    g.parse(file_name, format = type)
    prefix = 'http://dbpedia.org/ontology/'
    return (str(s)[len(prefix):] for s, p, o in g \
            if str(s).startswith(prefix) and len(str(s)) > len(prefix))

def write(src_file_name, trg_file_name):
    g = ontology(src_file_name)
    s = set(g)
    with open(trg_file_name, 'w') as file:
        for o in s:
            file.write(o)
            file.write("\n")

def test():
    # Input files can be downloaded from DBpedia.
    print('Processing nt.')
    write('dbpedia_2015-04.nt', 'tmp1.out')
    print('Processing owl.')
    write('dbpedia_2015-04.xml', 'tmp2.out')