Update app.py

app.py

@@ -40,6 +40,9 @@ if uploaded_file is not None:
     # 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")
@@ -50,20 +53,6 @@ if uploaded_file is not None:
     current_index = 0  # مؤشر للكلمة الحالية
     batch_size = 10  # عدد العناصر التي يتم معالجتها في الدفعة الواحدة
 
-    # دالة التحقق من صحة الملف
-    def validate_file(items, operation):
-        if operation == "Filter Keywords":
-            # التحقق من أن الملف يحتوي على كلمات مفتاحية فقط
-            for item in items:
-                if re.match(r'https?://', item):
-                    return False, "File should contain keywords only."
-        elif operation == "Extra & Filter Param (URLs)":
-            # التحقق من أن الملف يحتوي على روابط فقط
-            for item in items:
-                if not re.match(r'https?://', item):
-                    return False, "File should contain URLs only."
-        return True, ""
-
     # دالة تصنيف الكلمات باستخدام الدفعات
     def classify_keywords_batch(items, categories, start_index=0):
         global paused, stopped, current_index
@@ -144,6 +133,13 @@ if uploaded_file is not None:
                         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 = i + batch_size
 
@@ -174,21 +170,25 @@ if uploaded_file is not None:
         # 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
 
-        # التحقق من صحة الملف
-        is_valid, error_message = validate_file(items, operation)
-        if not is_valid:
-            st.error(error_message)
-        else:
-            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)
+        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:
@@ -216,21 +216,25 @@ if uploaded_file is not None:
         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
@@ -238,27 +242,39 @@ if uploaded_file is not None:
         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.")