langchain[patch]: Revert breaking change until 0.2 release (#21256)

Reverts a minor breaking change until 0.2 release

libs/langchain/langchain/agents/agent_toolkits/vectorstore/toolkit.py

@@ -1,11 +1,18 @@
 """Toolkit for interacting with a vector store."""
 from typing import List
 
+from langchain_community.agent_toolkits.base import BaseToolkit
+from langchain_community.llms.openai import OpenAI
+from langchain_community.tools.vectorstore.tool import (
+    VectorStoreQATool,
+    VectorStoreQAWithSourcesTool,
+)
 from langchain_core.language_models import BaseLanguageModel
 from langchain_core.pydantic_v1 import BaseModel, Field
-from langchain_core.tools import BaseTool, BaseToolkit
 from langchain_core.vectorstores import VectorStore
 
+from langchain.tools import BaseTool
+
 
 class VectorStoreInfo(BaseModel):
     """Information about a VectorStore."""
@@ -24,7 +31,7 @@ class VectorStoreToolkit(BaseToolkit):
     """Toolkit for interacting with a Vector Store."""
 
     vectorstore_info: VectorStoreInfo = Field(exclude=True)
-    llm: BaseLanguageModel
+    llm: BaseLanguageModel = Field(default_factory=lambda: OpenAI(temperature=0))
 
     class Config:
         """Configuration for this pydantic object."""
@@ -33,15 +40,6 @@ class VectorStoreToolkit(BaseToolkit):
 
     def get_tools(self) -> List[BaseTool]:
         """Get the tools in the toolkit."""
-        try:
-            from langchain_community.tools.vectorstore.tool import (
-                VectorStoreQATool,
-                VectorStoreQAWithSourcesTool,
-            )
-        except ImportError:
-            raise ImportError(
-                "You need to install langchain-community to use this toolkit."
-            )
         description = VectorStoreQATool.get_description(
             self.vectorstore_info.name, self.vectorstore_info.description
         )
@@ -67,7 +65,7 @@ class VectorStoreRouterToolkit(BaseToolkit):
     """Toolkit for routing between Vector Stores."""
 
     vectorstores: List[VectorStoreInfo] = Field(exclude=True)
-    llm: BaseLanguageModel
+    llm: BaseLanguageModel = Field(default_factory=lambda: OpenAI(temperature=0))
 
     class Config:
         """Configuration for this pydantic object."""
@@ -77,14 +75,6 @@ class VectorStoreRouterToolkit(BaseToolkit):
     def get_tools(self) -> List[BaseTool]:
         """Get the tools in the toolkit."""
         tools: List[BaseTool] = []
-        try:
-            from langchain_community.tools.vectorstore.tool import (
-                VectorStoreQATool,
-            )
-        except ImportError:
-            raise ImportError(
-                "You need to install langchain-community to use this toolkit."
-            )
         for vectorstore_info in self.vectorstores:
             description = VectorStoreQATool.get_description(
                 vectorstore_info.name, vectorstore_info.description

libs/langchain/langchain/chains/natbot/base.py

@@ -4,6 +4,7 @@ from __future__ import annotations
 import warnings
 from typing import Any, Dict, List, Optional
 
+from langchain_community.llms.openai import OpenAI
 from langchain_core.callbacks import CallbackManagerForChainRun
 from langchain_core.language_models import BaseLanguageModel
 from langchain_core.pydantic_v1 import Extra, root_validator
@@ -67,11 +68,8 @@ class NatBotChain(Chain):
     @classmethod
     def from_default(cls, objective: str, **kwargs: Any) -> NatBotChain:
         """Load with default LLMChain."""
-        raise NotImplementedError(
+        llm = OpenAI(temperature=0.5, best_of=10, n=3, max_tokens=50)
-            "This method is no longer implemented. Please use from_llm."
+        return cls.from_llm(llm, objective, **kwargs)
-            "llm = OpenAI(temperature=0.5, best_of=10, n=3, max_tokens=50)"
-            "For example, NatBotChain.from_llm(llm, objective)"
-        )
 
