image_embedding_loader_kg.py

import json
from neo4j import GraphDatabase
from dotenv import load_dotenv
import os
from tqdm import tqdm

load_dotenv()

def connect_to_neo4j():
    uri = os.getenv("NEO4J_URI")
    user = os.getenv("NEO4J_USER")
    password = os.getenv("NEO4J_PASSWORD")

    try:
        driver = GraphDatabase.driver(uri, auth=(user, password))
        return driver
    except Exception as e:
        print(f"Errore di connessione a Neo4j: {e}")
        return None

if __name__ == "__main__":
    driver = connect_to_neo4j()
    if driver is None:
        exit(1)

    with driver.session(database=os.getenv("NEO4J_DATABASE")) as session:
        print("Caricamento delle immagini e dei relativi embedding in Neo4j...")

        with open('image_embeddings_avg.json', 'r', encoding='utf-8') as f:
            image_embeddings = json.load(f)

        query = """
        UNWIND $rows AS row

        // Troviamo il giocatore corrispondente in base a playerId
        MATCH (p:Player {playerId: row.player_id})

        // Creiamo o aggiorniamo la proprietà imageEmbedding
        SET p.imageEmbedding = row.embedding,
            p.imageFolder = row.image_folder

        RETURN count(p) AS count
        """

        rows_to_insert = [
            {
                "player_id": player_id,
                "embedding": embedding_data["embedding"],
                "image_folder": embedding_data["image_folder"]
            }
            for player_id, embedding_data in image_embeddings.items()
        ]

        with driver.session(database=os.getenv("NEO4J_DATABASE")) as session:
            batch_size = 200
            for i in tqdm(range(0, len(rows_to_insert), batch_size), desc="Inserimento in batch"):
                batch = rows_to_insert[i:i + batch_size]
                result = session.run(query, rows=batch)

        print("Caricamento completato.")

    driver.close()
    print("Connessione a Neo4j chiusa.")