submit+app: toast-based feedback (validating / queued / refreshed)

Bundle 1+2 C9. Replaces the static markdown reply on the Submit
tab with gr.Info / gr.Error toasts, and adds a refresh toast on the
manual Refresh button (the Timer auto-refresh stays silent - a
toast every 10s would be noise).

submit.py:
- handle_submit no longer returns a markdown string. It side-
effects: gr.Info("Validating submission...") at the top of the
validation stage, gr.Info("Submission <id> queued ...") after
the row + zip are on the Hub and the worker has been spawned.
Every rejection path raises gr.Error so the user gets a red
toast and the handler aborts; the outer try/finally still runs
cleanup. Internal _ValidationError + _HubWriteError sentinels
preserved.
- _validate_form returns a plain message (no `**Error:**` markdown
wrap; the caller wraps into gr.Error).
- All four rejection paths covered:
1. Form (no zip attached).
2. Validation gate (extract / meta / fixtures / steps).
3. Dedup gate (sha256 match on existing row).
4. Hub write failure.

app.py:
- Submit tab drops `submit_out = gr.Markdown()`; submit_btn.click
wires straight to handle_submit with no outputs.
- New `_refresh_leaderboard_with_toast()` wraps `load_leaderboard_
split` for the manual Refresh button click; the existing Timer
binding still calls load_leaderboard_split directly so it
doesn't toast.

Verification (autonomous):

- 22/22 unit tests still green.
- Local boot probe: GET /config carries no leftover markdown
strings ("**Submission rejected.**", "**Queued.**", "**Error:**",
"please attach a submission zip"). All earlier-bundle features
remain (Citation + Validation Guidelines accordions, Download
CSV button, downloadbutton component).

Post-push live probe runs next.

app.py

@@ -147,6 +147,17 @@ def _build_report_iframe(html_bytes: bytes) -> str:
     )
 
 
+def _refresh_leaderboard_with_toast():
+    """Manual Refresh button handler: toast + fresh DataFrames.
+
+    The Timer auto-refresh wires straight to ``load_leaderboard_split``
+    so it stays silent (a toast every 10s would be noise). Only the
+    explicit click goes through this wrapper.
+    """
+    gr.Info("Leaderboard refreshed.")
+    return load_leaderboard_split()
+
+
 def _format_detail_and_report(
     df: pd.DataFrame | None, evt: gr.SelectData,
 ) -> tuple[str, str]:
@@ -293,7 +304,7 @@ with gr.Blocks(title="CADGenBench Leaderboard", theme=gr.themes.Soft()) as block
                 label="Download CSV", size="sm",
             )
         refresh_btn.click(
-            fn=load_leaderboard_split,
+            fn=_refresh_leaderboard_with_toast,
             outputs=[validated_view, unvalidated_view],
         )
         download_btn.click(fn=build_combined_csv, outputs=download_btn)
@@ -350,12 +361,9 @@ to publish the resulting row on the public leaderboard.
         )
         zip_in = gr.File(label="Submission ZIP", file_types=[".zip"])
         submit_btn = gr.Button("Submit", variant="primary")
-        submit_out = gr.Markdown()
-        submit_btn.click(
-            fn=handle_submit,
-            inputs=[zip_in],
-            outputs=submit_out,
-        )
+        # No static markdown output: handle_submit surfaces every
+        # status update via gr.Info / gr.Error toasts.
+        submit_btn.click(fn=handle_submit, inputs=[zip_in])
 
     with gr.Tab("About"):
         gr.Markdown(ABOUT_MD)

submit.py

@@ -77,6 +77,7 @@ from pathlib import Path
 from typing import Any
 
 import cadgenbench
+import gradio as gr
 from cadgenbench.common.paths import data_inputs_dir
 from cadgenbench.common.validity import parse_step
 from huggingface_hub import HfApi
@@ -130,21 +131,28 @@ class _HubWriteError(Exception):
     """Raised when a Hub upload fails after validation succeeded."""
 
 
-def handle_submit(zip_file) -> str:
-    """Validate a submission upload and return a markdown UI message.
+def handle_submit(zip_file) -> None:
+    """Validate a submission upload; surface progress + outcome via toasts.
 
