Should httpx.ReadTimeout be captured by pydantic-ai?

[yf-yang]

Question

When calling streaming model, due to network issue there could be ReadTimeout error. I am just wondering if this error should also be handled by pydantic-ai (so this is a bug) or it should be handled by developers.

Check the trace:

File "/workspaces/project/.venv/lib/python3.12/site-packages/pydantic_ai/models/__init__.py", line 579, in iterator_with_final_event
  async for event in iterator:
File "/workspaces/project/.venv/lib/python3.12/site-packages/pydantic_ai/models/openai.py", line 1432, in _get_event_iterator
  async for chunk in self._response:
File "/workspaces/project/.venv/lib/python3.12/site-packages/pydantic_ai/_utils.py", line 307, in __anext__
  return await self._source_iter.__anext__()
          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/workspaces/project/.venv/lib/python3.12/site-packages/openai/_streaming.py", line 147, in __aiter__
  async for item in self._iterator:
File "/workspaces/project/.venv/lib/python3.12/site-packages/openai/_streaming.py", line 160, in __stream__
  async for sse in iterator:
File "/workspaces/project/.venv/lib/python3.12/site-packages/openai/_streaming.py", line 151, in _iter_events
  async for sse in self._decoder.aiter_bytes(self.response.aiter_bytes()):
File "/workspaces/project/.venv/lib/python3.12/site-packages/openai/_streaming.py", line 302, in aiter_bytes
  async for chunk in self._aiter_chunks(iterator):
File "/workspaces/project/.venv/lib/python3.12/site-packages/openai/_streaming.py", line 313, in _aiter_chunks
  async for chunk in iterator:
File "/workspaces/project/.venv/lib/python3.12/site-packages/httpx/_models.py", line 997, in aiter_bytes
  async for raw_bytes in self.aiter_raw():
File "/workspaces/project/.venv/lib/python3.12/site-packages/httpx/_models.py", line 1055, in aiter_raw
  async for raw_stream_bytes in self.stream:
File "/workspaces/project/.venv/lib/python3.12/site-packages/httpx/_client.py", line 176, in __aiter__
  async for chunk in self._stream:
File "/workspaces/project/.venv/lib/python3.12/site-packages/httpx/_transports/default.py", line 270, in __aiter__
  with map_httpcore_exceptions():
File "/usr/lib/python3.12/contextlib.py", line 158, in __exit__
  self.gen.throw(value)
File "/workspaces/project/.venv/lib/python3.12/site-packages/httpx/_transports/default.py", line 118, in map_httpcore_exceptions
  raise mapped_exc(message) from exc
httpx.ReadTimeout

Additional Context

No response

[DouweM]

@yf-yang Ideally this would be handled by the OpenAI SDK, which already does this for non-streaming requests by automatically retrying and finally wrapping the timeout exception in their own APITimeoutError:

https://github.com/openai/openai-python/blob/4756247cee3d9548397b26a29109e76cc9522379/src/openai/_base_client.py#L987-L1000

So they really should do the same for streaming requests, although I suppose automatically retrying wouldn't really work as you'd get duplicate events.

Either way, for all of the models including OpenAI, Pydantic AI only currently handles HTTP status code exceptions from APIs, so not timeouts:

pydantic-ai/pydantic_ai_slim/pydantic_ai/models/openai.py

Lines 489 to 491 in ff0fbe3

	except APIStatusError as e:
	if (status_code := e.status_code) >= 400:
	raise ModelHTTPError(status_code=status_code, model_name=self.model_name, body=e.body) from e

You can implement your own retry behavior for specific network errors as documented here: https://ai.pydantic.dev/retries/#network-error-handling

As we don't handle timeouts from any other APIs and we have this retries functionality, I'm going to treat this as not-a-bug

[yf-yang]

Oh I haven't read that doc before. That sounds elegant