langchain 0.3.26__py3-none-any.whl → 0.4.0.dev0__py3-none-any.whl

langchain/chains/retrieval_qa/base.py

@@ -3,7 +3,6 @@
 from __future__ import annotations
 
 import inspect
-import warnings
 from abc import abstractmethod
 from typing import Any, Optional
 
@@ -19,6 +18,7 @@ from langchain_core.prompts import PromptTemplate
 from langchain_core.retrievers import BaseRetriever
 from langchain_core.vectorstores import VectorStore
 from pydantic import ConfigDict, Field, model_validator
+from typing_extensions import override
 
 from langchain.chains.base import Chain
 from langchain.chains.combine_documents.base import BaseCombineDocumentsChain
@@ -69,7 +69,7 @@ class BaseRetrievalQA(Chain):
         """
         _output_keys = [self.output_key]
         if self.return_source_documents:
-            _output_keys = _output_keys + ["source_documents"]
+            _output_keys = [*_output_keys, "source_documents"]
         return _output_keys
 
     @classmethod
@@ -84,10 +84,14 @@ class BaseRetrievalQA(Chain):
         """Initialize from LLM."""
         _prompt = prompt or PROMPT_SELECTOR.get_prompt(llm)
         llm_chain = LLMChain(
-            llm=llm, prompt=_prompt, callbacks=callbacks, **(llm_chain_kwargs or {})
+            llm=llm,
+            prompt=_prompt,
+            callbacks=callbacks,
+            **(llm_chain_kwargs or {}),
         )
         document_prompt = PromptTemplate(
-            input_variables=["page_content"], template="Context:\n{page_content}"
+            input_variables=["page_content"],
+            template="Context:\n{page_content}",
         )
         combine_documents_chain = StuffDocumentsChain(
             llm_chain=llm_chain,
@@ -113,7 +117,9 @@ class BaseRetrievalQA(Chain):
         """Load chain from chain type."""
         _chain_type_kwargs = chain_type_kwargs or {}
         combine_documents_chain = load_qa_chain(
-            llm, chain_type=chain_type, **_chain_type_kwargs
+            llm,
+            chain_type=chain_type,
+            **_chain_type_kwargs,
         )
         return cls(combine_documents_chain=combine_documents_chain, **kwargs)
 
@@ -141,6 +147,7 @@ class BaseRetrievalQA(Chain):
 
         res = indexqa({'query': 'This is my query'})
         answer, docs = res['result'], res['source_documents']
+
         """
         _run_manager = run_manager or CallbackManagerForChainRun.get_noop_manager()
         question = inputs[self.input_key]
@@ -152,13 +159,14 @@ class BaseRetrievalQA(Chain):
         else:
             docs = self._get_docs(question)  # type: ignore[call-arg]
         answer = self.combine_documents_chain.run(
-            input_documents=docs, question=question, callbacks=_run_manager.get_child()
+            input_documents=docs,
+            question=question,
+            callbacks=_run_manager.get_child(),
         )
 
         if self.return_source_documents:
             return {self.output_key: answer, "source_documents": docs}
-        else:
-            return {self.output_key: answer}
+        return {self.output_key: answer}
 
     @abstractmethod
     async def _aget_docs(
@@ -184,6 +192,7 @@ class BaseRetrievalQA(Chain):
 
         res = indexqa({'query': 'This is my query'})
         answer, docs = res['result'], res['source_documents']
+
         """
         _run_manager = run_manager or AsyncCallbackManagerForChainRun.get_noop_manager()
         question = inputs[self.input_key]
@@ -195,13 +204,14 @@ class BaseRetrievalQA(Chain):
         else:
             docs = await self._aget_docs(question)  # type: ignore[call-arg]
         answer = await self.combine_documents_chain.arun(
-            input_documents=docs, question=question, callbacks=_run_manager.get_child()
+            input_documents=docs,
+            question=question,
+            callbacks=_run_manager.get_child(),
         )
 
         if self.return_source_documents:
             return {self.output_key: answer, "source_documents": docs}
-        else:
-            return {self.output_key: answer}
+        return {self.output_key: answer}
 