     @classmethod
     def from_llm(

libs/langchain/langchain/chains/openai_functions/openapi.py

@@ -6,6 +6,8 @@ from collections import defaultdict
 from typing import TYPE_CHECKING, Any, Callable, Dict, List, Optional, Tuple, Union
 
 import requests
+from langchain_community.chat_models import ChatOpenAI
+from langchain_community.utilities.openapi import OpenAPISpec
 from langchain_core.callbacks import CallbackManagerForChainRun
 from langchain_core.language_models import BaseLanguageModel
 from langchain_core.output_parsers.openai_functions import JsonOutputFunctionsParser
@@ -19,7 +21,6 @@ from langchain.chains.sequential import SequentialChain
 from langchain.tools import APIOperation
 
 if TYPE_CHECKING:
-    from langchain_community.utilities.openapi import OpenAPISpec
     from openapi_pydantic import Parameter
 
 
@@ -255,13 +256,6 @@ def get_openapi_chain(
         prompt: Main prompt template to use.
         request_chain: Chain for taking the functions output and executing the request.
     """
-    try:
-        from langchain_community.utilities.openapi import OpenAPISpec
-    except ImportError as e:
-        raise ImportError(
-            "Could not import langchain_community.utilities.openapi. "
-            "Please install it with `pip install langchain-community`."
-        ) from e
     if isinstance(spec, str):
         for conversion in (
             OpenAPISpec.from_url,
@@ -278,12 +272,9 @@ def get_openapi_chain(
         if isinstance(spec, str):
             raise ValueError(f"Unable to parse spec from source {spec}")
     openai_fns, call_api_fn = openapi_spec_to_openai_fn(spec)
-    if not llm:
+    llm = llm or ChatOpenAI(
-        raise ValueError(
+        model="gpt-3.5-turbo-0613",
-            "Must provide an LLM for this chain.For example,\n"
+    )
-            "from langchain_openai import ChatOpenAI\n"
-            "llm = ChatOpenAI()\n"
-        )
     prompt = prompt or ChatPromptTemplate.from_template(
         "Use the provided API's to respond to this user query:\n\n{query}"
     )

libs/langchain/langchain/chains/router/multi_retrieval_qa.py

@@ -3,6 +3,7 @@ from __future__ import annotations
 
 from typing import Any, Dict, List, Mapping, Optional
 
+from langchain_community.chat_models import ChatOpenAI
 from langchain_core.language_models import BaseLanguageModel
 from langchain_core.prompts import PromptTemplate
 from langchain_core.retrievers import BaseRetriever
@@ -41,8 +42,6 @@ class MultiRetrievalQAChain(MultiRouteChain):
         default_retriever: Optional[BaseRetriever] = None,
         default_prompt: Optional[PromptTemplate] = None,
         default_chain: Optional[Chain] = None,
-        *,
-        default_chain_llm: Optional[BaseLanguageModel] = None,
         **kwargs: Any,
     ) -> MultiRetrievalQAChain:
         if default_prompt and not default_retriever:
@@ -79,20 +78,8 @@ class MultiRetrievalQAChain(MultiRouteChain):
             prompt = PromptTemplate(
                 template=prompt_template, input_variables=["history", "query"]
             )
-            if default_chain_llm is None:
-                raise NotImplementedError(
-                    "conversation_llm must be provided if default_chain is not "
-                    "specified. This API has been changed to avoid instantiating "
-                    "default LLMs on behalf of users."
-                    "You can provide a conversation LLM like so:\n"
-                    "from langchain_openai import ChatOpenAI\n"
-                    "llm = ChatOpenAI()"
-                )
             _default_chain = ConversationChain(
-                llm=default_chain_llm,
+                llm=ChatOpenAI(), prompt=prompt, input_key="query", output_key="result"
-                prompt=prompt,
-                input_key="query",
-                output_key="result",
             )
         return cls(
             router_chain=router_chain,

