Update Langchain to 0.1 and migrate most components to CustomComponent (

#1382)

This PR updates Langchain to 0.1.* and move components previously
created automatically to start using the CustomComponent framework.

This will improve maintainability and move Langflow to be closer to the
framework.

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "langflow"
-version = "0.6.5a12"
+version = "0.6.5a13"
 description = "A Python package with a built-in web application"
 authors = ["Logspace <[email protected]>"]
 maintainers = [
@@ -25,8 +25,6 @@ documentation = "https://docs.langflow.org"
 langflow = "langflow.__main__:main"
 
 [tool.poetry.dependencies]
-
-
 python = ">=3.9,<3.11"
 fastapi = "^0.108.0"
 uvicorn = "^0.25.0"
@@ -35,8 +33,8 @@ google-search-results = "^2.4.1"
 google-api-python-client = "^2.79.0"
 typer = "^0.9.0"
 gunicorn = "^21.2.0"
-langchain = "~0.0.345"
-openai = "^1.6.1"
+langchain = "~0.1.0"
+openai = "^1.10.0"
 pandas = "2.0.3"
 chromadb = "^0.4.0"
 huggingface-hub = { version = "^0.19.0", extras = ["inference"] }
@@ -55,15 +53,15 @@ tiktoken = "~0.5.0"
 wikipedia = "^1.4.0"
 qdrant-client = "^1.7.0"
 websockets = "^10.3"
-weaviate-client = { version = "^4.4b6", allow-prereleases = true }
+weaviate-client = "*"
 jina = "*"
 sentence-transformers = { version = "^2.2.2", optional = true }
 ctransformers = { version = "^0.2.10", optional = true }
 cohere = "^4.39.0"
 python-multipart = "^0.0.6"
 sqlmodel = "^0.0.14"
 faiss-cpu = "^1.7.4"
-anthropic = "^0.8.0"
+anthropic = "^0.13.0"
 orjson = "3.9.3"
 multiprocess = "^0.70.14"
 cachetools = "^5.3.1"
@@ -98,7 +96,7 @@ markupsafe = "^2.1.3"
 extract-msg = "^0.45.0"
 # jq is not available for windows
 jq = { version = "^1.6.0", markers = "sys_platform != 'win32'" }
-boto3 = "^1.28.63"
+boto3 = "^1.34.0"
 numexpr = "^2.8.6"
 qianfan = "0.2.0"
 pgvector = "^0.2.3"
@@ -107,12 +105,13 @@ langchain-google-genai = "^0.0.2"
 elasticsearch = "^8.11.1"
 pytube = "^15.0.0"
 llama-index = "^0.9.24"
+langchain-openai = "^0.0.2"
 
 [tool.poetry.group.dev.dependencies]
 pytest-asyncio = "^0.23.1"
 types-redis = "^4.6.0.5"
 ipykernel = "^6.27.0"
-mypy = "^1.7.1"
+mypy = "^1.8.0"
 ruff = "^0.1.5"
 httpx = "*"
 pytest = "^7.4.2"
@@ -154,7 +153,8 @@ exclude = ["src/backend/langflow/alembic/*"]
 line-length = 120
 
 [tool.mypy]
-plugins = "pydantic.mypy"
+plugins = ["pydantic.mypy"]
+follow_imports = "silent"
 
 [build-system]
 requires = ["poetry-core"]

src/backend/langflow/alembic/versions/b2fa308044b5_add_unique_constraints.py

@@ -28,7 +28,7 @@ def upgrade() -> None:
             batch_op.add_column(sa.Column('folder', sqlmodel.sql.sqltypes.AutoString(), nullable=True))
             batch_op.add_column(sa.Column('user_id', sqlmodel.sql.sqltypes.GUID(), nullable=True))
             batch_op.create_index(batch_op.f('ix_flow_user_id'), ['user_id'], unique=False)
-            batch_op.create_foreign_key(None, 'user', ['user_id'], ['id'])
+            batch_op.create_foreign_key('fk_flow_user_id_user', 'user', ['user_id'], ['id'])
     except Exception:
         pass
     # ### end Alembic commands ###
@@ -38,7 +38,7 @@ def downgrade() -> None:
     # ### commands auto generated by Alembic - please adjust! ###
     try:
         with op.batch_alter_table('flow', schema=None) as batch_op:
-            batch_op.drop_constraint(None, type_='foreignkey')
+            batch_op.drop_constraint('fk_flow_user_id_user', type_='foreignkey')
             batch_op.drop_index(batch_op.f('ix_flow_user_id'))
             batch_op.drop_column('user_id')
             batch_op.drop_column('folder')

