export_triplets.py

from neo4j import GraphDatabase
import pandas as pd
import os
from dotenv import load_dotenv

load_dotenv()

def get_neo4j_driver():
    uri = os.getenv("NEO4J_URI")
    user = os.getenv("NEO4J_USER")
    password = os.getenv("NEO4J_PASSWORD")
    try:
        driver = GraphDatabase.driver(uri, auth=(user, password))
        driver.verify_connectivity()
        return driver
    except Exception as e:
        print(f"Errore nella connessione a Neo4j: {e}")
        return None

def run_cypher_query(_driver, query):
    with _driver.session(database=os.getenv("NEO4J_DATABASE")) as session:
        results = session.run(query)
        return [record.data() for record in results]

if __name__ == "__main__":
    driver = get_neo4j_driver()
    if driver is None:
        exit(1)

    # Estraiamo tutte le triplette (head, relation, tail)
    query = """
    MATCH (p:Player)-[r:playsForTeam]->(t:Team) RETURN p.playerId AS head, type(r) AS relation, t.teamId AS tail
    UNION
    MATCH (m:Manager)-[r:coachesTeam]->(t:Team) RETURN m.managerId AS head, type(r) AS relation, t.teamId AS tail
    UNION
    MATCH (p:Player)-[r:participatedInMatch]->(m:Match) RETURN p.playerId AS head, type(r) AS relation, m.matchId AS tail
    UNION
    MATCH (t:Team)-[r:wasHomeTeamIn]->(m:Match) RETURN t.teamId AS head, type(r) AS relation, m.matchId AS tail
    UNION
    MATCH (t:Team)-[r:wasAwayTeamIn]->(m:Match) RETURN t.teamId AS head, type(r) AS relation, m.matchId AS tail
    UNION
    MATCH (m:Match)-[r:playedAt]->(s:Stadium) RETURN m.matchId AS head, type(r) AS relation, s.stadiumId AS tail
    UNION
    MATCH (g:Goal)-[r:scoredIn]->(m:Match) RETURN g.goalId AS head, type(r) AS relation, m.matchId AS tail
    UNION
    MATCH (g:Goal)-[r:scoredBy]->(p:Player) RETURN g.goalId AS head, type(r) AS relation, p.playerId AS tail
    UNION
    MATCH (b:Booking)-[r:givenIn]->(m:Match) RETURN b.bookingId AS head, type(r) AS relation, m.matchId AS tail
    UNION
    MATCH (b:Booking)-[r:receivedBy]->(p:Player) RETURN b.bookingId AS head, type(r) AS relation, p.playerId AS tail
    UNION
    MATCH (p:Player)-[r:wonAward]->(a:Award) RETURN p.playerId AS head, type(r) AS relation, a.awardId AS tail
    UNION
    MATCH (s:Substitution)-[r:occurredInMatch]->(m:Match) RETURN s.substitutionId AS head, type(r) AS relation, m.matchId AS tail
    UNION
    MATCH (s:Substitution)-[r:hasPlayerComingOn]->(p:Player) RETURN s.substitutionId AS head, type(r) AS relation, p.playerId AS tail
    UNION
    MATCH (s:Substitution)-[r:hasPlayerGoingOff]->(p:Player) RETURN s.substitutionId AS head, type(r) AS relation, p.playerId AS tail
    UNION
    MATCH (t:Tournament)-[r:hasParticipant]->(te:Team) RETURN t.tournamentId AS head, type(r) AS relation, te.teamId AS tail
    UNION
    MATCH (t:Tournament)-[r:includesMatch]->(m:Match) RETURN t.tournamentId AS head, type(r) AS relation, m.matchId AS tail
    UNION
    MATCH (t:Tournament)-[r:hasWinner]->(te:Team) RETURN t.tournamentId AS head, type(r) AS relation, te.teamId AS tail
    """

    results = run_cypher_query(driver, query)

    df = pd.DataFrame(results)
    df.to_csv('fifa_triplets.tsv', sep='\t', header=False, index=False)
    print("Triplets exported to fifa_triplets.tsv")