Bug/4299 cant instantiate simplechatmodel subclass without defining agenerate

langchain/chat_models/base.py

@@ -2,7 +2,7 @@
 import inspect
 import warnings
 from abc import ABC, abstractmethod
-from typing import Dict, List, Optional
+from typing import Any, Coroutine, Dict, List, Optional
 
 from pydantic import Extra, Field, root_validator
 
@@ -197,6 +197,14 @@ def _generate(
         generation = ChatGeneration(message=message)
         return ChatResult(generations=[generation])
 
+    def _agenerate(
+        self,
+        messages: List[BaseMessage],
+        stop: List[str] | None = None,
+        run_manager: AsyncCallbackManagerForLLMRun | None = None,
+    ) -> Coroutine[Any, Any, ChatResult]:
+        return self._generate(messages, stop=stop, run_manager=run_manager)
+
     @abstractmethod
     def _call(
         self,

langchain/utils.py

@@ -4,6 +4,8 @@
 
 from requests import HTTPError, Response
 
+from langchain.schema import BaseMessage
+
 
 def get_from_dict_or_env(
     data: Dict[str, Any], key: str, env_key: str, default: Optional[str] = None
@@ -78,3 +80,9 @@ def stringify_dict(data: dict) -> str:
     for key, value in data.items():
         text += key + ": " + stringify_value(value) + "\n"
     return text
+
+
+def serialize_msgs(msgs: list[BaseMessage], include_type=False) -> str:
+    return "\n\n".join(
+        (f"{msg.type}: {msg.content}" if include_type else msg.content) for msg in msgs
+    )

langchain/wrappers/__init__.py

@@ -0,0 +1,7 @@
+from langchain.wrappers.chat_model_facade import ChatModelFacade
+from langchain.wrappers.llm_facade import LLMFacade
+
+__all__ = [
+    "ChatModelFacade",
+    "LLMFacade",
+]

langchain/wrappers/chat_model_facade.py

@@ -0,0 +1,33 @@
+from __future__ import annotations
+
+from typing import List, Optional
+
+from langchain.chat_models.base import BaseChatModel, SimpleChatModel
+from langchain.schema import BaseMessage
+from langchain.llms.base import BaseLanguageModel
+from langchain.utils import serialize_msgs
+
+
+class ChatModelFacade(SimpleChatModel):
+    llm: BaseLanguageModel
+
+    def _call(self, messages: List[BaseMessage], stop: Optional[List[str]] = None) -> str:
+        if isinstance(self.llm, BaseChatModel):
+            return self.llm(messages, stop=stop).content
+        elif isinstance(self.llm, BaseLanguageModel):
+            return self.llm(serialize_msgs(messages), stop=stop)
+        else:
+            raise ValueError(
+                f"Invalid llm type: {type(self.llm)}. Must be a chat model or language model."
+            )
+
+    @classmethod
+    def of(cls, llm):
+        if isinstance(llm, BaseChatModel):
+            return llm
+        elif isinstance(llm, BaseLanguageModel):
+            return cls(llm)
+        else:
+            raise ValueError(
+                f"Invalid llm type: {type(llm)}. Must be a chat model or language model."
+            )

langchain/wrappers/llm_facade.py

@@ -0,0 +1,37 @@
+from __future__ import annotations
+
+from typing import Any, List, Mapping, Optional
+
+from langchain.chat_models.base import BaseChatModel
+from langchain.llms.base import LLM, BaseLanguageModel
+
+
+class LLMFacade(LLM):
+    chat_model: BaseChatModel
+
+    @property
+    def _llm_type(self) -> str:
+        return self.chat_model._llm_type
+
+    def _call(
+        self,
+        prompt: str,
+        stop: Optional[List[str]] = None,
+    ) -> str:
+        return self.chat_model.call_as_llm(prompt, stop=stop)
+
+    @property
+    def _identifying_params(self) -> Mapping[str, Any]:
+        """Get the identifying parameters."""
+        return self._chat._identifying_params
+
+    @staticmethod
+    def of(llm) -> LLMFacade:
+        if isinstance(llm, BaseChatModel):
+            return LLMFacade(llm)
+        elif isinstance(llm, BaseLanguageModel):
+            return llm
+        else:
+            raise ValueError(
+                f"Invalid llm type: {type(llm)}. Must be a chat model or language model."
+            )

tests/unit_tests/wrappers/test_chat_model_facade.py

@@ -0,0 +1,12 @@
+from langchain.llms.fake import FakeListLLM
+from langchain.schema import SystemMessage
+from langchain.wrappers.chat_model_facade import ChatModelFacade
+
+
+def test_chat_model_facade():
+    llm = FakeListLLM(responses=["hello", "goodbye"])
+    chat_model = ChatModelFacade.of(llm)
+    input_message = SystemMessage(content="hello")
+    output_message = chat_model([input_message])
+    assert output_message.content == "hello"
+    assert output_message.type == "ai"

tests/unit_tests/wrappers/test_llm_facade.py

@@ -0,0 +1,2 @@
+def test_llm_facade():
+    pass