-    Returns one of:
-      - An error string starting with ``**Error:**`` for form-level
-        rejects (no zip attached).
-      - A rejection string starting with ``**Submission rejected.**``
-        for any of the deeper validation gates or Hub-write failures.
-      - A success string starting with ``**Queued.**`` carrying the
-        minted ``submission_id`` once the pending row + zip are on
-        the Hub.
+    Side-effect-only (returns ``None``): every status update lands in
+    a ``gr.Info`` / ``gr.Error`` toast instead of a static markdown
+    output below the button. Rejection paths raise ``gr.Error``,
+    which Gradio surfaces as a red toast and aborts the handler; the
+    outer ``try/finally`` still runs to clean up the temp dir.
+
+    Toast sequence on the happy path:
+      1. ``gr.Info("Validating submission...")`` once the form-level
+         check is past.
+      2. ``gr.Info("Submission <id> queued for evaluation...")`` once
+         the pending row + zip are on the Hub and the worker has
+         been spawned.
+
+    On rejection (form-level, validation gate, dedup, or Hub write),
+    a single ``gr.Error`` toast carries the message; no second toast.
     """
     form_err = _validate_form(zip_file)
     if form_err is not None:
-        return form_err
+        raise gr.Error(form_err)
 
     zip_path = Path(zip_file.name)
 
@@ -156,13 +164,14 @@ def handle_submit(zip_file) -> str:
     run_dir = tmp / "run"
     run_dir.mkdir()
     try:
+        gr.Info("Validating submission...")
         try:
             _extract_zip(zip_path, run_dir)
             meta = _load_and_validate_meta(run_dir)
             fixture_names = _validate_fixture_set(run_dir)
             _validate_steps_parseable(run_dir, fixture_names)
         except _ValidationError as e:
-            return f"**Submission rejected.** {e}"
+            raise gr.Error(f"Submission rejected: {e}")
 
         # Dedup gate: hash the raw zip bytes and reject if an existing
         # row carries the same hash. Runs after validation so a clearly
@@ -170,10 +179,10 @@ def handle_submit(zip_file) -> str:
         zip_sha256 = _compute_sha256(zip_path)
         existing_id = _find_existing_submission_by_sha256(zip_sha256)
         if existing_id is not None:
-            return (
-                f"**Submission rejected.** This zip's contents are identical "
-                f"to an existing submission (`{existing_id}`). Resubmit only "
-                f"after changing at least one byte of the upload."
+            raise gr.Error(
+                f"This zip's contents are identical to an existing "
+                f"submission ({existing_id}). Resubmit only after changing "
+                f"at least one byte of the upload."
             )
 
         submission_id = _mint_submission_id(
@@ -186,27 +195,30 @@ def handle_submit(zip_file) -> str:
             )
             _append_pending_row(row)
         except _HubWriteError as e:
-            return f"**Submission rejected.** {e}"
+            raise gr.Error(f"Submission rejected: {e}")
 
         _spawn_worker(submission_id, tmp, run_dir)
         tmp = None  # ownership transferred; skip cleanup below
+        gr.Info(
+            f"Submission {submission_id} queued for evaluation "
+            f"({len(fixture_names)} fixtures). Evaluation typically "
+            f"takes 2-5 minutes; the row flips to completed when the "
+            f"worker finishes."
+        )
     finally:
         if tmp is not None:
             shutil.rmtree(tmp, ignore_errors=True)
 
-    return (
-        f"**Queued.** Submission `{submission_id}` has been accepted "
-        f"(submitter: `{meta['submitter_name']}`, system: "
-        f"`{meta['submission_name']}`, {len(fixture_names)} fixtures). "
-        f"Evaluation typically takes 2-5 minutes on this Space's "
-        f"`cpu-upgrade` tier; the row flips to `completed` with score "
-        f"columns populated when the worker finishes."
-    )
-
 
 def _validate_form(zip_file) -> str | None:
+    """Form-level check before any zip parsing.
+
+    Returns a plain-text rejection message (no markdown wrapping;
+    the caller wraps it into a ``gr.Error`` toast) or ``None`` when
+    the form is acceptable.
+    """
     if zip_file is None:
-        return "**Error:** please attach a submission zip."
+        return "Please attach a submission zip."
     return None