kris/ForcePilot
1
0
Fork 0
Code Issues Pull Requests Actions 1 Packages Projects Releases Wiki Activity

fix(agent): 修复因工具返回的 token 报告错误导致的频繁 summary

Browse Source
This commit is contained in:
Wenjie Zhang 2026-02-03 14:52:54 +08:00
parent b589521775
commit 34bfc0c97e
1 changed files with 2 additions and 23 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

25
src/agents/common/middlewares/summary_middleware.py
View File

@ -132,7 +132,8 @@ def _offload_tool_result(msg: ToolMessage, threshold: int, token_counter: TokenC
return None return None
# 计算 token 数 # 计算 token 数
if token_counter([msg]) <= threshold: msg_tokens = token_counter([msg])
if msg_tokens <= threshold:
return None return None
# 获取工具名称和参数 # 获取工具名称和参数
@ -450,23 +451,6 @@ class SummaryOffloadMiddleware(AgentMiddleware):
return result return result
def _should_summarize_based_on_reported_tokens(self, messages: list[AnyMessage], threshold: float) -> bool:
"""Check if reported token usage from last AIMessage exceeds threshold."""
last_ai_message = next(
(msg for msg in reversed(messages) if isinstance(msg, AIMessage)),
None,
)
if ( # noqa: SIM103
isinstance(last_ai_message, AIMessage)
and last_ai_message.usage_metadata is not None
and (reported_tokens := last_ai_message.usage_metadata.get("total_tokens", -1))
and reported_tokens >= threshold
and (message_provider := last_ai_message.response_metadata.get("model_provider"))
and message_provider == self.model._get_ls_params().get("ls_provider") # noqa: SLF001
):
return True
return False
def _should_summarize(self, messages: list[AnyMessage], total_tokens: int) -> bool: def _should_summarize(self, messages: list[AnyMessage], total_tokens: int) -> bool:
"""Determine whether summarization should run for the current token usage.""" """Determine whether summarization should run for the current token usage."""
if not self._trigger_conditions: if not self._trigger_conditions:
@ -477,8 +461,6 @@ class SummaryOffloadMiddleware(AgentMiddleware):
return True return True
if kind == "tokens" and total_tokens >= value: if kind == "tokens" and total_tokens >= value:
return True return True
if kind == "tokens" and self._should_summarize_based_on_reported_tokens(messages, value):
return True
if kind == "fraction": if kind == "fraction":
max_input_tokens = self._get_profile_limits() max_input_tokens = self._get_profile_limits()
if max_input_tokens is None: if max_input_tokens is None:
@ -488,9 +470,6 @@ class SummaryOffloadMiddleware(AgentMiddleware):
threshold = 1 threshold = 1
if total_tokens >= threshold: if total_tokens >= threshold:
return True return True
if self._should_summarize_based_on_reported_tokens(messages, threshold):
return True
return False return False
def _determine_cutoff_index(self, messages: list[AnyMessage]) -> int: def _determine_cutoff_index(self, messages: list[AnyMessage]) -> int: