Upload NavyBayes.py

NavyBayes.py

@@ -0,0 +1,171 @@
+import firebase_admin
+from firebase_admin import credentials, firestore
+from joblib import dump, load
+import datetime
+from sklearn.feature_extraction.text import TfidfVectorizer
+from sklearn.naive_bayes import MultinomialNB
+import pandas as pd
+from huggingface_hub import HfApi
+
+# التهيئة مرة واحدة فقط
+if not firebase_admin._apps:
+    # تأكد من وضع المسار الصحيح لملف التوثيق Firebase
+    cred = credentials.Certificate("D:/app-sentinel-7qnr19-firebase-adminsdk-kjmbe-f38e16a432.json")
+    firebase_admin.initialize_app(cred)
+
+db = firestore.client()
+
+# تحميل النموذج الحالي والمحول
+try:
+    model = load('model.joblib')
+    vectorizer = load('vectorizer.joblib')
+    print("Model and vectorizer loaded successfully.")
+except Exception as e:
+    model = None
+    vectorizer = None
+    print(f"Model and vectorizer not found. You need to train the model. Error: {e}")
+
+# 1. وظيفة لتحليل وتصنيف الرسائل وتخزينها في Firestore
+def classify_and_store_message(message):
+    global model, vectorizer
+    try:
+        if not model or not vectorizer:
+            raise ValueError("Model or vectorizer not loaded. Train or load the model first.")
+
+        # تحويل الرسالة إلى سمات رقمية
+        message_vector = vectorizer.transform([message])
+        classification = model.predict(message_vector)[0]
+
+        # إعداد البيانات للتخزين
+        message_data = {
+            'text': message,
+            'classification': classification,
+            'timestamp': datetime.datetime.now()
+        }
+
+        # تخزين الرسالة في مجموعة Firestore حسب التصنيف
+        if classification == "spam_phishing":
+            db.collection('spam').add(message_data)
+        elif classification == "news_phishing":
+            db.collection('news').add(message_data)
+        elif classification == "advertisement_phishing":
+            db.collection('advertisement').add(message_data)
+        elif classification == "social_phishing":
+            db.collection('social').add(message_data)
+
+        # تخزين الرسالة في مجموعة 'all_messages' لجميع الرسائل
+        db.collection('all_messages').add(message_data)
+
+        # تخزين الرسالة في مجموعة 'recently_analyzed_messages' للرسائل الحديثة
+        db.collection('recently_analyzed_messages').add(message_data)
+
+        print(f"Message classified as {classification} and stored in Firestore.")
+        return classification
+
+    except Exception as e:
+        print(f"Error classifying message: {e}")
+        return None
+
+# 2. رفع النموذج إلى Hugging Face Hub
+def upload_model_to_huggingface():
+    try:
+        # استخدام HfApi لرفع النموذج
+        api = HfApi()
+
+        # تحديد اسم النموذج والملفات
+        model_name = "Noufy/https-api_inference.huggingface.comodels-Noufy-naive_bayes_sms"
+        model_path = "model.joblib"
+        vectorizer_path = "vectorizer.joblib"
+
+        # رفع النموذج إلى Hugging Face
+        api.upload_file(
+            path_or_fileobj=model_path,
+            path_in_repo=f"{model_name}/model.joblib",
+            repo_id="Noufy/https-api_inference.huggingface.comodels-Noufy-naive_bayes_sms",  # ضع اسم المستخدم واسم المشروع
+            repo_type="model"
+        )
+
+        api.upload_file(
+            path_or_fileobj=vectorizer_path,
+            path_in_repo=f"{model_name}/vectorizer.joblib",
+            repo_id="Noufy/https-api_inference.huggingface.comodels-Noufy-naive_bayes_sms",
+            repo_type="model"
+        )
+
+        print("Model and vectorizer uploaded successfully to Hugging Face.")
+
+    except Exception as e:
+        print(f"Error uploading model: {e}")
+
+# 3. تحديث النموذج مع بيانات جديدة
+def update_model_with_new_data(new_messages, new_labels):
+    global model, vectorizer
+    try:
+        # تحميل البيانات الحالية
+        data = {
+            'message': new_messages,
+            'label': new_labels
+        }
+        df_new = pd.DataFrame(data)
+
+        # تحديث المحول والنموذج
+        if vectorizer is None or model is None:
+            vectorizer = TfidfVectorizer()
+            X_new = vectorizer.fit_transform(df_new['message'])
+        else:
+            X_new = vectorizer.transform(df_new['message'])
+
+        # جمع البيانات الجديدة مع القديمة وإعادة التدريب
+        y_new = df_new['label']
+        if model is None:
+            model = MultinomialNB()
+        model.partial_fit(X_new, y_new, classes=['spam_phishing', 'social_phishing', 'news_phishing', 'advertisement_phishing'])
+
+        # حفظ النموذج ا��جديد
+        dump(model, 'model.joblib')
+        dump(vectorizer, 'vectorizer.joblib')
+        print("Model updated and saved successfully.")
+
+    except Exception as e:
+        print(f"Error updating model: {e}")
+
+# تشغيل رفع النموذج إلى Hugging Face
+upload_model_to_huggingface()
+
+# 4. دالة لاختبار النظام
+def test_system():
+    test_messages = [
+        "Win a free vacation now!",
+        "Breaking news: Major stock updates today.",
+        "Don't forget our meeting tomorrow at 10 AM.",
+        "Click here to secure your bank account now!",
+        "Exclusive offers just for you, buy now!"
+    ]
+
+    for msg in test_messages:
+        classification = classify_and_store_message(msg)
+        print(f"Message: '{msg}' -> Classified as: {classification}")
+
+# 5. وظيفة للتصحيح اليدوي
+def correct_classification(message_id, correct_label):
+    try:
+        # جلب الرسالة من Firebase
+        message_ref = db.collection('view_history').document(message_id)
+        message_data = message_ref.get().to_dict()
+
+        if not message_data:
+            print("Message not found.")
+            return
+
+        # تحديث التصنيف في Firebase
+        message_data['classification'] = correct_label
+        message_ref.update({'classification': correct_label})
+
+        # إضافة البيانات إلى نموذج التدريب الجديد
+        update_model_with_new_data([message_data['text']], [correct_label])
+        print(f"Message classification corrected to {correct_label} and model updated.")
+    except Exception as e:
+        print(f"Error correcting classification: {e}")
+
+# اختبار
+test_system()