src/validate.py

from rdflib import Graph, Namespace
import pandas as pd

def run_validation():
    g = Graph()
    g.parse("cars_knowledge_graph.ttl", format="turtle")
    
    EX = Namespace("http://example.org/cars/")
    
    queries = {
        "1. List all cars manufactured by Ferrari": """
            PREFIX ex: <http://example.org/cars/>
            PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
            
            SELECT ?car_name
            WHERE {
                ?car ex:hasManufacturer ?manu .
                ?manu rdfs:label ?manu_name .
                FILTER (REGEX(?manu_name, "Ferrari", "i"))
                ?car rdfs:label ?car_name .
            }
        """,
        "2. Cars with HorsePower > 800": """
            PREFIX ex: <http://example.org/cars/>
            PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
            
            SELECT ?car_name ?hp
            WHERE {
                ?car ex:hasHorsePowerValue ?hp .
                ?car rdfs:label ?car_name .
                FILTER (?hp > 800)
            }
        """,
        "3. Count of 2-Seater cars": """
            PREFIX ex: <http://example.org/cars/>
            
            SELECT (COUNT(?car) as ?count)
            WHERE {
                ?car a ex:Coupe .
            }
        """,
        "4. Average Price of cars": """
            PREFIX ex: <http://example.org/cars/>
            
            SELECT (AVG(?price) as ?avg_price)
            WHERE {
                ?car ex:hasPriceValue ?price .
            }
        """,
        "5. (Federated) Get Manufacturer Description from DBpedia": """
            PREFIX ex: <http://example.org/cars/>
            PREFIX owl: <http://www.w3.org/2002/07/owl#>
            PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
            PREFIX dbo: <http://dbpedia.org/ontology/>
            
            SELECT ?manu_name ?desc
            WHERE {
                ?manu a ex:Manufacturer ;
                      rdfs:label ?manu_name ;
                      owl:sameAs ?dbpedia_link .
                
                SERVICE <http://dbpedia.org/sparql> {
                    ?dbpedia_link dbo:abstract ?desc .
                    FILTER (LANG(?desc) = 'en')
                }
            }
            LIMIT 3
        """
    }
    
    print(f"Loaded Graph with {len(g)} triples.\n")
    
    # Run Local Queries
    for title, query in list(queries.items())[:-1]: # Skip the last one for special handling
        print(f"--- {title} ---")
        try:
            results = g.query(query)
            for row in results:
                print(row)
        except Exception as e:
            print(f"Query Error: {e}")
        print("\n")

    # Run Federated Query (Special Handling)
    print(f"--- 5. (Federated) Get Manufacturer Description from DBpedia ---")
    print("Attempting to run SERVICE query via rdflib (may fail due to DBpedia restrictions)...")
    fed_query = queries["5. (Federated) Get Manufacturer Description from DBpedia"]
    
    try:
        results = g.query(fed_query)
        for row in results:
            print(row)
    except Exception as e:
        print(f"Standard SERVICE query failed ({e}).\nData is likely interlinked, but local engine cannot negotiate with DBpedia.")
        print("Fallback: Verifying Interlinking manually via SPARQLWrapper...")
        
        try:
            from SPARQLWrapper import SPARQLWrapper, JSON
            sparql = SPARQLWrapper("http://dbpedia.org/sparql")
            
            # Get a manufacturer link from our graph
            manu_link_query = """
                PREFIX ex: <http://example.org/cars/>
                PREFIX owl: <http://www.w3.org/2002/07/owl#>
                SELECT ?link WHERE { ?m a ex:Manufacturer ; owl:sameAs ?link . } LIMIT 1
            """
            res = g.query(manu_link_query)
            link = list(res)[0][0]
            print(f"Found Interlink: {link}")
            
            # Query DBpedia for that link
            dbpedia_q = f"""
                PREFIX dbo: <http://dbpedia.org/ontology/>
                SELECT ?desc WHERE {{ <{link}> dbo:abstract ?desc . FILTER (LANG(?desc) = 'en') }} LIMIT 1
            """
            sparql.setQuery(dbpedia_q)
            sparql.setReturnFormat(JSON)
            results = sparql.query().convert()
            
            for result in results["results"]["bindings"]:
                print(f"Description from DBpedia: {result['desc']['value'][:200]}...")
                
        except ImportError:
            print("SPARQLWrapper not installed. Cannot run fallback.")
        except Exception as ex:
            print(f"Fallback failed: {ex}")

if __name__ == "__main__":
    run_validation()