 
 @deprecated(
@@ -269,7 +279,8 @@ class RetrievalQA(BaseRetrievalQA):
     ) -> list[Document]:
         """Get docs."""
         return self.retriever.invoke(
-            question, config={"callbacks": run_manager.get_child()}
+            question,
+            config={"callbacks": run_manager.get_child()},
         )
 
     async def _aget_docs(
@@ -280,7 +291,8 @@ class RetrievalQA(BaseRetrievalQA):
     ) -> list[Document]:
         """Get docs."""
         return await self.retriever.ainvoke(
-            question, config={"callbacks": run_manager.get_child()}
+            question,
+            config={"callbacks": run_manager.get_child()},
         )
 
     @property
@@ -310,15 +322,6 @@ class VectorDBQA(BaseRetrievalQA):
     search_kwargs: dict[str, Any] = Field(default_factory=dict)
     """Extra search args."""
 
-    @model_validator(mode="before")
-    @classmethod
-    def raise_deprecation(cls, values: dict) -> Any:
-        warnings.warn(
-            "`VectorDBQA` is deprecated - "
-            "please use `from langchain.chains import RetrievalQA`"
-        )
-        return values
-
     @model_validator(mode="before")
     @classmethod
     def validate_search_type(cls, values: dict) -> Any:
@@ -326,9 +329,11 @@ class VectorDBQA(BaseRetrievalQA):
         if "search_type" in values:
             search_type = values["search_type"]
             if search_type not in ("similarity", "mmr"):
-                raise ValueError(f"search_type of {search_type} not allowed.")
+                msg = f"search_type of {search_type} not allowed."
+                raise ValueError(msg)
         return values
 
+    @override
     def _get_docs(
         self,
         question: str,
@@ -338,14 +343,19 @@ class VectorDBQA(BaseRetrievalQA):
         """Get docs."""
         if self.search_type == "similarity":
             docs = self.vectorstore.similarity_search(
-                question, k=self.k, **self.search_kwargs
+                question,
+                k=self.k,
+                **self.search_kwargs,
             )
         elif self.search_type == "mmr":
             docs = self.vectorstore.max_marginal_relevance_search(
-                question, k=self.k, **self.search_kwargs
+                question,
+                k=self.k,
+                **self.search_kwargs,
             )
         else:
-            raise ValueError(f"search_type of {self.search_type} not allowed.")
+            msg = f"search_type of {self.search_type} not allowed."
+            raise ValueError(msg)
         return docs
 
     async def _aget_docs(
@@ -355,7 +365,8 @@ class VectorDBQA(BaseRetrievalQA):
         run_manager: AsyncCallbackManagerForChainRun,
     ) -> list[Document]:
         """Get docs."""
-        raise NotImplementedError("VectorDBQA does not support async")
+        msg = "VectorDBQA does not support async"
+        raise NotImplementedError(msg)
 
     @property
     def _chain_type(self) -> str:

langchain/chains/retrieval_qa/prompt.py

@@ -1,4 +1,3 @@
-# flake8: noqa
 from langchain_core.prompts import PromptTemplate
 
 prompt_template = """Use the following pieces of context to answer the question at the end. If you don't know the answer, just say that you don't know, don't try to make up an answer.
@@ -6,7 +5,7 @@ prompt_template = """Use the following pieces of context to answer the question
 {context}
 
 Question: {question}
-Helpful Answer:"""
+Helpful Answer:"""  # noqa: E501
 PROMPT = PromptTemplate(
     template=prompt_template, input_variables=["context", "question"]
 )

langchain/chains/router/__init__.py

@@ -4,9 +4,9 @@ from langchain.chains.router.multi_prompt import MultiPromptChain
 from langchain.chains.router.multi_retrieval_qa import MultiRetrievalQAChain
 