src/backend/langflow/components/agents/CSVAgent.py

@@ -0,0 +1,23 @@
+from langflow import CustomComponent
+from langflow.field_typing import BaseLanguageModel, AgentExecutor
+from langchain_experimental.agents.agent_toolkits.csv.base import create_csv_agent
+
+
+class CSVAgentComponent(CustomComponent):
+    display_name = "CSVAgent"
+    description = "Construct a CSV agent from a CSV and tools."
+    documentation = "https://python.langchain.com/docs/modules/agents/toolkits/csv"
+
+    def build_config(self):
+        return {
+            "llm": {"display_name": "LLM", "type": BaseLanguageModel},
+            "path": {"display_name": "Path", "field_type": "file", "suffixes": [".csv"], "file_types": [".csv"]},
+        }
+
+    def build(
+        self,
+        llm: BaseLanguageModel,
+        path: str,
+    ) -> AgentExecutor:
+        # Instantiate and return the CSV agent class with the provided llm and path
+        return create_csv_agent(llm=llm, path=path)

src/backend/langflow/components/agents/JsonAgent.py

@@ -0,0 +1,24 @@
+from langflow import CustomComponent
+from langchain.agents import AgentExecutor, create_json_agent
+from langflow.field_typing import (
+    BaseLanguageModel,
+)
+from langchain_community.agent_toolkits.json.toolkit import JsonToolkit
+
+
+class JsonAgentComponent(CustomComponent):
+    display_name = "JsonAgent"
+    description = "Construct a json agent from an LLM and tools."
+
+    def build_config(self):
+        return {
+            "llm": {"display_name": "LLM"},
+            "toolkit": {"display_name": "Toolkit"},
+        }
+
+    def build(
+        self,
+        llm: BaseLanguageModel,
+        toolkit: JsonToolkit,
+    ) -> AgentExecutor:
+        return create_json_agent(llm=llm, toolkit=toolkit)

src/backend/langflow/components/agents/OpenAIConversationalAgent.py

@@ -3,7 +3,7 @@
 from langchain.agents.agent import AgentExecutor
 from langchain.agents.agent_toolkits.conversational_retrieval.openai_functions import _get_default_system_message
 from langchain.agents.openai_functions_agent.base import OpenAIFunctionsAgent
-from langchain.chat_models import ChatOpenAI
+from langchain_community.chat_models import ChatOpenAI
 from langchain.memory.token_buffer import ConversationTokenBufferMemory
 from langchain.prompts import SystemMessagePromptTemplate
 from langchain.prompts.chat import MessagesPlaceholder

src/backend/langflow/components/agents/SQLAgent.py

@@ -0,0 +1,29 @@
+from langflow import CustomComponent
+from typing import Union, Callable
+from langchain.agents import AgentExecutor
+from langflow.field_typing import BaseLanguageModel
+from langchain_community.agent_toolkits.sql.base import create_sql_agent
+from langchain.sql_database import SQLDatabase
+from langchain_community.agent_toolkits import SQLDatabaseToolkit
+
+
+class SQLAgentComponent(CustomComponent):
+    display_name = "SQLAgent"
+    description = "Construct an SQL agent from an LLM and tools."
+
+    def build_config(self):
+        return {
+            "llm": {"display_name": "LLM"},
+            "database_uri": {"display_name": "Database URI"},
+            "verbose": {"display_name": "Verbose", "value": False, "advanced": True},
+        }
+
+    def build(
+        self,
+        llm: BaseLanguageModel,
+        database_uri: str,
+        verbose: bool = False,
+    ) -> Union[AgentExecutor, Callable]:
+        db = SQLDatabase.from_uri(database_uri)
+        toolkit = SQLDatabaseToolkit(db=db, llm=llm)
+        return create_sql_agent(llm=llm, toolkit=toolkit)

