Update app.py

app.py

@@ -95,8 +95,119 @@ def get_all_files():
     except Exception as e:
         return []
 
+def get_file_choices():
+    """Get file choices for the dropdown - returns list of (title, file_id) tuples"""
+    try:
+        all_files = get_all_files()
+        if not all_files:
+            return []
+        
+        choices = []
+        for file_obj in all_files:
+            file_name = getattr(file_obj, 'name', 'Unknown File')
+            file_id = getattr(file_obj, 'id', 'unknown')
+            created_on = getattr(file_obj, 'created_on', '')
+            
+            # Format timestamp for display
+            try:
+                if created_on:
+                    created_formatted = parse_pinecone_timestamp(created_on).strftime('%Y-%m-%d %H:%M')
+                    display_name = f"{file_name} (uploaded: {created_formatted})"
+                else:
+                    display_name = file_name
+            except:
+                display_name = file_name
+            
+            choices.append((display_name, file_id))
+        
+        return choices
+    except Exception as e:
+        return []
+
+def refresh_delete_dropdown():
+    """Refresh the dropdown with current files"""
+    choices = get_file_choices()
+    if not choices:
+        return gr.update(choices=[], value=None, interactive=False)
+    return gr.update(choices=choices, value=None, interactive=True)
+
+def delete_selected_files(selected_file_ids, progress=gr.Progress()):
+    """Delete multiple selected files by their IDs"""
+    if not selected_file_ids:
+        return "❌ **Error:** No files selected for deletion", ""
+    
+    try:
+        progress(0.1, desc="🔧 Initializing Pinecone Assistant...")
+        assistant_name = os.getenv("PINECONE_ASSISTANT_NAME", "gstminutes")
+        assistant = pc.assistant.Assistant(assistant_name=assistant_name)
+        
+        # Get current files to map IDs to names
+        all_files = get_all_files()
+        file_id_to_name = {getattr(f, 'id', ''): getattr(f, 'name', 'Unknown') for f in all_files}
+        
+        total_files = len(selected_file_ids)
+        deleted_files = []
+        failed_files = []
+        
+        progress(0.2, desc=f"🗑️ Starting deletion of {total_files} files...")
+        
+        for i, file_id in enumerate(selected_file_ids):
+            try:
+                file_name = file_id_to_name.get(file_id, f"File ID: {file_id}")
+                progress((0.2 + (i / total_files) * 0.7), desc=f"🗑️ Deleting: {file_name}...")
+                
+                # Delete the file
+                response = assistant.delete_file(file_id=file_id)
+                deleted_files.append({
+                    'name': file_name,
+                    'id': file_id,
+                    'status': 'success'
+                })
+                
+                time.sleep(0.2)  # Small delay between deletions
+                
+            except Exception as delete_error:
+                failed_files.append({
+                    'name': file_id_to_name.get(file_id, f"File ID: {file_id}"),
+                    'id': file_id,
+                    'error': str(delete_error)
+                })
+        
+        progress(1.0, desc="✅ Deletion process completed!")
+        
+        # Format results
+        success_count = len(deleted_files)
+        error_count = len(failed_files)
+        
+        status_message = f"📊 **Deletion Complete**\n\n"
+        status_message += f"✅ **Successfully deleted:** {success_count} files\n"
+        status_message += f"❌ **Failed to delete:** {error_count} files\n"
+        status_message += f"📁 **Total processed:** {total_files} files\n\n"
+        
+        # Detailed results
+        detailed_results = "## 🗑️ **Deletion Results**\n\n"
+        
+        if deleted_files:
+            detailed_results += "### ✅ **Successfully Deleted Files:**\n"
+            for file_info in deleted_files:
+                detailed_results += f"- **{file_info['name']}** (`{file_info['id']}`)\n"
+            detailed_results += "\n"
+        
+        if failed_files:
+            detailed_results += "### ❌ **Failed Deletions:**\n"
+            for file_info in failed_files:
+                detailed_results += f"- **{file_info['name']}** (`{file_info['id']}`)\n"
+                detailed_results += f"  - Error: {file_info['error']}\n"
+            detailed_results += "\n"
+        
+        return status_message, detailed_results
+        
+    except Exception as e:
+        error_msg = f"❌ **Critical Error during deletion:** {str(e)}"
+        return error_msg, ""
+
 def list_uploaded_files_paginated(page_num=0, progress=gr.Progress()):
-    """List files with pagination - 10 files per page"""
+    """List files with pagination - 100 files per page"""
     try:
         progress(0.1, desc="🔍 Getting files...")
         
