Update app.py

app.py

@@ -37,10 +37,8 @@ if uploaded_file is not None:
         "unknown_params": []
     }
 
-    # قوائم لتحليل الصيغ وأنواع الروابط
+    # قائمة لتحليل الصيغ (Extensions)
     extensions = {}
-    internal_links = []
-    external_links = []
 
     # متغيرات للتحكم في العملية
     progress_bar = st.progress(0)
@@ -116,27 +114,21 @@ if uploaded_file is not None:
                     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)}")
+                        param_categories["shop_params"].append(param)
                     elif best_category == "game" and score > 0.5:
-                        param_categories["game_params"].append(f"{param}={re.search(param + r'=([^&]*)', url).group(1)}")
+                        param_categories["game_params"].append(param)
                     elif best_category == "stream" and score > 0.5:
-                        param_categories["stream_params"].append(f"{param}={re.search(param + r'=([^&]*)', url).group(1)}")
+                        param_categories["stream_params"].append(param)
                     else:
-                        param_categories["unknown_params"].append(f"{param}={re.search(param + r'=([^&]*)', url).group(1)}")
-
-                # تصنيف الرابط بناءً على نوعه (داخلي/خارجي)
-                if re.match(r'^https?://[^/]+\.example\.com', url):
-                    internal_links.append(url)
-                else:
-                    external_links.append(url)
+                        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] = []
-                    extensions[ext].append(url)
+                        extensions[ext] = 0
+                    extensions[ext] += 1
 
             # تحديث المؤشر الحالي
             current_index = i + batch_size
@@ -160,17 +152,13 @@ if uploaded_file is not None:
         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"])
-
-        # تحديث محتوى المربعات الخاصة بأنواع الروابط
-        st.session_state.internal_links = "\n".join(internal_links)
-        st.session_state.external_links = "\n".join(external_links)
+        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 المربعات الخاصة بالصيغ
-        st.session_state.extensions_text = "\n\n".join([f"{k}: {', '.join(v)}" for k, v in extensions.items()])
+        # تحديث محتوى المربع الخاص بالصيغ
+        st.session_state.extensions_text = "\n".join(extensions.keys())
 
     # زر البدء
     if st.button("Start"):
@@ -247,18 +235,7 @@ if uploaded_file is not None:
             st.session_state.unknown_params = ""
         st.text_area("Copy the unknown parameters here:", value=st.session_state.unknown_params, height=200, key="unknown_params")
 
-        # عرض أنواع الروابط
-        st.header("Internal Links")
-        if 'internal_links' not in st.session_state:
-            st.session_state.internal_links = ""
-        st.text_area("Copy the internal links here:", value=st.session_state.internal_links, height=200, key="internal_links")
-
-        st.header("External Links")
-        if 'external_links' not in st.session_state:
-            st.session_state.external_links = ""
-        st.text_area("Copy the external links here:", value=st.session_state.external_links, height=200, key="external_links")
-
-        # عرض الصيغ
+        # عرض الصيغ (Extensions)
         st.header("File Extensions")
         if 'extensions_text' not in st.session_state:
             st.session_state.extensions_text = ""