Fix

app.py

@@ -26,26 +26,18 @@ def _extract_top_level(repo_id: str) -> tuple[list[str], list[str]]:
     return top_level_dirs, top_level_files
 
 
-def _summarize_directory(repo_id: str, directory: str) -> dict:
-    """Return a lightweight summary of the contents of a top-level directory."""
+def _get_subdirectories(repo_id: str, directory: str) -> list[str]:
+    """Return the direct subdirectories of the given directory."""
     if not directory:
-        return {}
+        return []
 
     files = [path for path in _list_repo_files(repo_id) if path.startswith(f"{directory}/")]
     relative_paths = [path[len(directory) + 1 :] for path in files]
-
+    
+    # Get immediate subdirectories (first level only)
     child_dirs = sorted({rel.split("/", 1)[0] for rel in relative_paths if "/" in rel})
-    child_files = sorted(rel for rel in relative_paths if rel and "/" not in rel)
-
-    sample_files = child_files[:10]
-    has_more_files = len(child_files) > len(sample_files)
-
-    return {
-        "folder": directory,
-        "total_files": len(files),
-        "direct_subdirectories": child_dirs,
-        "sample_files": sample_files + (["..."] if has_more_files else []),
-    }
+    
+    return child_dirs
 
 
 def refresh_repo(repo_id: str):
@@ -55,17 +47,17 @@ def refresh_repo(repo_id: str):
         print(f"[refresh_repo] Hub HTTP error for {repo_id}: {error}", flush=True)
         print(traceback.format_exc(), flush=True)
         return (
+            gr.update(choices=[], value=None, interactive=False),
             gr.update(choices=[], value=None, interactive=False),
             gr.update(value=f"❌ Unable to load repo `{repo_id}`: {error}"),
-            {}
         )
     except Exception as error:  # pragma: no cover - network and auth edge cases
         print(f"[refresh_repo] Unexpected error for {repo_id}: {error}", flush=True)
         print(traceback.format_exc(), flush=True)
         return (
+            gr.update(choices=[], value=None, interactive=False),
             gr.update(choices=[], value=None, interactive=False),
             gr.update(value=f"❌ Unexpected error loading `{repo_id}`: {error}"),
-            {}
         )
 
     status_lines = [
@@ -79,26 +71,50 @@ def refresh_repo(repo_id: str):
         status_lines.append("• No sub-folders found at root.")
 
     dropdown_value = top_dirs[0] if top_dirs else None
-    dropdown_update = gr.update(
+    
+    # Get subdirectories for the first top-level folder
+    subdirs = _get_subdirectories(repo_id, dropdown_value) if dropdown_value else []
+    subdir_value = subdirs[0] if subdirs else None
+    
+    first_dropdown_update = gr.update(
         choices=top_dirs,
         value=dropdown_value,
         interactive=bool(top_dirs),
         label="Top-level folders",
         info="Choose a folder to explore"
     )
+    
+    second_dropdown_update = gr.update(
+        choices=subdirs,
+        value=subdir_value,
+        interactive=bool(subdirs),
+        label="Subdirectories",
+        info="Choose a subdirectory"
+    )
 
-    folder_summary = _summarize_directory(repo_id, dropdown_value) if dropdown_value else {}
-
-    return dropdown_update, gr.update(value="\n".join(status_lines)), folder_summary
+    return first_dropdown_update, second_dropdown_update, gr.update(value="\n".join(status_lines))
 
 
-def update_directory(repo_id: str, directory: str):
+def update_second_dropdown(repo_id: str, top_level_dir: str):
+    """Update the second dropdown when the first dropdown selection changes."""
     try:
-        return _summarize_directory(repo_id, directory)
+        if not top_level_dir:
+            return gr.update(choices=[], value=None, interactive=False)
+        
+        subdirs = _get_subdirectories(repo_id, top_level_dir)
+        subdir_value = subdirs[0] if subdirs else None
+        
+        return gr.update(
+            choices=subdirs,
+            value=subdir_value,
+            interactive=bool(subdirs),
+            label="Subdirectories",
+            info="Choose a subdirectory"
+        )
     except Exception as error:
-        print(f"[update_directory] Error for {repo_id}/{directory}: {error}", flush=True)
+        print(f"[update_second_dropdown] Error for {repo_id}/{top_level_dir}: {error}", flush=True)
         print(traceback.format_exc(), flush=True)
-        return {}
+        return gr.update(choices=[], value=None, interactive=False)
 
 
 with gr.Blocks(title="HF Dataset Explorer") as demo:
@@ -119,24 +135,24 @@ Provide a dataset repository ID (e.g. `org/dataset`) to list its top-level folde
 
     status_display = gr.Markdown()
     folder_dropdown = gr.Dropdown(label="Top-level folders", interactive=False)
-    folder_details = gr.JSON(label="Folder summary")
+    subfolder_dropdown = gr.Dropdown(label="Subdirectories", interactive=False)
 
     reload_button.click(
         refresh_repo,
         inputs=repo_id_input,
-        outputs=[folder_dropdown, status_display, folder_details],
+        outputs=[folder_dropdown, subfolder_dropdown, status_display],
     )
 
     folder_dropdown.change(
-        update_directory,
+        update_second_dropdown,
         inputs=[repo_id_input, folder_dropdown],
-        outputs=folder_details,
+        outputs=subfolder_dropdown,
     )
 
     demo.load(
         refresh_repo,
         inputs=repo_id_input,
-        outputs=[folder_dropdown, status_display, folder_details],
+        outputs=[folder_dropdown, subfolder_dropdown, status_display],
     )
 
 if __name__ == "__main__":