@@ -110,7 +221,7 @@ def list_uploaded_files_paginated(page_num=0, progress=gr.Progress()):
         progress(0.5, desc="📊 Processing page...")
         
         # Pagination settings
-        files_per_page = 100  # Changed to 100
+        files_per_page = 100
         start_idx = page_num * files_per_page
         end_idx = start_idx + files_per_page
         
@@ -133,10 +244,23 @@ def list_uploaded_files_paginated(page_num=0, progress=gr.Progress()):
             try:
                 # Get only the file name/title
                 file_name = getattr(file_obj, 'name', 'Unknown File')
+                file_id = getattr(file_obj, 'id', 'Unknown ID')
+                created_on = getattr(file_obj, 'created_on', '')
+                
                 global_index = start_idx + i
                 
-                # Display only the title vertically
-                detailed_info += f"{global_index}. **{file_name}**\n\n"
+                # Format timestamp
+                try:
+                    if created_on:
+                        created_formatted = parse_pinecone_timestamp(created_on).strftime('%Y-%m-%d %H:%M')
+                    else:
+                        created_formatted = 'Unknown'
+                except:
+                    created_formatted = 'Unknown'
+                
+                # Display file info
+                detailed_info += f"{global_index}. **{file_name}**\n"
+                detailed_info += f"   📅 Uploaded: {created_formatted} | 🆔 ID: `{file_id}`\n\n"
                 