 __all__ = [
-    "RouterChain",
-    "MultiRouteChain",
+    "LLMRouterChain",
     "MultiPromptChain",
     "MultiRetrievalQAChain",
-    "LLMRouterChain",
+    "MultiRouteChain",
+    "RouterChain",
 ]

langchain/chains/router/base.py

@@ -12,11 +12,14 @@ from langchain_core.callbacks import (
     Callbacks,
 )
 from pydantic import ConfigDict
+from typing_extensions import override
 
 from langchain.chains.base import Chain
 
 
 class Route(NamedTuple):
+    """A route to a destination chain."""
+
     destination: Optional[str]
     next_inputs: dict[str, Any]
 
@@ -25,12 +28,12 @@ class RouterChain(Chain, ABC):
     """Chain that outputs the name of a destination chain and the inputs to it."""
 
     @property
+    @override
     def output_keys(self) -> list[str]:
         return ["destination", "next_inputs"]
 
     def route(self, inputs: dict[str, Any], callbacks: Callbacks = None) -> Route:
-        """
-        Route inputs to a destination chain.
+        """Route inputs to a destination chain.
 
         Args:
             inputs: inputs to the chain
@@ -43,8 +46,19 @@ class RouterChain(Chain, ABC):
         return Route(result["destination"], result["next_inputs"])
 
     async def aroute(
-        self, inputs: dict[str, Any], callbacks: Callbacks = None
+        self,
+        inputs: dict[str, Any],
+        callbacks: Callbacks = None,
     ) -> Route:
+        """Route inputs to a destination chain.
+
+        Args:
+            inputs: inputs to the chain
+            callbacks: callbacks to use for the chain
+
+        Returns:
+            a Route object
+        """
         result = await self.acall(inputs, callbacks=callbacks)
         return Route(result["destination"], result["next_inputs"])
 
@@ -59,7 +73,7 @@ class MultiRouteChain(Chain):
     default_chain: Chain
     """Default chain to use when none of the destination chains are suitable."""
     silent_errors: bool = False
-    """If True, use default_chain when an invalid destination name is provided. 
+    """If True, use default_chain when an invalid destination name is provided.
     Defaults to False."""
 
     model_config = ConfigDict(
@@ -93,20 +107,20 @@ class MultiRouteChain(Chain):
         route = self.router_chain.route(inputs, callbacks=callbacks)
 
         _run_manager.on_text(
-            str(route.destination) + ": " + str(route.next_inputs), verbose=self.verbose
+            str(route.destination) + ": " + str(route.next_inputs),
+            verbose=self.verbose,
         )
         if not route.destination:
             return self.default_chain(route.next_inputs, callbacks=callbacks)
-        elif route.destination in self.destination_chains:
+        if route.destination in self.destination_chains:
             return self.destination_chains[route.destination](
-                route.next_inputs, callbacks=callbacks
+                route.next_inputs,
+                callbacks=callbacks,
             )
-        elif self.silent_errors:
+        if self.silent_errors:
             return self.default_chain(route.next_inputs, callbacks=callbacks)
-        else:
-            raise ValueError(
-                f"Received invalid destination chain name '{route.destination}'"
-            )
+        msg = f"Received invalid destination chain name '{route.destination}'"
+        raise ValueError(msg)
 
     async def _acall(
         self,
@@ -118,21 +132,23 @@ class MultiRouteChain(Chain):
         route = await self.router_chain.aroute(inputs, callbacks=callbacks)
 
         await _run_manager.on_text(
-            str(route.destination) + ": " + str(route.next_inputs), verbose=self.verbose
+            str(route.destination) + ": " + str(route.next_inputs),
+            verbose=self.verbose,
         )
         if not route.destination:
             return await self.default_chain.acall(
-                route.next_inputs, callbacks=callbacks
+                route.next_inputs,
+                callbacks=callbacks,
             )
-        elif route.destination in self.destination_chains:
+        if route.destination in self.destination_chains:
             return await self.destination_chains[route.destination].acall(
-                route.next_inputs, callbacks=callbacks
+                route.next_inputs,
+                callbacks=callbacks,
             )
-        elif self.silent_errors:
+        if self.silent_errors:
             return await self.default_chain.acall(
-                route.next_inputs, callbacks=callbacks
-            )
-        else:
-            raise ValueError(
-                f"Received invalid destination chain name '{route.destination}'"
+                route.next_inputs,
+                callbacks=callbacks,
             )
+        msg = f"Received invalid destination chain name '{route.destination}'"
+        raise ValueError(msg)

