geonames-clean-rdflib.py

import rdflib
# from io import StringIO
# import xml.etree.ElementTree as etree

#builtGraph is where I'm going to accumulate triples that I've scraped
builtGraph=rdflib.Graph()

#addedGraph contains triples that I got from a particular GeoNames RDF file
addedGraph=rdflib.Graph()

with open("all-geonames-rdf.txt", encoding="utf8") as f:
    fileIndex=0
    while fileIndex<1000:
        if fileIndex/10 == int(fileIndex/10):
            print(fileIndex)
        junk=f.readline()
        rdfXmlText=f.readline()
        out = open('output.txt', 'w', encoding='utf-8')
        out.write(rdfXmlText)
        out.close()
        
        result=addedGraph.parse('output.txt')
        builtGraph = builtGraph + addedGraph
        fileIndex=fileIndex+1
    s = builtGraph.serialize(destination='geonames1000.n3', format='n3')