delete docs

app.py

@@ -1158,6 +1158,114 @@ class PDFSearchApp:
         else:
             return "medium"
     
+    def delete_documents(self, collection_name=None):
+        """
+        Delete documents and their associated collections from the system
+        
+        Args:
+            collection_name: Name of the collection to delete. If None, deletes all collections.
+            
+        Returns:
+            Status message about the deletion operation
+        """
+        try:
+            print(f"🗑️ DELETE DOCUMENTS REQUESTED")
+            print(f"📁 Collection to delete: {collection_name if collection_name else 'ALL COLLECTIONS'}")
+            
+            if not self.indexed_docs:
+                return "❌ No documents found to delete. Please upload some documents first."
+            
+            deleted_collections = []
+            deleted_files = []
+            
+            if collection_name:
+                # Delete specific collection
+                if collection_name in self.indexed_docs:
+                    collection_info = self.indexed_docs[collection_name]
+                    
+                    # Delete from Milvus
+                    try:
+                        middleware = Middleware(collection_name, create_collection=False)
+                        middleware.drop_collection()
+                        print(f"✅ Dropped Milvus collection: {collection_name}")
+                    except Exception as e:
+                        print(f"⚠️ Warning: Could not drop Milvus collection {collection_name}: {e}")
+                    
+                    # Delete page images
+                    try:
+                        base_output_dir = self._ensure_base_directory()
+                        collection_dir = os.path.join(base_output_dir, collection_name)
+                        if os.path.exists(collection_dir):
+                            shutil.rmtree(collection_dir)
+                            print(f"✅ Deleted page images directory: {collection_dir}")
+                            deleted_files.append(f"Page images: {collection_dir}")
+                    except Exception as e:
+                        print(f"⚠️ Warning: Could not delete page images for {collection_name}: {e}")
+                    
+                    # Remove from indexed_docs
+                    del self.indexed_docs[collection_name]
+                    deleted_collections.append(collection_name)
+                    
+                    return f"✅ Successfully deleted collection '{collection_name}'\n📁 Deleted: {len(deleted_files)} file/directory items"
+                else:
+                    return f"❌ Collection '{collection_name}' not found. Available collections: {list(self.indexed_docs.keys())}"
+            else:
+                # Delete all collections
+                for coll_name in list(self.indexed_docs.keys()):
+                    try:
+                        # Delete from Milvus
+                        middleware = Middleware(coll_name, create_collection=False)
+                        middleware.drop_collection()
+                        print(f"✅ Dropped Milvus collection: {coll_name}")
+                    except Exception as e:
+                        print(f"⚠️ Warning: Could not drop Milvus collection {coll_name}: {e}")
+                    
+                    # Delete page images
+                    try:
+                        base_output_dir = self._ensure_base_directory()
+                        collection_dir = os.path.join(base_output_dir, coll_name)
+                        if os.path.exists(collection_dir):
+                            shutil.rmtree(collection_dir)
+                            print(f"✅ Deleted page images directory: {collection_dir}")
+                            deleted_files.append(f"Page images: {collection_dir}")
+                    except Exception as e:
+                        print(f"⚠️ Warning: Could not delete page images for {coll_name}: {e}")
+                    
+                    deleted_collections.append(coll_name)
+                
+                # Clear all indexed docs
+                self.indexed_docs.clear()
+                
+                return f"✅ Successfully deleted ALL collections ({len(deleted_collections)} total)\n📁 Deleted: {len(deleted_files)} file/directory items\n🗑️ Collections deleted: {', '.join(deleted_collections)}"
+                
+        except Exception as e:
+            error_msg = f"❌ Error during document deletion: {str(e)}"
+            print(f"{error_msg}")
+            print(f"❌ Traceback: {traceback.format_exc()}")
+            return error_msg
+    
+    def get_available_collections(self):
+        """
+        Get list of available collections for deletion
+        
+        Returns:
+            List of collection names and their details
+        """
+        if not self.indexed_docs:
+            return "No collections available for deletion."
+        
+        collection_list = []
+        for collection_name, collection_info in self.indexed_docs.items():
+            collection_list.append(f"📁 {collection_name}")
+            if isinstance(collection_info, dict):
+                if 'files' in collection_info:
+                    collection_list.append(f"   📄 Files: {len(collection_info['files'])}")
+                if 'pages' in collection_info:
+                    collection_list.append(f"   📄 Pages: {collection_info['pages']}")
+            collection_list.append("")
+        
+        return "\n".join(collection_list)
+    
     def _optimize_consecutive_pages(self, selected, all_results, target_count=None):
         """
         Optimize selection to include consecutive pages when beneficial
@@ -3498,14 +3606,45 @@ def create_ui():
                             visible=True
                         )
         
-
-        
-
+        # Delete Documents Tab
+        with gr.Tab("🗑️ Delete Documents"):
+            with gr.Column():
+                gr.Markdown("### Delete Document Collections")
+                gr.Markdown("⚠️ **Warning**: This will permanently delete documents and their associated data from the system.")
                 
-
-
-
-
+                # Show available collections
+                gr.Markdown("#### Available Collections")
+                collections_display = gr.Textbox(
+                    label="Current Collections",
+                    interactive=False,
+                    lines=8,
+                    value="No collections available. Upload some documents first."
+                )
+                
+                # Collection selection
+                collection_dropdown = gr.Dropdown(
+                    label="Select Collection to Delete",
+                    choices=[],
+                    value=None,
+                    allow_custom_value=True,
+                    info="Select a specific collection to delete, or leave empty to delete all collections"
+                )
+                
+                # Delete options
+                with gr.Row():
+                    delete_specific_btn = gr.Button("🗑️ Delete Selected Collection", variant="secondary")
+                    delete_all_btn = gr.Button("🗑️ Delete ALL Collections", variant="stop")
+                
+                # Status output
+                delete_status = gr.Textbox(
+                    label="Deletion Status",
+                    interactive=False,
+                    lines=6
+                )
+                
+                # Refresh button
+                refresh_collections_btn = gr.Button("🔄 Refresh Collections List", variant="secondary")
+        
 
         
         # Event handlers
@@ -3522,6 +3661,46 @@ def create_ui():
             outputs=[path, images, llm_answer, cited_pages_display, csv_download, doc_download, excel_download]
         )
         
+        # Delete events
+        def refresh_collections():
+            """Refresh the collections list and dropdown"""
+            collections_text = app.get_available_collections()
+            collection_choices = list(app.indexed_docs.keys()) if app.indexed_docs else []
+            return collections_text, gr.Dropdown(choices=collection_choices)
+        
+        def delete_specific_collection(collection_name):
+            """Delete a specific collection"""
+            if not collection_name or collection_name.strip() == "":
+                return "❌ Please select a collection to delete."
+            return app.delete_documents(collection_name.strip())
+        
+        def delete_all_collections():
+            """Delete all collections"""
+            return app.delete_documents()
+        
+        # Delete event handlers
+        refresh_collections_btn.click(
+            fn=refresh_collections,
+            outputs=[collections_display, collection_dropdown]
+        )
+        
+        delete_specific_btn.click(
+            fn=delete_specific_collection,
+            inputs=[collection_dropdown],
+            outputs=[delete_status]
+        )
+        
+        delete_all_btn.click(
+            fn=delete_all_collections,
+            outputs=[delete_status]
+        )
+        
+        # Initialize collections on page load
+        demo.load(
+            fn=refresh_collections,
+            outputs=[collections_display, collection_dropdown]
+        )
+