langchain/chains/router/embedding_router.py

@@ -11,6 +11,7 @@ from langchain_core.documents import Document
 from langchain_core.embeddings import Embeddings
 from langchain_core.vectorstores import VectorStore
 from pydantic import ConfigDict
+from typing_extensions import override
 
 from langchain.chains.router.base import RouterChain
 
@@ -34,6 +35,7 @@ class EmbeddingRouterChain(RouterChain):
         """
         return self.routing_keys
 
+    @override
     def _call(
         self,
         inputs: dict[str, Any],
@@ -43,6 +45,7 @@ class EmbeddingRouterChain(RouterChain):
         results = self.vectorstore.similarity_search(_input, k=1)
         return {"next_inputs": inputs, "destination": results[0].metadata["name"]}
 
+    @override
     async def _acall(
         self,
         inputs: dict[str, Any],
@@ -63,10 +66,12 @@ class EmbeddingRouterChain(RouterChain):
         """Convenience constructor."""
         documents = []
         for name, descriptions in names_and_descriptions:
-            for description in descriptions:
-                documents.append(
+            documents.extend(
+                [
                     Document(page_content=description, metadata={"name": name})
-                )
+                    for description in descriptions
+                ]
+            )
         vectorstore = vectorstore_cls.from_documents(documents, embeddings)
         return cls(vectorstore=vectorstore, **kwargs)
 
@@ -80,10 +85,12 @@ class EmbeddingRouterChain(RouterChain):
     ) -> EmbeddingRouterChain:
         """Convenience constructor."""
         documents = []
-        for name, descriptions in names_and_descriptions:
-            for description in descriptions:
-                documents.append(
-                    Document(page_content=description, metadata={"name": name})
-                )
+        documents.extend(
+            [
+                Document(page_content=description, metadata={"name": name})
+                for name, descriptions in names_and_descriptions
+                for description in descriptions
+            ]
+        )
         vectorstore = await vectorstore_cls.afrom_documents(documents, embeddings)
         return cls(vectorstore=vectorstore, **kwargs)

langchain/chains/router/llm_router.py

@@ -15,7 +15,7 @@ from langchain_core.output_parsers import BaseOutputParser
 from langchain_core.prompts import BasePromptTemplate
 from langchain_core.utils.json import parse_and_check_json_markdown
 from pydantic import model_validator
-from typing_extensions import Self
+from typing_extensions import Self, override
 
 from langchain.chains import LLMChain
 from langchain.chains.router.base import RouterChain
@@ -27,7 +27,7 @@ from langchain.chains.router.base import RouterChain
     message=(
         "Use RunnableLambda to select from multiple prompt templates. See example "
         "in API reference: "
-        "https://api.python.langchain.com/en/latest/chains/langchain.chains.router.llm_router.LLMRouterChain.html"  # noqa: E501
+        "https://api.python.langchain.com/en/latest/chains/langchain.chains.router.llm_router.LLMRouterChain.html"
     ),
 )
 class LLMRouterChain(RouterChain):
@@ -96,21 +96,23 @@ class LLMRouterChain(RouterChain):
             )
 
             chain.invoke({"query": "what color are carrots"})
+
     """  # noqa: E501
 
     llm_chain: LLMChain
     """LLM chain used to perform routing"""
 
     @model_validator(mode="after")
-    def validate_prompt(self) -> Self:
+    def _validate_prompt(self) -> Self:
         prompt = self.llm_chain.prompt
         if prompt.output_parser is None:
-            raise ValueError(
+            msg = (
                 "LLMRouterChain requires base llm_chain prompt to have an output"
                 " parser that converts LLM text output to a dictionary with keys"
                 " 'destination' and 'next_inputs'. Received a prompt with no output"
                 " parser."
             )
+            raise ValueError(msg)
         return self
 
