spoi-cleanup/exploreData.py

#from csv import reader
#import json
from pprint import pprint
from SPARQLWrapper import SPARQLWrapper, JSON
#import ssl
#import sys
import time
#import urllib.request, urllib.parse, urllib.error
import rdflib
from rdflib.plugins import sparql

def getPOIsWithEqualNameAndTypeWithinDistance(distance):
    fc_sparql = SPARQLWrapper("https://www.foodie-cloud.org/sparql")
    q = """
        PREFIX unit: <http://www.opengis.net/def/uom/OGC/1.0/>
        SELECT DISTINCT ?Resource1 ?Resource2 ?distance
        FROM <http://www.sdi4apps.eu/poi/czech>
        WHERE {{
            ?Resource1 a ?type1 .
            ?Resource1 rdfs:label ?label1 .
            ?Resource1 ogcgs:asWKT ?geo1 .
            ?Resource2 a ?type2 .
            ?Resource2 rdfs:label ?label2 .
            ?Resource2 ogcgs:asWKT ?geo2 .
            FILTER (lcase(str(?label1)) = lcase(str(?label2) )) .
            FILTER ( ?Resource1 != ?Resource2 && ?Resource1 < ?Resource2)
            FILTER (NOT EXISTS {{?x a ?Resource1}} && NOT EXISTS {{?y a ?Resource2}}) .
            BIND (ogcgsf:distance ( ?geo1 , ?geo2 , unit:meter ) as ?distance) .
            FILTER ( ?distance < {0})
        }}
        LIMIT 100
        OFFSET 0
        """.format(distance)
    print(q)
    fc_sparql.setQuery(q)
    fc_sparql.setReturnFormat(JSON)
    results = fc_sparql.query().convert()
    #for result in results["results"]["bindings"]:
        #print(result)
    return results["results"]["bindings"]

#g = rdflib.Graph()
print("= LOOKING FOR DUPLICATE POIS =")
distance = input("Distance to search within: ")
print("Searching for duplicate POIs within {0} m ...".format(distance))
t_1 = time.time()
result = getPOIsWithEqualNameAndTypeWithinDistance(distance)
print("Finished in {}".format(time.time()-t_1))
pprint(result)
#g.parse("ontOSM&qids.owl", format="turtle")
#print("Input RDF graph read with {} statements!\n".format(len(g)))

#with open("ontOSM_beta.owl", 'w', encoding="utf-8") as fh:
#    fh.write(g.serialize(format="turtle").decode())
#print("File \"ontOSM_beta.owl\" succesfully created")