feat: 不再预先计算前文token数而是在报错时提醒用户重置

pkg/provider/modelmgr/entities.py

@@ -0,0 +1,25 @@
+from __future__ import annotations
+
+import typing
+
+import pydantic
+
+from . import api
+from . import token
+
+
+class LLMModelInfo(pydantic.BaseModel):
+    """模型"""
+
+    name: str
+
+    model_name: typing.Optional[str] = None
+
+    token_mgr: token.TokenManager
+
+    requester: api.LLMAPIRequester
+
+    tool_call_supported: typing.Optional[bool] = False
+
+    class Config:
+        arbitrary_types_allowed = True