app.py

import streamlit as st
from transformers import pipeline
import re
import time

# تحميل النموذج
classifier = pipeline("zero-shot-classification", model="cross-encoder/nli-distilroberta-base")

# عنوان التطبيق
st.title("Text Classification App")

# اختيار العملية
operation = st.radio("Choose an operation:", ["Filter Keywords", "Extra & Filter Param (URLs)"])

# إدخال الملف النصي
uploaded_file = st.file_uploader("Upload a text file", type=["txt"])

if uploaded_file is not None:
    # قراءة الملف النصي
    content = uploaded_file.read().decode("utf-8")
    items = [line.strip() for line in content.splitlines() if line.strip()]

    # تحديد الفئات
    categories = ["shop", "game", "stream"]

    # قوائم لتخزين النتائج
    shopping_items = []
    gaming_items = []
    streaming_items = []
    unknown_items = []

    # قوائم خاصة بالباراميترات
    param_categories = {
        "shop_params": [],
        "game_params": [],
        "stream_params": [],
        "unknown_params": []
    }

    # متغيرات للتحكم في العملية
    progress_bar = st.progress(0)
    pause_button = st.button("Pause")
    stop_button = st.button("Stop")
    continue_button = st.button("Continue")
    paused = False
    stopped = False
    current_index = 0

    # دالة تصنيف العناصر (للكلمات المفتاحية)
    def classify_keywords(items, categories, start_index=0):
        global paused, stopped, current_index
        total_items = len(items)
        for i, item in enumerate(items[start_index:], start=start_index):
            current_index = i
            if stopped:
                break
            if paused:
                time.sleep(0.5)
                continue

            # تصنيف الكلمة باستخدام zero-shot-classification
            result = classifier(item, categories)
            best_category = result['labels'][0]
            score = result['scores'][0]

            if best_category == "shop" and score > 0.5:
                shopping_items.append(item)
            elif best_category == "game" and score > 0.5:
                gaming_items.append(item)
            elif best_category == "stream" and score > 0.5:
                streaming_items.append(item)
            else:
                unknown_items.append(item)

            # تحديث شريط التقدم
            progress = (current_index + 1) / total_items
            progress_bar.progress(progress)

            # تحديث النتائج في الوقت الحقيقي
            update_results()

            # إبطاء العملية قليلاً للسماح بتحديث الواجهة
            time.sleep(0.1)

    # دالة تصنيف الباراميترات
    def classify_parameters(items, categories, start_index=0):
        global paused, stopped, current_index
        total_items = len(items)
        for i, url in enumerate(items[start_index:], start=start_index):
            current_index = i
            if stopped:
                break
            if paused:
                time.sleep(0.5)
                continue

            # استخراج الباراميترات من الرابط باستخدام RegEx
            params = re.findall(r'(\w+)=\w+', url)
            for param in params:
                # تصنيف الباراميتر باستخدام zero-shot-classification
                result = classifier(param, categories)
                best_category = result['labels'][0]
                score = result['scores'][0]

                if best_category == "shop" and score > 0.5:
                    param_categories["shop_params"].append(f"{param}={re.search(param + r'=([^&]*)', url).group(1)}")
                elif best_category == "game" and score > 0.5:
                    param_categories["game_params"].append(f"{param}={re.search(param + r'=([^&]*)', url).group(1)}")
                elif best_category == "stream" and score > 0.5:
                    param_categories["stream_params"].append(f"{param}={re.search(param + r'=([^&]*)', url).group(1)}")
                else:
                    param_categories["unknown_params"].append(f"{param}={re.search(param + r'=([^&]*)', url).group(1)}")

            # تحديث شريط التقدم
            progress = (current_index + 1) / total_items
            progress_bar.progress(progress)

            # تحديث النتائج في الوقت الحقيقي
            update_results()

            # إبطاء العملية قليلاً للسماح بتحديث الواجهة
            time.sleep(0.1)

    # دالة تحديث النتائج
    def update_results():
        # تحديث محتوى المربعات النصية
        st.session_state.shopping_text = "\n".join(shopping_items)
        st.session_state.gaming_text = "\n".join(gaming_items)
        st.session_state.streaming_text = "\n".join(streaming_items)
        st.session_state.unknown_text = "\n".join(unknown_items)

        # تحديث محتوى المربعات الخاصة بالباراميترات
        st.session_state.shop_params = "\n".join(param_categories["shop_params"])
        st.session_state.game_params = "\n".join(param_categories["game_params"])
        st.session_state.stream_params = "\n".join(param_categories["stream_params"])
        st.session_state.unknown_params = "\n".join(param_categories["unknown_params"])

    # زر البدء
    if st.button("Start"):
        stopped = False
        paused = False
        current_index = 0

        if operation == "Filter Keywords":
            classify_keywords(items, categories, start_index=current_index)
        elif operation == "Extra & Filter Param (URLs)":
            classify_parameters(items, categories, start_index=current_index)

    # زر الإيقاف المؤقت
    if pause_button:
        paused = True
        st.write("Classification paused.")

    # زر الاستمرار
    if continue_button and paused:
        paused = False
        st.write("Classification resumed.")
        if operation == "Filter Keywords":
            classify_keywords(items, categories, start_index=current_index)
        elif operation == "Extra & Filter Param (URLs)":
            classify_parameters(items, categories, start_index=current_index)

    # زر التوقف الكامل
    if stop_button:
        stopped = True
        st.write("Classification stopped.")

    # عرض النتائج بناءً على الخيار المختار
    if operation == "Filter Keywords":
        # عرض النتائج للكلمات المفتاحية
        st.header("Shopping Keywords")
        if 'shopping_text' not in st.session_state:
            st.session_state.shopping_text = ""
        st.text_area("Copy the shopping keywords here:", value=st.session_state.shopping_text, height=200, key="shopping")

        st.header("Gaming Keywords")
        if 'gaming_text' not in st.session_state:
            st.session_state.gaming_text = ""
        st.text_area("Copy the gaming keywords here:", value=st.session_state.gaming_text, height=200, key="gaming")

        st.header("Streaming Keywords")
        if 'streaming_text' not in st.session_state:
            st.session_state.streaming_text = ""
        st.text_area("Copy the streaming keywords here:", value=st.session_state.streaming_text, height=200, key="streaming")

        st.header("Unknown Keywords")
        if 'unknown_text' not in st.session_state:
            st.session_state.unknown_text = ""
        st.text_area("Copy the unknown keywords here:", value=st.session_state.unknown_text, height=200, key="unknown")

    elif operation == "Extra & Filter Param (URLs)":
        # عرض النتائج للباراميترات
        st.header("Shop Parameters")
        if 'shop_params' not in st.session_state:
            st.session_state.shop_params = ""
        st.text_area("Copy the shop parameters here:", value=st.session_state.shop_params, height=200, key="shop_params")

        st.header("Game Parameters")
        if 'game_params' not in st.session_state:
            st.session_state.game_params = ""
        st.text_area("Copy the game parameters here:", value=st.session_state.game_params, height=200, key="game_params")

        st.header("Stream Parameters")
        if 'stream_params' not in st.session_state:
            st.session_state.stream_params = ""
        st.text_area("Copy the stream parameters here:", value=st.session_state.stream_params, height=200, key="stream_params")

        st.header("Unknown Parameters")
        if 'unknown_params' not in st.session_state:
            st.session_state.unknown_params = ""
        st.text_area("Copy the unknown parameters here:", value=st.session_state.unknown_params, height=200, key="unknown_params")

else:
    st.warning("Please upload a text file to start classification.")