Update app.py

app.py

@@ -28,7 +28,200 @@ pc = Pinecone(api_key=pinecone_api_key)
 UPLOAD_FOLDER = "uploads"
 os.makedirs(UPLOAD_FOLDER, exist_ok=True)
 
-def list_uploaded_files(progress=gr.Progress()):
+def parse_pinecone_timestamp(iso_string):
+    """
+    Parses an ISO 8601 string from Pinecone, handling nanosecond precision.
+    
+    Args:
+        iso_string (str): The ISO-formatted timestamp string.
+        
+    Returns:
+        datetime: The parsed datetime object.
+    """
+    if not isinstance(iso_string, str) or not iso_string:
+        return datetime.min # Return a very old date for sorting purposes
+
+    # Handle ISO strings ending with 'Z' (convert to +00:00)
+    if iso_string.endswith('Z'):
+        iso_string = iso_string[:-1] + '+00:00'
+    
+    # Find the decimal point for fractional seconds
+    decimal_point = iso_string.find('.')
+    
+    if decimal_point != -1:
+        # Find where the timezone info starts (+ or -)
+        tz_start = max(iso_string.rfind('+'), iso_string.rfind('-'))
+        
+        if tz_start > decimal_point:
+            # Extract the fractional seconds part
+            fractional_part = iso_string[decimal_point+1:tz_start]
+            
+            # Truncate to 6 digits (microseconds) if longer
+            if len(fractional_part) > 6:
+                fractional_part = fractional_part[:6]
+            
+            # Reconstruct the ISO string with truncated fractional seconds
+            iso_string = iso_string[:decimal_point+1] + fractional_part + iso_string[tz_start:]
+    
+    return datetime.fromisoformat(iso_string)
+
+def get_all_files():
+    """Get all files from Pinecone Assistant and sort them"""
+    try:
+        assistant_name = os.getenv("PINECONE_ASSISTANT_NAME", "gstminutes")
+        assistant = pc.assistant.Assistant(assistant_name=assistant_name)
+        
+        # List all files in the assistant
+        files_response = assistant.list_files()
+        
+        # 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:
+            return []
+        
+        # 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
+        )
+        
+        return sorted_files
+        
+    except Exception as e:
+        return []
+
+def list_uploaded_files_paginated(page_num=0, progress=gr.Progress()):
+    """List files with pagination - 10 files per page"""
+    try:
+        progress(0.1, desc="🔍 Getting files...")
+        
+        # Get all files
+        all_files = get_all_files()
+        
+        if not all_files:
+            progress(1.0, desc="✅ Complete - No files found")
+            return "📋 **No files found in Pinecone Assistant**", "", "No files available", gr.update(visible=False), gr.update(visible=False)
+        
+        progress(0.5, desc="📊 Processing page...")
+        
+        # Pagination settings
+        files_per_page = 10
+        start_idx = page_num * files_per_page
+        end_idx = start_idx + files_per_page
+        
+        # Get files for current page
+        page_files = all_files[start_idx:end_idx]
+        total_files = len(all_files)
+        total_pages = (total_files + files_per_page - 1) // files_per_page
+        
+        # Create summary
+        summary = f"📊 **Files Summary (Page {page_num + 1} of {total_pages})**\n\n"
+        summary += f"📁 **Total files:** {total_files}\n"
+        summary += f"📋 **Showing:** {start_idx + 1}-{min(end_idx, total_files)} of {total_files}\n"
+        summary += f"🕐 **Last updated:** {datetime.now().strftime('%Y-%m-%d %H:%M:%S')}\n\n"
+        
+        # Show current page files in summary
+        if page_files:
+            summary += f"🆕 **Files on this page:**\n"
+            for i, file_obj in enumerate(page_files, 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"{start_idx + i}. {file_name} ({created_formatted})\n"
+            summary += "\n"
+        
+        # Create detailed file list for current page only
+        detailed_info = f"## 📋 **File Details - Page {page_num + 1}**\n\n"
+        
+        progress(0.8, desc="📝 Formatting page results...")
+        
+        for i, file_obj in enumerate(page_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)
+                
+                global_index = start_idx + i
+                detailed_info += f"### {global_index}. 📄 **{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"
+                detailed_info += "\n---\n\n"
+                
+            except Exception as file_error:
+                detailed_info += f"### {start_idx + i}. ❌ **Error processing file**\n"
+                detailed_info += f"- **Error:** {str(file_error)}\n\n---\n\n"
+        
+        # Pagination info
+        pagination_info = f"📄 Page {page_num + 1} of {total_pages} | Total: {total_files} files"
+        
+        # Show/hide next/prev buttons
+        show_prev = page_num > 0
+        show_next = page_num < total_pages - 1
+        
+        progress(1.0, desc="✅ Page loaded successfully!")
+        
+        return summary, detailed_info, pagination_info, gr.update(visible=show_prev), gr.update(visible=show_next)
+        
+    except Exception as e:
+        error_msg = f"❌ **Error retrieving file list:** {str(e)}"
+        return error_msg, "", "Error", gr.update(visible=False), gr.update(visible=False)
+
+def load_next_page(current_page_info):
+    """Load next page of files"""
+    try:
+        # Extract current page number from pagination info
+        current_page = int(current_page_info.split("Page ")[1].split(" of")[0]) - 1
+        return list_uploaded_files_paginated(current_page + 1)
+    except:
+        return list_uploaded_files_paginated(0)
+
+def load_prev_page(current_page_info):
+    """Load previous page of files"""
+    try:
+        # Extract current page number from pagination info
+        current_page = int(current_page_info.split("Page ")[1].split(" of")[0]) - 1
+        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...")
@@ -44,27 +237,55 @@ def list_uploaded_files(progress=gr.Progress()):
         progress(0.7, desc="📊 Processing file information...")
         time.sleep(0.3)
         
