demo_script.py

import csv
import os
import sys
import base64
import requests
import logging
from pathlib import Path

# ==================== CONFIGURAZIONE ====================

API_KEY = "ChiaveAPI" 
MODEL_NAME = "gpt-4o"
IMAGE_DIR = Path("images")
CSV_PATH = Path("Dataset_Art-0725.csv")
OUTPUT_CSV = Path("risposte_output.csv")

CATEGORIE = {
    "AR": "art_recognition",
    "CR": "chronological_reasoning",
    "CS": "contextual_summary",
    "VR": "vision_reading",
    "VB": "vision_basic",
    "VL": "vision_logic",
    "VRS": "vision_reasoning",
    "IG" : "img_gen"
}

# ==================== LOGGING ====================

logging.basicConfig(
    filename='task_log.txt',
    level=logging.INFO,
    format='%(asctime)s - %(levelname)s - %(message)s'
)

# ==================== FUNZIONI ====================

def encode_image(image_path: Path) -> str:
    with open(image_path, "rb") as img:
        return base64.b64encode(img.read()).decode("utf-8")

def validate_image_paths(row):
    missing = []
    for col in ['immagine_1_path', 'immagine_2_path']:
        if row[col].strip():
            path = IMAGE_DIR / row[col].strip()
            if not path.exists():
                missing.append(path)
    return missing

def invia_task(prompt, image_paths):
    content = [{"type": "text", "text": prompt}]
    for img_path in image_paths:
        if img_path.exists():
            b64 = encode_image(img_path)
            content.append({
                "type": "image_url",
                "image_url": {
                    "url": f"data:image/jpeg;base64,{b64}"
                }
            })

    payload = {
        "model": MODEL_NAME,
        "messages": [
            {
                "role": "system",
                "content": (
                    "Sei uno storico dell’arte specializzato in analisi iconografiche e storico-stilistiche. "
                    "Rispondi sempre in italiano, in stile accademico, formale e neutrale. "
                    "Analizza secondo i criteri storico-artistici e rispondi in modo rigoroso e preciso."
                )
            },
            {"role": "user", "content": content}
        ],
        "max_tokens": 1000
    }

    response = requests.post(
        "https://api.openai.com/v1/chat/completions",
        headers={"Authorization": f"Bearer {API_KEY}", "Content-Type": "application/json"},
        json=payload
    )

    if response.status_code == 200:
        return response.json()["choices"][0]["message"]["content"]
    else:
        raise Exception(f"Errore API: {response.status_code} - {response.text}")

# ==================== MAIN ====================

def main():
    if len(sys.argv) < 2:
        print("Uso: python test_01.py AR CR ...")
        sys.exit(1)

    categorie_input = sys.argv[1:]
    categorie_mapped = {CATEGORIE[c] for c in categorie_input if c in CATEGORIE}

    if not categorie_mapped:
        print("Nessuna categoria valida fornita.")
        sys.exit(1)

    results = []

    with open(CSV_PATH, encoding='utf-8') as f:
        reader = csv.DictReader(f)
        for idx, row in enumerate(reader, 1):
            if row["categoria"] not in categorie_mapped:
                continue

            image_paths = [IMAGE_DIR / row["immagine_1_path"].strip()]
            if row["immagine_2_path"].strip():
                image_paths.append(IMAGE_DIR / row["immagine_2_path"].strip())

            missing = validate_image_paths(row)
            if missing:
                msg = f"[TASK {idx}] Immagini mancanti: {[str(m) for m in missing]}"
                logging.warning(msg)
                print(msg)
                continue

            # Costruzione del prompt
            base_prompt = row["prompt"].strip() + " " + row["instructions"].strip()
            opzioni = [row.get("opzione_1", ""), row.get("opzione_2", ""), row.get("opzione_3", "")]
            opzioni = [opt.strip() for opt in opzioni if opt.strip()]
            if opzioni:
                base_prompt += "\n\nOpzioni disponibili:\n" + "\n".join(f"- {opt}" for opt in opzioni)

            try:
                result = invia_task(base_prompt, image_paths)
                logging.info(f"[TASK {idx}] Completato con successo.")
                results.append({
                    "task_id": idx,
                    "risposta": result,
                    "prompt_inviato": base_prompt
                })
            except Exception as e:
                logging.error(f"[TASK {idx}] Errore durante l'invio: {e}")
                print(f"[TASK {idx}] Errore - vedi log")

    # Scrittura risultati su CSV
    if results:
        with open(OUTPUT_CSV, mode='w', encoding='utf-8', newline='') as out_csv:
            writer = csv.DictWriter(out_csv, fieldnames=["task_id", "risposta", "prompt_inviato"])
            writer.writeheader()
            for r in results:
                writer.writerow(r)

    print(f"Esecuzione completata. Risposte salvate in {OUTPUT_CSV}.")

if __name__ == "__main__":
    main()