     @property
@@ -124,7 +126,7 @@ class LLMRouterChain(RouterChain):
     def _validate_outputs(self, outputs: dict[str, Any]) -> None:
         super()._validate_outputs(outputs)
         if not isinstance(outputs["next_inputs"], dict):
-            raise ValueError
+            raise ValueError  # noqa: TRY004
 
     def _call(
         self,
@@ -135,11 +137,10 @@ class LLMRouterChain(RouterChain):
         callbacks = _run_manager.get_child()
 
         prediction = self.llm_chain.predict(callbacks=callbacks, **inputs)
-        output = cast(
-            dict[str, Any],
+        return cast(
+            "dict[str, Any]",
             self.llm_chain.prompt.output_parser.parse(prediction),
         )
-        return output
 
     async def _acall(
         self,
@@ -148,15 +149,17 @@ class LLMRouterChain(RouterChain):
     ) -> dict[str, Any]:
         _run_manager = run_manager or CallbackManagerForChainRun.get_noop_manager()
         callbacks = _run_manager.get_child()
-        output = cast(
-            dict[str, Any],
+        return cast(
+            "dict[str, Any]",
             await self.llm_chain.apredict_and_parse(callbacks=callbacks, **inputs),
         )
-        return output
 
     @classmethod
     def from_llm(
-        cls, llm: BaseLanguageModel, prompt: BasePromptTemplate, **kwargs: Any
+        cls,
+        llm: BaseLanguageModel,
+        prompt: BasePromptTemplate,
+        **kwargs: Any,
     ) -> LLMRouterChain:
         """Convenience constructor."""
         llm_chain = LLMChain(llm=llm, prompt=prompt)
@@ -170,16 +173,17 @@ class RouterOutputParser(BaseOutputParser[dict[str, str]]):
     next_inputs_type: type = str
     next_inputs_inner_key: str = "input"
 
+    @override
     def parse(self, text: str) -> dict[str, Any]:
         try:
             expected_keys = ["destination", "next_inputs"]
             parsed = parse_and_check_json_markdown(text, expected_keys)
             if not isinstance(parsed["destination"], str):
-                raise ValueError("Expected 'destination' to be a string.")
+                msg = "Expected 'destination' to be a string."
+                raise TypeError(msg)
             if not isinstance(parsed["next_inputs"], self.next_inputs_type):
-                raise ValueError(
-                    f"Expected 'next_inputs' to be {self.next_inputs_type}."
-                )
+                msg = f"Expected 'next_inputs' to be {self.next_inputs_type}."
+                raise TypeError(msg)
             parsed["next_inputs"] = {self.next_inputs_inner_key: parsed["next_inputs"]}
             if (
                 parsed["destination"].strip().lower()
@@ -188,8 +192,7 @@ class RouterOutputParser(BaseOutputParser[dict[str, str]]):
                 parsed["destination"] = None
             else:
                 parsed["destination"] = parsed["destination"].strip()
-            return parsed
         except Exception as e:
-            raise OutputParserException(
-                f"Parsing text\n{text}\n raised following error:\n{e}"
-            )
+            msg = f"Parsing text\n{text}\n raised following error:\n{e}"
+            raise OutputParserException(msg) from e
+        return parsed

langchain/chains/router/multi_prompt.py

@@ -7,6 +7,7 @@ from typing import Any, Optional
 from langchain_core._api import deprecated
 from langchain_core.language_models import BaseLanguageModel
 from langchain_core.prompts import PromptTemplate
+from typing_extensions import override
 
 from langchain.chains import ConversationChain
 from langchain.chains.base import Chain
@@ -21,7 +22,7 @@ from langchain.chains.router.multi_prompt_prompt import MULTI_PROMPT_ROUTER_TEMP
     removal="1.0",
     message=(
         "Please see migration guide here for recommended implementation: "
-        "https://python.langchain.com/docs/versions/migrating_chains/multi_prompt_chain/"  # noqa: E501
+        "https://python.langchain.com/docs/versions/migrating_chains/multi_prompt_chain/"
     ),
 )
 class MultiPromptChain(MultiRouteChain):
@@ -139,9 +140,11 @@ class MultiPromptChain(MultiRouteChain):
             result = await app.ainvoke({"query": "what color are carrots"})
             print(result["destination"])
             print(result["answer"])
+
     """  # noqa: E501
 