             except Exception as file_error:
                 detailed_info += f"{start_idx + i}. ❌ **Error loading file**\n\n"
@@ -173,140 +297,6 @@ def load_prev_page(current_page_info):
         return list_uploaded_files_paginated(max(0, current_page - 1))
     except:
         return list_uploaded_files_paginated(0)
-    """List all files uploaded to Pinecone Assistant with their metadata and timestamps"""
-    try:
-        progress(0.1, desc="🔍 Connecting to Pinecone Assistant...")
-        assistant_name = os.getenv("PINECONE_ASSISTANT_NAME", "gstminutes")
-        assistant = pc.assistant.Assistant(assistant_name=assistant_name)
-        
-        progress(0.3, desc="📋 Fetching file list...")
-        time.sleep(0.5)
-        
-        # List all files in the assistant
-        files_response = assistant.list_files()
-        
-        progress(0.7, desc="📊 Processing file information...")
-        time.sleep(0.3)
-        
-        # Check if files_response is directly a list or has a files attribute
-        if hasattr(files_response, 'files'):
-            files_list = files_response.files
-        else:
-            files_list = files_response
-        
-        if not files_list:
-            progress(1.0, desc="✅ Complete - No files found")
-            return "📋 **No files found in Pinecone Assistant**", ""
-        total_files = len(files_list)
-        
-        # Sort files by creation time (most recent first) using robust timestamp parsing
-        sorted_files = sorted(
-            files_list, 
-            key=lambda x: parse_pinecone_timestamp(getattr(x, 'created_on', '')), 
-            reverse=True
-        )
-        
-        # Only show last 10 files for faster loading
-        recent_files = sorted_files[:10]
-        
-        progress(0.9, desc="📝 Formatting results...")
-        
-        # Create summary focusing on newly uploaded files
-        summary = f"📊 **Files Summary**\n\n"
-        summary += f"📁 **Total files:** {total_files}\n"
-        summary += f"� **Showing:** Last 10 files (newest first)\n"
-        summary += f"�🕐 **Last updated:** {datetime.now().strftime('%Y-%m-%d %H:%M:%S')}\n\n"
-        
-        # Show latest files in summary for quick view
-        if len(recent_files) > 0:
-            summary += f"🆕 **Latest uploads:**\n"
-            for i, file_obj in enumerate(recent_files[:5], 1):
-                file_name = getattr(file_obj, 'name', 'Unknown')
-                created_on = getattr(file_obj, 'created_on', 'Unknown')
-                try:
-                    if created_on != 'Unknown':
-                        created_formatted = parse_pinecone_timestamp(created_on).strftime('%m-%d %H:%M')
-                    else:
-                        created_formatted = 'Unknown'
-                except:
-                    created_formatted = 'Unknown'
-                summary += f"{i}. {file_name} ({created_formatted})\n"
-            summary += "\n"
-        
-        # Create detailed file list (only last 10 files)
-        detailed_info = "## 📋 **Latest 10 Files**\n\n"
-        
-        for i, file_obj in enumerate(recent_files, 1):
-            try:
-                # Get basic file information
-                file_name = getattr(file_obj, 'name', 'Unknown')
-                file_id = getattr(file_obj, 'id', 'Unknown')
-                file_size = getattr(file_obj, 'size', 0)
-                created_on = getattr(file_obj, 'created_on', 'Unknown')
-                updated_on = getattr(file_obj, 'updated_on', created_on)
-                
-                # Format file size
-                if file_size > 1024 * 1024:
-                    size_str = f"{file_size / (1024 * 1024):.2f} MB"
-                elif file_size > 1024:
-                    size_str = f"{file_size / 1024:.2f} KB"
-                else:
-                    size_str = f"{file_size} bytes"
-                
-                # Format timestamps
-                try:
-                    if created_on != 'Unknown':
-                        created_formatted = parse_pinecone_timestamp(created_on).strftime('%Y-%m-%d %H:%M:%S UTC')
-                    else:
-                        created_formatted = 'Unknown'
-                        
-                    if updated_on != 'Unknown' and updated_on != created_on:
-                        updated_formatted = parse_pinecone_timestamp(updated_on).strftime('%Y-%m-%d %H:%M:%S UTC')
-                    else:
-                        updated_formatted = created_formatted
-                except Exception as e:
-                    created_formatted = str(created_on)
-                    updated_formatted = str(updated_on)
-                
-                detailed_info += f"### {i}. 📄 **{file_name}**\n"
-                detailed_info += f"- **🆔 File ID:** `{file_id}`\n"
-                detailed_info += f"- **📏 Size:** {size_str}\n"
-                detailed_info += f"- **📅 Uploaded:** {created_formatted}\n"
-                detailed_info += f"- **🔄 Last Updated:** {updated_formatted}\n"
-                
-                # Try to get metadata if available
-                try:
-                    # Get file details for metadata
-                    file_details = assistant.describe_file(file_id=file_id)
-                    if hasattr(file_details, 'metadata') and file_details.metadata:
-                        metadata = file_details.metadata
-                        detailed_info += f"- **🏷️ Metadata:**\n"
-                        
-                        if isinstance(metadata, dict):
-                            for key, value in metadata.items():
-                                if isinstance(value, list):
-                                    detailed_info += f"  - **{key.title()}:** {', '.join(map(str, value))}\n"
-                                else:
-                                    detailed_info += f"  - **{key.title()}:** {value}\n"
-                        else:
-                            detailed_info += f"  - {metadata}\n"
-                except Exception as metadata_error:
-                    detailed_info += f"- **🏷️ Metadata:** Could not retrieve metadata\n"
-                
-                detailed_info += "\n---\n\n"
-                
-            except Exception as file_error:
-                detailed_info += f"### {i}. ❌ **Error processing file**\n"
-                detailed_info += f"- **Error:** {str(file_error)}\n\n---\n\n"
-        
-        progress(1.0, desc="✅ File list retrieved successfully!")
-        time.sleep(0.3)
-        
-        return summary, detailed_info
-        
-    except Exception as e:
-        error_msg = f"❌ **Error retrieving file list:** {str(e)}"
-        return error_msg, ""
 
 def refresh_file_list():
     """Refresh the file list"""
@@ -456,6 +446,10 @@ def clear_form():
     """Clear all form fields"""
     return [None] + [""] * 30 + ["", "", "🟢 **Ready to process documents**"]
 
+def clear_delete_form():
+    """Clear delete form"""
+    return gr.update(value=[]), "", ""
+
 def start_processing():
     """Show processing started status"""
     return "🔄 **Processing documents... Please wait**"
@@ -480,6 +474,12 @@ with gr.Blocks(
         text-align: center;
         background-color: #f8f9fa;
     }
+    .delete-container {
+        border: 2px dashed #f44336;
+        border-radius: 10px;
+        padding: 20px;
+        background-color: #ffebee;
+    }
     .tab-nav {
         margin-bottom: 20px;
     }
@@ -490,12 +490,13 @@ with gr.Blocks(
         """
         # 📄 Tax Document Ingestion System
         
-        Upload and manage documents in the Pinecone Assistant for GST Minutes processing.
+        Upload, manage, and delete documents in the Pinecone Assistant for GST Minutes processing.
         
