[Code scan] Propagate frame-level data modifier failures

[njzjz]

This issue comes from a Codex global scan of deepmodeling/deepmd-kit at commit 73de44b1f94471b2e3bdb6b11f57b34d7bc791bb.

Problem

DeepmdData.get_single_frame() runs a data modifier through a one-off ThreadPoolExecutor, but never calls future.result():

deepmd-kit/deepmd/utils/data.py

Lines 525 to 535 in 73de44b

	if self.modifier is not None:
	with ThreadPoolExecutor(max_workers=num_worker) as executor:
	# Apply modifier if it exists
	future = executor.submit(
	self.modifier.modify_data,
	frame_data,
	self,
	)
	if self.use_modifier_cache:
	# Cache the modified frame to avoid recomputation
	self._modified_frame_cache[index] = copy.deepcopy(frame_data)

The context manager waits for the submitted task to finish, but exceptions raised inside the future stay stored on the Future object unless result() is called. The same class calls modifiers directly in other paths, so those paths do propagate errors:

deepmd-kit/deepmd/utils/data.py

Lines 337 to 339 in 73de44b

	if self.modifier is not None:
	self.modifier.modify_data(ret, self)
	return ret

deepmd-kit/deepmd/utils/data.py

Lines 613 to 618 in 73de44b

	def _load_batch_set(self, set_name: DPPath) -> None:
	if not hasattr(self, "batch_set") or self.get_numb_set() > 1:
	self.batch_set = self._load_set(set_name)
	if self.modifier is not None:
	self.modifier.modify_data(self.batch_set, self)
	self.batch_set, _ = self._shuffle_data(self.batch_set)

Impact

Frame-level data loading can silently ignore modifier failures. If modifier.modify_data(...) raises, callers receive the unmodified frame and, when caching is enabled, that bad frame can be cached for future use.

Suggested fix

Call future.result() before leaving the modifier block, or remove the single-task executor and call self.modifier.modify_data(frame_data, self) directly. If parallelism is needed, keep exception propagation explicit before caching the modified frame.