Update app.py

app.py

@@ -10,7 +10,6 @@ HF_BUCKET = "hf://buckets/Krixter/storage"
 def get_fs():
     """Initialize the Hugging Face File System. Uses a token if available, otherwise runs anonymously."""
     token = os.environ.get("HF_TOKEN")
-    # If there's no token, HfFileSystem will just run in public/anonymous mode
     return HfFileSystem(token=token)
 
 def format_size(size_bytes):
@@ -19,40 +18,80 @@ def format_size(size_bytes):
         if size_bytes < 1024.0:
             return f"{size_bytes:.2f} {unit}"
         size_bytes /= 1024.0
+    return f"{size_bytes:.2f} PB" # Fallback
 
-def list_files():
-    """Fetch files from the bucket and return a formatted DataFrame."""
+def list_files(sort_by="Date (Newest)"):
+    """Fetch files from the bucket, extract accurate metadata, and sort."""
     try:
         fs = get_fs()
-        # Fetch file list with details (size, modified date)
         files = fs.ls(HF_BUCKET, detail=True)
         
         data = []
         for f in files:
-            if f["type"] == "file":
-                name = os.path.basename(f["name"])
-                size_str = format_size(f["size"])
-                last_modified = f.get("last_modified", "Unknown")
+            # Skip directories
+            if f.get("type") == "directory":
+                continue
                 
-                data.append({
-                    "File Name": name,
-                    "Last Modified": last_modified,
-                    "Size": size_str
-                })
-        
+            name = os.path.basename(f["name"])
+            raw_size = f.get("size", 0)
+            size_str = format_size(raw_size)
+            
+            # --- AGGRESSIVE DATE EXTRACTION ---
+            # Different storage backends (like S3/Xet) use different dictionary keys. 
+            # We check all possible variations to ensure we catch the Upload Date.
+            raw_date = None
+            date_keys = ["last_modified", "LastModified", "updated", "updatedAt", "client_modified", "created_at"]
+            
+            for k in date_keys:
+                if k in f and f[k] is not None:
+                    raw_date = f[k]
+                    break
+            
+            # If missing, attempt to ping the file directly for deeper metadata
+            if raw_date is None:
+                try:
+                    info = fs.info(f["name"])
+                    for k in date_keys:
+                        if k in info and info[k] is not None:
+                            raw_date = info[k]
+                            break
+                except Exception:
+                    pass
+            
+            data.append({
+                "File Name": name,
+                "Size": size_str,
+                "_raw_date": raw_date,
+                "_raw_size": raw_size
+            })
+            
         df = pd.DataFrame(data)
         if not df.empty:
-            # Safely convert to datetime using errors='coerce' to turn "Unknown" into NaT (Not a Time)
-            dt_col = pd.to_datetime(df["Last Modified"], errors='coerce')
+            # Safely parse dates, converting valid dates to datetime objects and invalid ones to NaT
+            dt_col = pd.to_datetime(df["_raw_date"], errors='coerce').dt.tz_localize(None)
+            df["_raw_date"] = dt_col
             
-            # Remove timezone info for clean sorting/display
-            dt_col = dt_col.dt.tz_localize(None)
-            
-            # Convert back to a clean string format, replacing NaT back with "Unknown" for the UI
+            # Format display column, filling NaT back to "Unknown"
             df["Last Modified"] = dt_col.dt.strftime('%Y-%m-%d %H:%M:%S').fillna("Unknown")
             
-            # Default sort by File Name (alphabetical)
-            df = df.sort_values(by="File Name", ascending=True)
+            # --- EXPLICIT SORTING ---
+            # We sort against the raw hidden values, NOT the formatted strings
+            if sort_by == "Name (A-Z)":
+                df = df.sort_values(by="File Name", ascending=True)
+            elif sort_by == "Name (Z-A)":
+                df = df.sort_values(by="File Name", ascending=False)
+            elif sort_by == "Date (Newest)":
+                df = df.sort_values(by="_raw_date", ascending=False, na_position='last')
+            elif sort_by == "Date (Oldest)":
+                df = df.sort_values(by="_raw_date", ascending=True, na_position='last')
+            elif sort_by == "Size (Largest)":
+                df = df.sort_values(by="_raw_size", ascending=False)
+            elif sort_by == "Size (Smallest)":
+                df = df.sort_values(by="_raw_size", ascending=True)
+                
+            # Clean up raw columns and enforce column order
+            df = df.drop(columns=["_raw_date", "_raw_size"])
+            df = df[["File Name", "Last Modified", "Size"]]
             
         file_names = df["File Name"].tolist() if not df.empty else []
         return df, gr.update(choices=file_names)
@@ -70,11 +109,9 @@ def download_file(filename):
         fs = get_fs()
         src_path = f"{HF_BUCKET}/{filename}"
         
-        # Create a temporary directory to hold the downloaded file
         temp_dir = tempfile.mkdtemp()
         dest_path = os.path.join(temp_dir, filename)
         
-        # Pull the file down from the bucket
         print(f"📥 Downloading {filename}...")
         fs.get(src_path, dest_path)
         print("✅ Download complete!")
@@ -85,14 +122,21 @@ def download_file(filename):
         return None
 
 # --- UI LAYOUT ---
-# Theme moved from here to the launch command at the bottom
 with gr.Blocks() as app:
     gr.Markdown("# 🗄️ AudioNovel Bucket Explorer")
-    gr.Markdown("View, sort, and download your processed audio files. **Click the column headers below to sort by Name, Date, or Size.**")
+    gr.Markdown("View, sort, and download your processed audio files. Use the dropdown to sort your files perfectly.")
     
     with gr.Row():
-        refresh_btn = gr.Button("🔄 Refresh Bucket contents", variant="secondary")
-        
+        with gr.Column(scale=2):
+            refresh_btn = gr.Button("🔄 Refresh Bucket contents", variant="secondary")
+        with gr.Column(scale=1):
+            sort_dropdown = gr.Dropdown(
+                label="Sort Files By",
+                choices=["Name (A-Z)", "Name (Z-A)", "Date (Newest)", "Date (Oldest)", "Size (Largest)", "Size (Smallest)"],
+                value="Date (Newest)",
+                interactive=True
+            )
+            
     # Interactive table
     file_table = gr.Dataframe(
         headers=["File Name", "Last Modified", "Size"],
@@ -113,12 +157,11 @@ with gr.Blocks() as app:
     
     # --- EVENT TRIGGERS ---
     # Trigger refresh
-    refresh_btn.click(fn=list_files, inputs=[], outputs=[file_table, file_dropdown])
-    # Trigger download
-    download_btn.click(fn=download_file, inputs=[file_dropdown], outputs=[file_output])
+    refresh_btn.click(fn=list_files, inputs=[sort_dropdown], outputs=[file_table, file_dropdown])
+    # Trigger sort instantly when dropdown changes
+    sort_dropdown.change(fn=list_files, inputs=[sort_dropdown], outputs=[file_table, file_dropdown])
     # Load data automatically on startup
-    app.load(fn=list_files, inputs=[], outputs=[file_table, file_dropdown])
+    app.load(fn=list_files, inputs=[sort_dropdown], outputs=[file_table, file_dropdown])
 
 if __name__ == "__main__":
-    # Theme is applied here now to avoid the Gradio 6.0 warning
     app.launch(theme=gr.themes.Soft(primary_hue="indigo"))