src/backend/langflow/components/agents/VectorStoreAgent.py

@@ -0,0 +1,23 @@
+from langflow import CustomComponent
+from langchain.agents import AgentExecutor, create_vectorstore_agent
+from langchain.agents.agent_toolkits.vectorstore.toolkit import VectorStoreToolkit
+from typing import Union, Callable
+from langflow.field_typing import BaseLanguageModel
+
+
+class VectorStoreAgentComponent(CustomComponent):
+    display_name = "VectorStoreAgent"
+    description = "Construct an agent from a Vector Store."
+
+    def build_config(self):
+        return {
+            "llm": {"display_name": "LLM"},
+            "vector_store_toolkit": {"display_name": "Vector Store Info"},
+        }
+
+    def build(
+        self,
+        llm: BaseLanguageModel,
+        vector_store_toolkit: VectorStoreToolkit,
+    ) -> Union[AgentExecutor, Callable]:
+        return create_vectorstore_agent(llm=llm, toolkit=vector_store_toolkit)

src/backend/langflow/components/agents/VectorStoreRouterAgent.py

@@ -0,0 +1,19 @@
+from langflow import CustomComponent
+from langchain_core.language_models.base import BaseLanguageModel
+from langchain.agents.agent_toolkits.vectorstore.toolkit import VectorStoreRouterToolkit
+from langchain.agents import create_vectorstore_router_agent
+from typing import Callable
+
+
+class VectorStoreRouterAgentComponent(CustomComponent):
+    display_name = "VectorStoreRouterAgent"
+    description = "Construct an agent from a Vector Store Router."
+
+    def build_config(self):
+        return {
+            "llm": {"display_name": "LLM"},
+            "vectorstoreroutertoolkit": {"display_name": "Vector Store Router Toolkit"},
+        }
+
+    def build(self, llm: BaseLanguageModel, vectorstoreroutertoolkit: VectorStoreRouterToolkit) -> Callable:
+        return create_vectorstore_router_agent(llm=llm, toolkit=vectorstoreroutertoolkit)

src/backend/langflow/components/chains/LLMChain.py

@@ -28,5 +28,5 @@ def build(
         prompt: BasePromptTemplate,
         llm: BaseLanguageModel,
         memory: Optional[BaseMemory] = None,
-    ) -> Union[Chain, Callable]:
+    ) -> Union[Chain, Callable, LLMChain]:
         return LLMChain(prompt=prompt, llm=llm, memory=memory)

src/backend/langflow/components/chains/LLMCheckerChain.py

@@ -0,0 +1,24 @@
+from langflow import CustomComponent
+from langchain.chains import LLMCheckerChain
+from typing import Union, Callable
+from langflow.field_typing import (
+    BaseLanguageModel,
+    Chain,
+)
+
+
+class LLMCheckerChainComponent(CustomComponent):
+    display_name = "LLMCheckerChain"
+    description = ""
+    documentation = "https://python.langchain.com/docs/modules/chains/additional/llm_checker"
+
+    def build_config(self):
+        return {
+            "llm": {"display_name": "LLM"},
+        }
+
+    def build(
+        self,
+        llm: BaseLanguageModel,
+    ) -> Union[Chain, Callable]:
+        return LLMCheckerChain(llm=llm)

