fix: codex OAuth credential pool disconnect + expired token import (#5681)
Three bugs causing OpenAI Codex sessions to fail silently: 1. Credential pool vs legacy store disconnect: hermes auth and hermes model store device_code tokens in the credential pool, but get_codex_auth_status(), resolve_codex_runtime_credentials(), and _model_flow_openai_codex() only read from the legacy provider state. Fresh pool tokens were invisible to the auth status checks and model selection flow. 2. _import_codex_cli_tokens() imported expired tokens from ~/.codex/ without checking JWT expiry. Combined with _login_openai_codex() saying 'Login successful!' for expired credentials, users got stuck in a loop of dead tokens being recycled. 3. _login_openai_codex() accepted expired tokens from resolve_codex_runtime_credentials() without validating expiry before telling the user login succeeded. Fixes: - get_codex_auth_status() now checks credential pool first, falls back to legacy provider state - _model_flow_openai_codex() uses pool-aware auth status for token retrieval when fetching model lists - _import_codex_cli_tokens() validates JWT exp claim, rejects expired - _login_openai_codex() verifies resolved token isn't expiring before accepting existing credentials - _run_codex_stream() logs response.incomplete/failed terminal events with status and incomplete_details for diagnostics - Codex empty output recovery: captures streamed text during streaming and synthesizes a response when get_final_response() returns empty output (handles chatgpt.com backend-api edge cases)
This commit is contained in:
Teknium
63
run_agent.py
63
run_agent.py
@@ -3868,6 +3868,10 @@ class AIAgent:
|
||||
has_tool_calls = False
|
||||
first_delta_fired = False
|
||||
self._reasoning_deltas_fired = False
|
||||
# Accumulate streamed text so we can recover if get_final_response()
|
||||
# returns empty output (e.g. chatgpt.com backend-api sends
|
||||
# response.incomplete instead of response.completed).
|
||||
self._codex_streamed_text_parts: list = []
|
||||
for attempt in range(max_stream_retries + 1):
|
||||
try:
|
||||
with active_client.responses.stream(**api_kwargs) as stream:
|
||||
@@ -3887,6 +3891,7 @@ class AIAgent:
|
||||
except Exception:
|
||||
pass
|
||||
self._fire_stream_delta(delta_text)
|
||||
self._codex_streamed_text_parts.append(delta_text)
|
||||
# Track tool calls to suppress text streaming
|
||||
elif "function_call" in event_type:
|
||||
has_tool_calls = True
|
||||
@@ -3895,6 +3900,18 @@ class AIAgent:
|
||||
reasoning_text = getattr(event, "delta", "")
|
||||
if reasoning_text:
|
||||
self._fire_reasoning_delta(reasoning_text)
|
||||
# Log non-completed terminal events for diagnostics
|
||||
elif event_type in ("response.incomplete", "response.failed"):
|
||||
resp_obj = getattr(event, "response", None)
|
||||
status = getattr(resp_obj, "status", None) if resp_obj else None
|
||||
incomplete_details = getattr(resp_obj, "incomplete_details", None) if resp_obj else None
|
||||
logger.warning(
|
||||
"Codex Responses stream received terminal event %s "
|
||||
"(status=%s, incomplete_details=%s, streamed_chars=%d). %s",
|
||||
event_type, status, incomplete_details,
|
||||
sum(len(p) for p in self._codex_streamed_text_parts),
|
||||
self._client_log_context(),
|
||||
)
|
||||
return stream.get_final_response()
|
||||
except (_httpx.RemoteProtocolError, _httpx.ReadTimeout, _httpx.ConnectError, ConnectionError) as exc:
|
||||
if attempt < max_stream_retries:
|
||||
@@ -7366,8 +7383,50 @@ class AIAgent:
|
||||
response_invalid = True
|
||||
error_details.append("response.output is not a list")
|
||||
elif len(output_items) == 0:
|
||||
response_invalid = True
|
||||
error_details.append("response.output is empty")
|
||||
# Log diagnostics for empty output
|
||||
_resp_status = getattr(response, "status", None)
|
||||
_resp_incomplete = getattr(response, "incomplete_details", None)
|
||||
_streamed_parts = getattr(self, "_codex_streamed_text_parts", [])
|
||||
_streamed_text = "".join(_streamed_parts).strip() if _streamed_parts else ""
|
||||
logging.warning(
|
||||
"Codex response.output is empty "
|
||||
"(status=%s, incomplete_details=%s, streamed_chars=%d, "
|
||||
"output_text=%r, model=%s). %s",
|
||||
_resp_status, _resp_incomplete, len(_streamed_text),
|
||||
getattr(response, "output_text", None),
|
||||
getattr(response, "model", None),
|
||||
f"api_mode={self.api_mode} provider={self.provider}",
|
||||
)
|
||||
# Recovery: if we streamed text but the final response
|
||||
# lost it (e.g. response.incomplete from chatgpt backend-api),
|
||||
# synthesize a minimal response so the user gets the answer
|
||||
# the model already delivered.
|
||||
if _streamed_text:
|
||||
logging.info(
|
||||
"Recovering %d chars of streamed text as response "
|
||||
"(status was %s).", len(_streamed_text), _resp_status,
|
||||
)
|
||||
response = SimpleNamespace(
|
||||
output=[SimpleNamespace(
|
||||
type="message",
|
||||
role="assistant",
|
||||
status="completed",
|
||||
content=[SimpleNamespace(
|
||||
type="output_text",
|
||||
text=_streamed_text,
|
||||
)],
|
||||
)],
|
||||
status=_resp_status or "completed",
|
||||
model=getattr(response, "model", self.model),
|
||||
usage=getattr(response, "usage", None),
|
||||
id=getattr(response, "id", None),
|
||||
output_text=_streamed_text,
|
||||
)
|
||||
# Clear the accumulated parts so we don't double-recover
|
||||
self._codex_streamed_text_parts = []
|
||||
else:
|
||||
response_invalid = True
|
||||
error_details.append("response.output is empty")
|
||||
elif self.api_mode == "anthropic_messages":
|
||||
content_blocks = getattr(response, "content", None) if response is not None else None
|
||||
if response is None:
|
||||
|
||||
Reference in New Issue
Block a user