evaluate_bot.py

import json
import random
from app import get_response, predict_intent


with open("intents_augmented.json", encoding="utf-8") as f:
    intents = json.load(f)["intents"]

total = 0
correct = 0
results = []

print("\n بدء اختبار الردود الفعلية من البوت...\n")

for intent in intents:
    tag = intent["tag"]
    patterns = intent["patterns"]
    responses = intent["responses"]

    for example in patterns:
        total += 1
        reply = get_response(example)
        predicted_tag, conf = predict_intent(example)


        is_correct = any(r.strip() == reply.strip() for r in responses)

        results.append({
            "input": example,
            "expected_tag": tag,
            "predicted_tag": predicted_tag,
            "confidence": round(conf, 3),
            "reply": reply,
            "is_correct": is_correct
        })

        status = "T" if is_correct else "F"
        print(f"{status} {example} → {predicted_tag} ({conf:.2f})")

        if is_correct:
            correct += 1

accuracy = (correct / total) * 100
print(f"\n دقة الردود الفعلية: {accuracy:.2f}% ({correct}/{total})\n")


with open("results/full_eval.json", "w", encoding="utf-8") as f:
    json.dump(results, f, ensure_ascii=False, indent=2)

print("تم حفظ التفاصيل في results/full_eval.json")