         ## 🚀 Features:
         - ✅ **Multiple file upload** - Select and upload multiple documents at once
         - 🏷️ **Metadata tagging** - Add sections, keywords, and descriptions
         - 🔄 **Batch processing** - All files processed with individual metadata
+        - 🗑️ **File deletion** - Delete multiple files by selecting from dropdown
         - 📊 **File management** - View uploaded files with timestamps and metadata
         - 📋 **Detailed reporting** - See success/failure status for each operation
         
@@ -574,15 +575,70 @@ with gr.Blocks(
                         value="",
                         max_height=400  # Add height limit for scrolling
                     )
+
+        # Tab 2: Delete Documents
+        with gr.TabItem("🗑️ Delete Documents", id="delete_tab"):
+            gr.Markdown("### 🗑️ **Delete Multiple Documents**")
+            gr.Markdown("Select multiple files from the dropdown to delete them from the Pinecone Assistant.")
+            
+            with gr.Row():
+                with gr.Column(scale=2):
+                    file_dropdown = gr.Dropdown(
+                        label="📋 Select Files to Delete (Multiple Selection)",
+                        choices=[],
+                        multiselect=True,
+                        interactive=False,
+                        elem_classes=["delete-container"]
+                    )
+                    
+                with gr.Column(scale=1):
+                    refresh_dropdown_btn = gr.Button(
+                        "🔄 Refresh File List",
+                        variant="secondary",
+                        size="lg"
+                    )
+            
+            with gr.Row():
+                with gr.Column(scale=1):
+                    delete_btn = gr.Button(
+                        "🗑️ Delete Selected Files",
+                        variant="stop",
+                        size="lg"
+                    )
+                    
+                with gr.Column(scale=1):
+                    clear_delete_btn = gr.Button(
+                        "↺ Clear Selection",
+                        variant="secondary",
+                        size="lg"
+                    )
+            
+            gr.Markdown("---")
+            
+            # Delete results section
+            with gr.Row():
+                with gr.Column():
+                    delete_status_output = gr.Markdown(
+                        label="📊 Deletion Status",
+                        value="*Select files to delete...*"
+                    )
+            
+            with gr.Row():
+                with gr.Column():
+                    delete_results_output = gr.Markdown(
+                        label="🗑️ Deletion Results",
+                        value="",
+                        max_height=400
+                    )
         
-        # Tab 2: View Uploaded Files
+        # Tab 3: View Uploaded Files
         with gr.TabItem("📋 View Uploaded Files", id="view_tab"):
             gr.Markdown("### 📋 **Uploaded Files Management**")
             gr.Markdown("View all files currently uploaded to the Pinecone Assistant with their metadata and timestamps.")
             
             with gr.Row():
                 refresh_btn = gr.Button(
-                    "� Fetch Files",
+                    "🔄 Fetch Files",
                     variant="primary",
                     size="lg"
                 )
@@ -653,6 +709,27 @@ with gr.Blocks(
         outputs=[files_input] + metadata_fields + [status_output, results_output, processing_status]
     )
     
+    # Event handlers for Delete tab
+    
+    # Refresh dropdown with current files
+    refresh_dropdown_btn.click(
+        fn=refresh_delete_dropdown,
+        outputs=[file_dropdown]
+    )
+    
+    # Delete selected files
+    delete_btn.click(
+        fn=delete_selected_files,
+        inputs=[file_dropdown],
+        outputs=[delete_status_output, delete_results_output]
+    )
+    
+    # Clear delete form
+    clear_delete_btn.click(
+        fn=clear_delete_form,
+        outputs=[file_dropdown, delete_status_output, delete_results_output]
+    )
+    
     # Event handlers for View Files tab
     
     # Fetch files - load page 1
@@ -692,10 +769,22 @@ with gr.Blocks(
         - Each file gets its own sections, keywords, and description
         - Check the results section for upload status
         
+        **Delete Documents:**
+        - Click 'Refresh File List' to load current files in dropdown
+        - Select multiple files using the dropdown (supports multi-select)
+        - Click 'Delete Selected Files' to remove them permanently
+        - View deletion results for success/failure status
+        
         **View Uploaded Files:**
-        - Click 'Refresh File List' to see all uploaded files
+        - Click 'Fetch Files' to see all uploaded files
         - View file details including upload timestamps and metadata
         - Files are sorted by most recent first
+        - Use pagination to navigate through large file lists
+        
+        ### ⚠️ **Important Notes:**
+        - File deletion is **permanent** and cannot be undone
+        - Always verify your selection before deleting files
+        - The system maps file titles to IDs internally for deletion
         
         ### 📞 **Support:**
         For issues or questions, contact the development team.