Initial commit

app.py

@@ -0,0 +1,132 @@
+from functools import lru_cache
+
+import gradio as gr
+from huggingface_hub import HfApi, HfHubHTTPError
+
+
+DEFAULT_REPO_ID = "mlfoundations-cua-dev/human_eval"
+
+
+api = HfApi()
+
+
+@lru_cache(maxsize=16)
+def _list_repo_files(repo_id: str) -> list[str]:
+    """Return all file paths contained in a Hugging Face dataset repository."""
+    return api.list_repo_files(repo_id=repo_id, repo_type="dataset")
+
+
+def _extract_top_level(repo_id: str) -> tuple[list[str], list[str]]:
+    """Split top-level folders and files for the given repository."""
+    files = _list_repo_files(repo_id)
+    top_level_dirs = sorted({path.split("/", 1)[0] for path in files if "/" in path})
+    top_level_files = sorted(path for path in files if "/" not in path)
+    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."""
+    if not directory:
+        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]
+
+    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 []),
+    }
+
+
+def refresh_repo(repo_id: str):
+    try:
+        top_dirs, top_files = _extract_top_level(repo_id)
+    except HfHubHTTPError as error:
+        return (
+            gr.Dropdown.update(choices=[], value=None, interactive=False),
+            gr.Markdown.update(value=f"❌ Unable to load repo `{repo_id}`: {error}"),
+            {}
+        )
+    except Exception as error:  # pragma: no cover - network and auth edge cases
+        return (
+            gr.Dropdown.update(choices=[], value=None, interactive=False),
+            gr.Markdown.update(value=f"❌ Unexpected error loading `{repo_id}`: {error}"),
+            {}
+        )
+
+    status_lines = [
+        f"✅ Loaded `{repo_id}`",
+        f"• Top-level folders: {len(top_dirs)}",
+    ]
+
+    if top_files:
+        status_lines.append(f"• Loose files at root: {len(top_files)}")
+    if not top_dirs:
+        status_lines.append("• No sub-folders found at root.")
+
+    dropdown_value = top_dirs[0] if top_dirs else None
+    dropdown_update = gr.Dropdown.update(
+        choices=top_dirs,
+        value=dropdown_value,
+        interactive=bool(top_dirs),
+        label="Top-level folders",
+        info="Choose a folder to explore"
+    )
+
+    folder_summary = _summarize_directory(repo_id, dropdown_value) if dropdown_value else {}
+
+    return dropdown_update, gr.Markdown.update(value="\n".join(status_lines)), folder_summary
+
+
+def update_directory(repo_id: str, directory: str):
+    return _summarize_directory(repo_id, directory)
+
+
+with gr.Blocks(title="HF Dataset Explorer") as demo:
+    gr.Markdown(
+        """# Hugging Face Dataset Explorer
+
+Provide a dataset repository ID (e.g. `org/dataset`) to list its top-level folders."""
+    )
+
+    with gr.Row():
+        repo_id_input = gr.Textbox(
+            value=DEFAULT_REPO_ID,
+            label="Dataset repo ID",
+            placeholder="owner/dataset",
+            info="Any public dataset on the Hugging Face Hub"
+        )
+        reload_button = gr.Button("Load repo", variant="primary")
+
+    status_display = gr.Markdown()
+    folder_dropdown = gr.Dropdown(label="Top-level folders", interactive=False)
+    folder_details = gr.JSON(label="Folder summary")
+
+    reload_button.click(
+        refresh_repo,
+        inputs=repo_id_input,
+        outputs=[folder_dropdown, status_display, folder_details],
+    )
+
+    folder_dropdown.change(
+        update_directory,
+        inputs=[repo_id_input, folder_dropdown],
+        outputs=folder_details,
+    )
+
+    demo.load(
+        refresh_repo,
+        inputs=repo_id_input,
+        outputs=[folder_dropdown, status_display, folder_details],
+    )
+
+if __name__ == "__main__":
+    demo.launch()