Spaces:
Sleeping
Sleeping
| 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": [] | |
| } | |
| # قائمة لتحليل الصيغ (Extensions) | |
| extensions = {} | |
| # قائمة لتحليل أنماط الصفحات الكاملة (Full PageType) | |
| full_page_types = [] | |
| # 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 # مؤشر للكلمة الحالية | |
| batch_size = 10 # عدد العناصر التي يتم معالجتها في الدفعة الواحدة | |
| # دالة تصنيف الكلمات باستخدام الدفعات | |
| def classify_keywords_batch(items, categories, start_index=0): | |
| global paused, stopped, current_index | |
| total_items = len(items) | |
| for i in range(start_index, total_items, batch_size): | |
| if stopped: | |
| break | |
| if paused: | |
| time.sleep(0.5) | |
| continue | |
| # معالجة دفعة من العناصر | |
| batch = items[i:i + batch_size] | |
| results = classifier(batch, categories) | |
| for j, result in enumerate(results): | |
| best_category = result['labels'][0] | |
| score = result['scores'][0] | |
| if best_category == "shop" and score > 0.5: | |
| shopping_items.append(batch[j]) | |
| elif best_category == "game" and score > 0.5: | |
| gaming_items.append(batch[j]) | |
| elif best_category == "stream" and score > 0.5: | |
| streaming_items.append(batch[j]) | |
| else: | |
| unknown_items.append(batch[j]) | |
| # تحديث المؤشر الحالي | |
| current_index = min(i + batch_size, total_items) # تأكد من عدم تجاوز المؤشر للعدد الإجمالي | |
| # تحديث شريط التقدم | |
| progress = min((current_index) / total_items, 1.0) # تأكد من أن قيمة التقدم لا تتجاوز 1.0 | |
| 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 in range(start_index, total_items, batch_size): | |
| if stopped: | |
| break | |
| if paused: | |
| time.sleep(0.5) | |
| continue | |
| # معالجة دفعة من الروابط | |
| batch = items[i:i + batch_size] | |
| for url in batch: | |
| # 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(param) | |
| elif best_category == "game" and score > 0.5: | |
| param_categories["game_params"].append(param) | |
| elif best_category == "stream" and score > 0.5: | |
| param_categories["stream_params"].append(param) | |
| else: | |
| param_categories["unknown_params"].append(param) | |
| # استخراج الصيغ (Extensions) من الروابط | |
| match = re.search(r'\.([a-zA-Z0-9]+)$', url) | |
| if match: | |
| ext = match.group(1) | |
| if ext not in extensions: | |
| extensions[ext] = 0 | |
| extensions[ext] += 1 | |
| # استخراج أنماط الصفحات الكاملة (Full PageType) | |
| page_type_match = re.search(r'(\w+\.php|\w+\.html)\?', url) | |
| if page_type_match: | |
| page_type = page_type_match.group(1) | |
| if page_type not in full_page_types: | |
| full_page_types.append(page_type) | |
| # تحديث المؤشر الحالي | |
| current_index = min(i + batch_size, total_items) # تأكد من عدم تجاوز المؤشر للعدد الإجمالي | |
| # تحديث شريط التقدم | |
| progress = min((current_index) / total_items, 1.0) # تأكد من أن قيمة التقدم لا تتجاوز 1.0 | |
| 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(set(param_categories["shop_params"])) | |
| st.session_state.game_params = "\n".join(set(param_categories["game_params"])) | |
| st.session_state.stream_params = "\n".join(set(param_categories["stream_params"])) | |
| st.session_state.unknown_params = "\n".join(set(param_categories["unknown_params"])) | |
| # تحديث محتوى المربع الخاص بالصيغ | |
| st.session_state.extensions_text = "\n".join(extensions.keys()) | |
| # تحديث محتوى المربع الخاص بأنماط الصفحات الكاملة | |
| st.session_state.full_page_types = "\n".join(full_page_types) | |
| # دالة تصدير النتائج | |
| def export_results(key, filename): | |
| with open(filename, "w") as f: | |
| f.write(st.session_state[key]) | |
| st.success(f"Results exported to {filename}") | |
| # زر البدء | |
| if st.button("Start"): | |
| stopped = False | |
| paused = False | |
| current_index = 0 | |
| if operation == "Filter Keywords": | |
| classify_keywords_batch(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_batch(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.button("Export Shopping Keywords", on_click=export_results, args=("shopping_text", "shopping_keywords.txt")) | |
| 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.button("Export Gaming Keywords", on_click=export_results, args=("gaming_text", "gaming_keywords.txt")) | |
| 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.button("Export Streaming Keywords", on_click=export_results, args=("streaming_text", "streaming_keywords.txt")) | |
| 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") | |
| st.button("Export Unknown Keywords", on_click=export_results, args=("unknown_text", "unknown_keywords.txt")) | |
| 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.button("Export Shop Parameters", on_click=export_results, args=("shop_params", "shop_params.txt")) | |
| 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.button("Export Game Parameters", on_click=export_results, args=("game_params", "game_params.txt")) | |
| 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.button("Export Stream Parameters", on_click=export_results, args=("stream_params", "stream_params.txt")) | |
| 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") | |
| st.button("Export Unknown Parameters", on_click=export_results, args=("unknown_params", "unknown_params.txt")) | |
| # عرض الصيغ (Extensions) | |
| st.header("File Extensions") | |
| if 'extensions_text' not in st.session_state: | |
| st.session_state.extensions_text = "" | |
| st.text_area("Copy the file extensions here:", value=st.session_state.extensions_text, height=200, key="extensions") | |
| st.button("Export File Extensions", on_click=export_results, args=("extensions_text", "file_extensions.txt")) | |
| # عرض أنماط الصفحات الكاملة (Full PageType) | |
| st.header("Full PageType") | |
| if 'full_page_types' not in st.session_state: | |
| st.session_state.full_page_types = "" | |
| st.text_area("Copy the full page types here:", value=st.session_state.full_page_types, height=200, key="full_page_types") | |
| st.button("Export Full PageTypes", on_click=export_results, args=("full_page_types", "full_page_types.txt")) | |
| else: | |
| st.warning("Please upload a text file to start classification.") |