mirror of
https://github.com/NousResearch/hermes-agent.git
synced 2026-05-18 04:41:56 +00:00
fix(agent): retry malformed anthropic stream parser errors
This commit is contained in:
parent
53637fb17d
commit
9c304a7f56
2 changed files with 131 additions and 9 deletions
57
run_agent.py
57
run_agent.py
|
|
@ -3027,6 +3027,24 @@ class AIAgent:
|
|||
parts.append(f"{type(e).__name__}({msg})" if msg else type(e).__name__)
|
||||
return " <- ".join(parts) if parts else type(error).__name__
|
||||
|
||||
def _is_provider_stream_parse_error(self, error: BaseException) -> bool:
|
||||
"""Return True for malformed provider streaming data from SDK parsers.
|
||||
|
||||
Some Anthropic-compatible streaming providers can send a malformed
|
||||
event-stream frame. The Anthropic SDK surfaces that as a plain
|
||||
``ValueError`` such as ``expected ident at line 1 column 149``. That
|
||||
is provider wire-format trouble, not local request validation, so it
|
||||
should follow the same retry path as a truncated JSON body.
|
||||
"""
|
||||
if getattr(self, "api_mode", None) != "anthropic_messages":
|
||||
return False
|
||||
if not isinstance(error, ValueError):
|
||||
return False
|
||||
if isinstance(error, (UnicodeEncodeError, json.JSONDecodeError)):
|
||||
return False
|
||||
message = str(error).strip().lower()
|
||||
return "expected ident at line" in message
|
||||
|
||||
def _log_stream_retry(
|
||||
self,
|
||||
*,
|
||||
|
|
@ -5080,6 +5098,12 @@ class AIAgent:
|
|||
"""
|
||||
raw = str(error)
|
||||
|
||||
if (
|
||||
isinstance(error, ValueError)
|
||||
and "expected ident at line" in raw.lower()
|
||||
):
|
||||
return f"Malformed provider streaming response: {raw[:300]}"
|
||||
|
||||
# Cloudflare / proxy HTML pages: grab the <title> for a clean summary
|
||||
if "<!DOCTYPE" in raw or "<html" in raw:
|
||||
m = re.search(r"<title[^>]*>([^<]+)</title>", raw, re.IGNORECASE)
|
||||
|
|
@ -8528,6 +8552,7 @@ class AIAgent:
|
|||
_is_conn_err = isinstance(
|
||||
e, (_httpx.ConnectError, _httpx.RemoteProtocolError, ConnectionError)
|
||||
)
|
||||
_is_stream_parse_err = self._is_provider_stream_parse_error(e)
|
||||
|
||||
# If the stream died AFTER some tokens were delivered:
|
||||
# normally we don't retry (the user already saw text,
|
||||
|
|
@ -8567,7 +8592,10 @@ class AIAgent:
|
|||
for phrase in _SSE_PREVIEW_PHRASES
|
||||
)
|
||||
_is_transient = (
|
||||
_is_timeout or _is_conn_err or _is_sse_conn_err_preview
|
||||
_is_timeout
|
||||
or _is_conn_err
|
||||
or _is_sse_conn_err_preview
|
||||
or _is_stream_parse_err
|
||||
)
|
||||
_can_silent_retry = (
|
||||
_partial_tool_in_flight
|
||||
|
|
@ -8665,7 +8693,7 @@ class AIAgent:
|
|||
for phrase in _SSE_CONN_PHRASES
|
||||
)
|
||||
|
||||
if _is_timeout or _is_conn_err or _is_sse_conn_err:
|
||||
if _is_timeout or _is_conn_err or _is_sse_conn_err or _is_stream_parse_err:
|
||||
# Transient network / timeout error. Retry the
|
||||
# streaming request with a fresh connection first.
|
||||
if _stream_attempt < _max_stream_retries:
|
||||
|
|
@ -8706,12 +8734,20 @@ class AIAgent:
|
|||
mid_tool_call=False,
|
||||
diag=request_client_holder.get("diag"),
|
||||
)
|
||||
self._emit_status(
|
||||
"❌ Connection to provider failed after "
|
||||
f"{_max_stream_retries + 1} attempts. "
|
||||
"The provider may be experiencing issues — "
|
||||
"try again in a moment."
|
||||
)
|
||||
if _is_stream_parse_err:
|
||||
self._emit_status(
|
||||
"❌ Provider returned malformed streaming data after "
|
||||
f"{_max_stream_retries + 1} attempts. "
|
||||
"The provider may be experiencing issues — "
|
||||
"try again in a moment."
|
||||
)
|
||||
else:
|
||||
self._emit_status(
|
||||
"❌ Connection to provider failed after "
|
||||
f"{_max_stream_retries + 1} attempts. "
|
||||
"The provider may be experiencing issues — "
|
||||
"try again in a moment."
|
||||
)
|
||||
else:
|
||||
_err_lower = str(e).lower()
|
||||
_is_stream_unsupported = (
|
||||
|
|
@ -14509,11 +14545,16 @@ class AIAgent:
|
|||
# provider/network failure (malformed response body,
|
||||
# truncated stream, routing layer corruption), not a
|
||||
# local programming bug, and should be retried (#14782).
|
||||
# Exclude Anthropic stream parser ValueErrors for the
|
||||
# same reason: third-party Anthropic-compatible providers
|
||||
# can emit malformed event-stream frames that SDK parsers
|
||||
# raise as plain ValueError.
|
||||
is_local_validation_error = (
|
||||
isinstance(api_error, (ValueError, TypeError))
|
||||
and not isinstance(
|
||||
api_error, (UnicodeEncodeError, json.JSONDecodeError)
|
||||
)
|
||||
and not self._is_provider_stream_parse_error(api_error)
|
||||
# ssl.SSLError (and its subclass SSLCertVerificationError)
|
||||
# inherits from OSError *and* ValueError via Python MRO,
|
||||
# so the isinstance(ValueError) check above would
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue