from sklearn.pipeline import Pipeline
from sklearn.naive_bayes import MultinomialNB
from sklearn.feature_extraction.text import TfidfVectorizer
from joblib import dump, load
import firebase_admin
from firebase_admin import credentials, firestore
import logging
import datetime
import re
import pandas as pd
import os
from flask import Flask, request, jsonify

# إعداد السجلات
logging.basicConfig(level=logging.DEBUG, format="%(asctime)s - %(levelname)s - %(message)s")

# Firebase Initialization
try:
    cred_path = r"D:\app-sentinel-7qnr19-firebase-adminsdk-kjmbe-533749ec1a.json"
    if not firebase_admin._apps:
        cred = credentials.Certificate(cred_path)
        firebase_admin.initialize_app(cred)
    db = firestore.client()
    logging.info("Firebase initialized successfully.")
except Exception as e:
    logging.error(f"Error initializing Firebase: {e}")
    db = None

# Load or Train Model with Pipeline
try:
    model_path = os.path.join(os.getcwd(), "model_pipeline.joblib")
    model_pipeline = load(model_path)
    logging.info("Model pipeline loaded successfully.")
except Exception as e:
    logging.warning(f"Model pipeline not found. Training new one. Error: {e}")
    try:
        # Train new model and vectorizer as part of pipeline
        messages = ["example message 1", "example message 2"]
        labels = ["label1", "label2"]
        model_pipeline = Pipeline([
            ('vectorizer', TfidfVectorizer()),  # تحويل النصوص إلى تمثيل رقمي
            ('classifier', MultinomialNB())      # تصنيف باستخدام Naive Bayes
        ])
        model_pipeline.fit(messages, labels)
        dump(model_pipeline, model_path)
        logging.info("New model pipeline trained and saved.")
    except Exception as e:
        logging.error(f"Error training new model: {e}")

# Flask API
app = Flask(__name__)

@app.route('/classify', methods=['POST'])
def classify_message():
    data = request.json
    message = data.get("message", "")
    
    # التصنيف باستخدام الـ pipeline
    if message.strip() == "":
        return jsonify({"error": "Message cannot be empty"}), 400
    
    classification = model_pipeline.predict([message])[0]
    
    # تخزين الرسالة في Firestore
    message_data = {
        "text": message,
        "classification": classification,
        "timestamp": datetime.datetime.now(),
    }

    if db:
        db.collection("all_messages").add(message_data)
        logging.info(f"Message classified as {classification} and stored in Firestore.")
    else:
        logging.warning("Firestore is not initialized. Data not stored.")
    
    return jsonify({"classification": classification})

if __name__ == "__main__":
    app.run(debug=True)