-        if not files_response or not hasattr(files_response, 'files') or not files_response.files:
+        # 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**", ""
-        
-        files_list = files_response.files
         total_files = len(files_list)
         
-        # Sort files by creation time (most recent first)
-        sorted_files = sorted(files_list, key=lambda x: getattr(x, 'created_on', ''), reverse=True)
+        # 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
+        # Create summary focusing on newly uploaded files
         summary = f"📊 **Files Summary**\n\n"
         summary += f"📁 **Total files:** {total_files}\n"
-        summary += f"🕐 **Last updated:** {datetime.now().strftime('%Y-%m-%d %H:%M:%S')}\n\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
-        detailed_info = "## 📋 **File Details**\n\n"
+        # Create detailed file list (only last 10 files)
+        detailed_info = "## 📋 **Latest 10 Files**\n\n"
         
-        for i, file_obj in enumerate(sorted_files, 1):
+        for i, file_obj in enumerate(recent_files, 1):
             try:
                 # Get basic file information
                 file_name = getattr(file_obj, 'name', 'Unknown')
@@ -84,15 +305,15 @@ def list_uploaded_files(progress=gr.Progress()):
                 # Format timestamps
                 try:
                     if created_on != 'Unknown':
-                        created_formatted = datetime.fromisoformat(created_on.replace('Z', '+00:00')).strftime('%Y-%m-%d %H:%M:%S UTC')
+                        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 = datetime.fromisoformat(updated_on.replace('Z', '+00:00')).strftime('%Y-%m-%d %H:%M:%S UTC')
+                        updated_formatted = parse_pinecone_timestamp(updated_on).strftime('%Y-%m-%d %H:%M:%S UTC')
                     else:
                         updated_formatted = created_formatted
-                except:
+                except Exception as e:
                     created_formatted = str(created_on)
                     updated_formatted = str(updated_on)
                 
@@ -399,7 +620,8 @@ with gr.Blocks(
                 with gr.Column():
                     results_output = gr.Markdown(
                         label="📋 Detailed Results",
-                        value=""
+                        value="",
+                        max_height=400  # Add height limit for scrolling
                     )
         
         # Tab 2: View Uploaded Files
@@ -409,25 +631,37 @@ with gr.Blocks(
             
             with gr.Row():
                 refresh_btn = gr.Button(
-                    "🔄 Refresh File List",
+                    "� Fetch Files",
                     variant="primary",
                     size="lg"
                 )
-                auto_refresh_btn = gr.Button(
-                    "📋 Load Files on Startup",
-                    variant="secondary",
-                    size="lg"
-                )
             
             # File list status
             with gr.Row():
                 file_list_status = gr.Markdown(
-                    value="🟡 **Click 'Refresh File List' to load uploaded files**",
+                    value="🟡 **Click 'Fetch Files' to load uploaded files**",
                     visible=True
                 )
             
             gr.Markdown("---")
             
+            # Pagination controls
+            with gr.Row():
+                prev_btn = gr.Button(
+                    "⬅️ Previous 10",
+                    variant="secondary",
+                    visible=False
+                )
+                pagination_info = gr.Markdown(
+                    value="📄 Page 1 of 1 | Total: 0 files",
+                    elem_classes=["pagination-info"]
+                )
+                next_btn = gr.Button(
+                    "Next 10 ➡️",
+                    variant="secondary", 
+                    visible=False
+                )
+            
             # File list results
             with gr.Row():
                 with gr.Column(scale=1):
@@ -440,7 +674,8 @@ with gr.Blocks(
                 with gr.Column():
                     file_details = gr.Markdown(
                         label="📋 File Details",
-                        value="*Click refresh to load file details...*"
+                        value="*Click refresh to load file details...*",
+                        max_height=600  # Add height limit for scrolling
                     )
     
     # Event handlers for Upload tab
@@ -469,22 +704,28 @@ with gr.Blocks(
     
     # Event handlers for View Files tab
     
-    # Refresh file list
+    # Fetch files - load page 1
     refresh_btn.click(
         fn=refresh_file_list,
         outputs=[file_list_status]
     ).then(
-        fn=list_uploaded_files,
-        outputs=[file_summary, file_details]
+        fn=list_uploaded_files_paginated,
+        inputs=[],
+        outputs=[file_summary, file_details, pagination_info, prev_btn, next_btn]
     )
     
-    # Auto load files on startup
-    auto_refresh_btn.click(
-        fn=refresh_file_list,
-        outputs=[file_list_status]
-    ).then(
-        fn=list_uploaded_files,
-        outputs=[file_summary, file_details]
+    # Next page button
+    next_btn.click(
+        fn=load_next_page,
+        inputs=[pagination_info],
+        outputs=[file_summary, file_details, pagination_info, prev_btn, next_btn]
+    )
+    
+    # Previous page button  
+    prev_btn.click(
+        fn=load_prev_page,
+        inputs=[pagination_info],
+        outputs=[file_summary, file_details, pagination_info, prev_btn, next_btn]
     )
     
     # Footer