     @property
+    @override
     def output_keys(self) -> list[str]:
         return ["text"]
 
@@ -157,7 +160,7 @@ class MultiPromptChain(MultiRouteChain):
         destinations = [f"{p['name']}: {p['description']}" for p in prompt_infos]
         destinations_str = "\n".join(destinations)
         router_template = MULTI_PROMPT_ROUTER_TEMPLATE.format(
-            destinations=destinations_str
+            destinations=destinations_str,
         )
         router_prompt = PromptTemplate(
             template=router_template,

langchain/chains/router/multi_retrieval_qa.py

@@ -8,6 +8,7 @@ from typing import Any, Optional
 from langchain_core.language_models import BaseLanguageModel
 from langchain_core.prompts import PromptTemplate
 from langchain_core.retrievers import BaseRetriever
+from typing_extensions import override
 
 from langchain.chains import ConversationChain
 from langchain.chains.base import Chain
@@ -32,6 +33,7 @@ class MultiRetrievalQAChain(MultiRouteChain):
     """Default chain to use when router doesn't map input to one of the destinations."""
 
     @property
+    @override
     def output_keys(self) -> list[str]:
         return ["result"]
 
@@ -47,15 +49,32 @@ class MultiRetrievalQAChain(MultiRouteChain):
         default_chain_llm: Optional[BaseLanguageModel] = None,
         **kwargs: Any,
     ) -> MultiRetrievalQAChain:
+        """Create a multi retrieval qa chain from an LLM and a default chain.
+
+        Args:
+            llm: The language model to use.
+            retriever_infos: Dictionaries containing retriever information.
+            default_retriever: Optional default retriever to use if no default chain
+                is provided.
+            default_prompt: Optional prompt template to use for the default retriever.
+            default_chain: Optional default chain to use when router doesn't map input
+                to one of the destinations.
+            default_chain_llm: Optional language model to use if no default chain and
+                no default retriever are provided.
+            **kwargs: Additional keyword arguments to pass to the chain.
+        Returns:
+            An instance of the multi retrieval qa chain.
+        """
         if default_prompt and not default_retriever:
-            raise ValueError(
+            msg = (
                 "`default_retriever` must be specified if `default_prompt` is "
                 "provided. Received only `default_prompt`."
             )
+            raise ValueError(msg)
         destinations = [f"{r['name']}: {r['description']}" for r in retriever_infos]
         destinations_str = "\n".join(destinations)
         router_template = MULTI_RETRIEVAL_ROUTER_TEMPLATE.format(
-            destinations=destinations_str
+            destinations=destinations_str,
         )
         router_prompt = PromptTemplate(
             template=router_template,
@@ -74,15 +93,18 @@ class MultiRetrievalQAChain(MultiRouteChain):
             _default_chain = default_chain
         elif default_retriever:
             _default_chain = RetrievalQA.from_llm(
-                llm, prompt=default_prompt, retriever=default_retriever
+                llm,
+                prompt=default_prompt,
+                retriever=default_retriever,
             )
         else:
             prompt_template = DEFAULT_TEMPLATE.replace("input", "query")
             prompt = PromptTemplate(
-                template=prompt_template, input_variables=["history", "query"]
+                template=prompt_template,
+                input_variables=["history", "query"],
             )
             if default_chain_llm is None:
-                raise NotImplementedError(
+                msg = (
                     "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."
@@ -90,6 +112,7 @@ class MultiRetrievalQAChain(MultiRouteChain):
                     "from langchain_openai import ChatOpenAI\n"
                     "llm = ChatOpenAI()"
                 )
+                raise NotImplementedError(msg)
             _default_chain = ConversationChain(
                 llm=default_chain_llm,
                 prompt=prompt,