libs/langchain/langchain/indexes/vectorstore.py

@@ -1,6 +1,9 @@
 from typing import Any, Dict, List, Optional, Type
 
-from langchain_core.document_loaders import BaseLoader
+from langchain_community.document_loaders.base import BaseLoader
+from langchain_community.embeddings.openai import OpenAIEmbeddings
+from langchain_community.llms.openai import OpenAI
+from langchain_community.vectorstores.inmemory import InMemoryVectorStore
 from langchain_core.documents import Document
 from langchain_core.embeddings import Embeddings
 from langchain_core.language_models import BaseLanguageModel
@@ -35,14 +38,7 @@ class VectorStoreIndexWrapper(BaseModel):
         **kwargs: Any,
     ) -> str:
         """Query the vectorstore."""
-        if llm is None:
+        llm = llm or OpenAI(temperature=0)
-            raise NotImplementedError(
-                "This API has been changed to require an LLM. "
-                "Please provide an llm to use for querying the vectorstore.\n"
-                "For example,\n"
-                "from langchain_openai import OpenAI\n"
-                "llm = OpenAI(temperature=0)"
-            )
         retriever_kwargs = retriever_kwargs or {}
         chain = RetrievalQA.from_chain_type(
             llm, retriever=self.vectorstore.as_retriever(**retriever_kwargs), **kwargs
@@ -57,14 +53,7 @@ class VectorStoreIndexWrapper(BaseModel):
         **kwargs: Any,
     ) -> str:
         """Query the vectorstore."""
-        if llm is None:
+        llm = llm or OpenAI(temperature=0)
-            raise NotImplementedError(
-                "This API has been changed to require an LLM. "
-                "Please provide an llm to use for querying the vectorstore.\n"
-                "For example,\n"
-                "from langchain_openai import OpenAI\n"
-                "llm = OpenAI(temperature=0)"
-            )
         retriever_kwargs = retriever_kwargs or {}
         chain = RetrievalQA.from_chain_type(
             llm, retriever=self.vectorstore.as_retriever(**retriever_kwargs), **kwargs
@@ -79,14 +68,7 @@ class VectorStoreIndexWrapper(BaseModel):
         **kwargs: Any,
     ) -> dict:
         """Query the vectorstore and get back sources."""
-        if llm is None:
+        llm = llm or OpenAI(temperature=0)
-            raise NotImplementedError(
-                "This API has been changed to require an LLM. "
-                "Please provide an llm to use for querying the vectorstore.\n"
-                "For example,\n"
-                "from langchain_openai import OpenAI\n"
-                "llm = OpenAI(temperature=0)"
-            )
         retriever_kwargs = retriever_kwargs or {}
         chain = RetrievalQAWithSourcesChain.from_chain_type(
             llm, retriever=self.vectorstore.as_retriever(**retriever_kwargs), **kwargs
@@ -101,14 +83,7 @@ class VectorStoreIndexWrapper(BaseModel):
         **kwargs: Any,
     ) -> dict:
         """Query the vectorstore and get back sources."""
-        if llm is None:
+        llm = llm or OpenAI(temperature=0)
-            raise NotImplementedError(
-                "This API has been changed to require an LLM. "
-                "Please provide an llm to use for querying the vectorstore.\n"
-                "For example,\n"
-                "from langchain_openai import OpenAI\n"
-                "llm = OpenAI(temperature=0)"
-            )
         retriever_kwargs = retriever_kwargs or {}
         chain = RetrievalQAWithSourcesChain.from_chain_type(
             llm, retriever=self.vectorstore.as_retriever(**retriever_kwargs), **kwargs
@@ -116,31 +91,11 @@ class VectorStoreIndexWrapper(BaseModel):
         return await chain.ainvoke({chain.question_key: question})
 
 
-def _get_in_memory_vectorstore() -> Type[VectorStore]:
-    """Get the InMemoryVectorStore."""
-    import warnings
-
-    try:
-        from langchain_community.vectorstores.inmemory import InMemoryVectorStore
-    except ImportError:
-        raise ImportError(
-            "Please install langchain-community to use the InMemoryVectorStore."
-        )
-    warnings.warn(
-        "Using InMemoryVectorStore as the default vectorstore."
-        "This memory store won't persist data. You should explicitly"
-        "specify a vectorstore when using VectorstoreIndexCreator"
-    )
-    return InMemoryVectorStore
-
-
 class VectorstoreIndexCreator(BaseModel):
     """Logic for creating indexes."""
 
-    vectorstore_cls: Type[VectorStore] = Field(
+    vectorstore_cls: Type[VectorStore] = InMemoryVectorStore
-        default_factory=_get_in_memory_vectorstore
+    embedding: Embeddings = Field(default_factory=OpenAIEmbeddings)
-    )
-    embedding: Embeddings
     text_splitter: TextSplitter = Field(default_factory=_get_default_text_splitter)
     vectorstore_kwargs: dict = Field(default_factory=dict)