src/backend/langflow/components/chains/LLMMathChain.py

@@ -0,0 +1,31 @@
+from typing import Callable, Optional, Union
+
+from langchain.chains import LLMChain, LLMMathChain
+
+from langflow import CustomComponent
+from langflow.field_typing import BaseLanguageModel, BaseMemory, Chain
+
+
+class LLMMathChainComponent(CustomComponent):
+    display_name = "LLMMathChain"
+    description = "Chain that interprets a prompt and executes python code to do math."
+    documentation = "https://python.langchain.com/docs/modules/chains/additional/llm_math"
+
+    def build_config(self):
+        return {
+            "llm": {"display_name": "LLM"},
+            "llm_chain": {"display_name": "LLM Chain"},
+            "memory": {"display_name": "Memory"},
+            "input_key": {"display_name": "Input Key"},
+            "output_key": {"display_name": "Output Key"},
+        }
+
+    def build(
+        self,
+        llm: BaseLanguageModel,
+        llm_chain: LLMChain,
+        input_key: str = "question",
+        output_key: str = "answer",
+        memory: Optional[BaseMemory] = None,
+    ) -> Union[LLMMathChain, Callable, Chain]:
+        return LLMMathChain(llm=llm, llm_chain=llm_chain, input_key=input_key, output_key=output_key, memory=memory)

src/backend/langflow/components/chains/RetrievalQA.py

@@ -0,0 +1,39 @@
+from typing import Callable, Optional, Union
+
+from langchain.chains.combine_documents.base import BaseCombineDocumentsChain
+from langchain.chains.retrieval_qa.base import BaseRetrievalQA, RetrievalQA
+from langflow import CustomComponent
+from langflow.field_typing import BaseMemory, BaseRetriever
+
+
+class RetrievalQAComponent(CustomComponent):
+    display_name = "RetrievalQA"
+    description = "Chain for question-answering against an index."
+
+    def build_config(self):
+        return {
+            "combine_documents_chain": {"display_name": "Combine Documents Chain"},
+            "retriever": {"display_name": "Retriever"},
+            "memory": {"display_name": "Memory", "required": False},
+            "input_key": {"display_name": "Input Key", "advanced": True},
+            "output_key": {"display_name": "Output Key", "advanced": True},
+            "return_source_documents": {"display_name": "Return Source Documents"},
+        }
+
+    def build(
+        self,
+        combine_documents_chain: BaseCombineDocumentsChain,
+        retriever: BaseRetriever,
+        memory: Optional[BaseMemory] = None,
+        input_key: str = "query",
+        output_key: str = "result",
+        return_source_documents: bool = True,
+    ) -> Union[BaseRetrievalQA, Callable]:
+        return RetrievalQA(
+            combine_documents_chain=combine_documents_chain,
+            retriever=retriever,
+            memory=memory,
+            input_key=input_key,
+            output_key=output_key,
+            return_source_documents=return_source_documents,
+        )

src/backend/langflow/components/chains/RetrievalQAWithSourcesChain.py

@@ -0,0 +1,42 @@
+from typing import Optional
+
+from langchain.chains import RetrievalQAWithSourcesChain
+from langchain.chains.qa_with_sources.base import BaseQAWithSourcesChain
+from langchain.chains.combine_documents.base import BaseCombineDocumentsChain
+
+from langflow import CustomComponent
+from langflow.field_typing import BaseLanguageModel, BaseMemory, BaseRetriever
+
+
+class RetrievalQAWithSourcesChainComponent(CustomComponent):
+    display_name = "RetrievalQAWithSourcesChain"
+    description = "Question-answering with sources over an index."
+
+    def build_config(self):
+        return {
+            "llm": {"display_name": "LLM"},
+            "chain_type": {
+                "display_name": "Chain Type",
+                "options": ["stuff", "map_reduce", "map_rerank", "refine"],
+            },
+            "memory": {"display_name": "Memory"},
+            "return_source_documents": {"display_name": "Return Source Documents"},
+        }
+
+    def build(
+        self,
+        retriever: BaseRetriever,
+        llm: BaseLanguageModel,
+        combine_documents_chain: BaseCombineDocumentsChain,
+        chain_type: str,
+        memory: Optional[BaseMemory] = None,
+        return_source_documents: Optional[bool] = True,
+    ) -> BaseQAWithSourcesChain:
+        return RetrievalQAWithSourcesChain.from_chain_type(
+            llm=llm,
+            chain_type=chain_type,
+            combine_documents_chain=combine_documents_chain,
+            memory=memory,
+            return_source_documents=return_source_documents,
+            retriever=retriever,
+        )