chore: langchain and dependencies fixed

django_app/pyproject.toml

@@ -64,7 +64,6 @@ duckdb = "^1.4.4"
 opentelemetry-api = "1.25.0"
 opentelemetry-sdk = "1.25.0"
 
-
 [tool.poetry.group.dev.dependencies]
 pytest = "^8.3.2"
 pytest-django = "^4.12.0"

notebooks/pyproject.toml

@@ -15,7 +15,6 @@ package-mode = false
 python = ">=3.12,<3.13"
 redbox = {path="../redbox", develop=false}
 redbox_app = {path="../django_app", develop=false}
-langchain-mcp-adapters = "^0.0.11"
 django = "5.2.11"
 
 [tool.poetry.group.dev.dependencies]

redbox/pyproject.toml

@@ -1,59 +1,71 @@
 [tool.ruff]
 line-length = 120
-target-version = 'py312'
+target-version = "py312"
 
 [tool.poetry]
 name = "redbox"
 version = "0.4.0"
 description = "RAG application that uses GenAI to chat with and summarise civil service documents"
-authors = ["DBT Redbox <redbox@businessandtrade.gov.uk>", "i.AI <i-dot-ai-enquiries@cabinetoffice.gov.uk>"]
+authors = [
+    "DBT Redbox <redbox@businessandtrade.gov.uk>",
+    "i.AI <i-dot-ai-enquiries@cabinetoffice.gov.uk>"
+]
 license = "MIT"
 readme = "../README.md"
 
 [tool.poetry.dependencies]
 python = ">=3.12,<3.13"
-pydantic = "^2.7.1"
-elasticsearch = "^8.15.0"
-langchain-community = "0.3.27"
-langchain = "^0.3.4"
-langchain_openai = ">0.1.21"
-boto3 = "^1.34.160"
-pydantic-settings = "^2.3.4"
-langchain-elasticsearch = ">0.2.2"
-pytest-dotenv = "^0.5.2"
-kneed = "^0.8.5"
-langgraph = "^0.2.76"
-langchain-aws = "^0.2.3"
-wikipedia = "^1.4.0"
-opensearch-py = "^2.7.1"
-scikit-learn = "^1.5.2"
-django-environ = "^0.12.0"
-mohawk = "^1.1.0"
-django-waffle = "^4.2.0"
-langchain-mcp-adapters = "^0.0.11"
-pymupdf = "^1.26.0"
-ddtrace = "^3.10.0"
+
+pydantic = "^2.7"
+pydantic-settings = "^2.3"
+
+langchain = "^1.2"
+langchain-community = "^0.4.1"
+langchain-openai = "^1.1"
+langchain-aws = "^1.0.0"
+langchain-elasticsearch = "^1.0.0"
+langgraph = "^1.1"
+
+elasticsearch = "^8.15"
+opensearch-py = "^2.7"
+
+boto3 = "^1.34"
+
+starlette = "^0.49"
+
+django = "^5.2"
+django-environ = "^0.12"
+django-waffle = "^4.2"
 django-requestlogs = "^0.8"
-django-log-formatter-asim = "^1.3.0"
-pytest = "^8.4.2"
-urllib3 = "2.6.3"
-langchain-text-splitters = "0.3.9"
-starlette = "0.49.1"
-duckdb = "^1.4.3"
-django = "5.2.11"
-protobuf = "5.29.6"
-pandas = "^2.3.3"
+django-log-formatter-asim = "^1.3"
+
+pandas = "^2.3"
+scikit-learn = "^1.5"
+duckdb = "^1.4"
+pymupdf = "^1.26"
+kneed = "^0.8"
+
+mcp = "^1.7.1"
+
+wikipedia = "^1.4"
+mohawk = "^1.1"
+ddtrace = "^3.10"
+
+urllib3 = "^2.2"
 
 [tool.poetry.group.dev.dependencies]
-pytest = "^8.3.2"
-moto = "^5.0.12"
-pytest-cov = "^5.0.0"
-pytest-asyncio = "^0.23.6"
-jsonlines = "^4.0.0"
-deepeval = "^3.8.4"
-pytest-mock = "^3.14.0"
-boto3-stubs = {extras = ["essential"], version = "^1.35.28"}
-requests-mock = "^1.12.1"
+pytest = "^8.4"
+pytest-cov = "^5.0"
+pytest-asyncio = "^0.23"
+pytest-dotenv = "^0.5"
+pytest-mock = "^3.14"
+requests-mock = "^1.12"
+
+moto = "^5.0"
+jsonlines = "^4.0"
+deepeval = "^3.8"
+
+boto3-stubs = { extras = ["essential"], version = "^1.35" }
 
 [build-system]
 requires = ["poetry-core"]

redbox/redbox/chains/ingest.py

@@ -3,7 +3,7 @@
 from io import BytesIO
 from typing import TYPE_CHECKING, Iterator
 
-from langchain.vectorstores import VectorStore
+from langchain_core.vectorstores import VectorStore
 from langchain_core.documents.base import Document
 from langchain_core.runnables import Runnable, RunnableLambda, chain
 

redbox/redbox/graph/nodes/processes.py

@@ -14,7 +14,7 @@
 
 import pandas as pd
 from botocore.exceptions import EventStreamError
-from langchain.schema import StrOutputParser
+from langchain_core.output_parsers import StrOutputParser
 from langchain_core.callbacks.manager import dispatch_custom_event
 from langchain_core.documents import Document
 from langchain_core.messages import AIMessage, HumanMessage, RemoveMessage

redbox/redbox/graph/nodes/tools.py

@@ -7,7 +7,7 @@
 import threading
 import time
 from io import StringIO
-from typing import Annotated, Callable, Iterable, Literal, Union
+from typing import Callable, Iterable, Literal, Union
 
 import boto3
 import duckdb
@@ -20,7 +20,6 @@
 from langchain_core.embeddings.embeddings import Embeddings
 from langchain_core.messages import ToolCall
 from langchain_core.tools import Tool, tool
-from langgraph.prebuilt import InjectedState
 from mohawk import Sender
 from opensearchpy import OpenSearch
 from sklearn.metrics.pairwise import cosine_similarity
@@ -62,9 +61,7 @@ def format_result(loop, content, artifact, status, is_intermediate_step):
 
 def build_document_from_prompt_tool(loop: bool = False):
     @tool(response_format="content_and_artifact")
-    def _retrieve_document_from_prompt(
-        state: Annotated[RedboxState, InjectedState], is_intermediate_step: bool = False
-    ) -> tuple:
+    def _retrieve_document_from_prompt(state: RedboxState, is_intermediate_step: bool = False) -> tuple:
         """
         Retrieve document from user prompt
 
@@ -87,9 +84,7 @@ def _retrieve_document_from_prompt(
 
 def build_retrieve_document_full_text(es_client: Union[Elasticsearch, OpenSearch], index_name: str, loop: bool = False):
     @tool(response_format="content_and_artifact")
-    def _retrieve_document_full_text(
-        state: Annotated[RedboxState, InjectedState], is_intermediate_step: bool = False
-    ) -> tuple:
+    def _retrieve_document_full_text(state: RedboxState, is_intermediate_step: bool = False) -> tuple:
         """
         Retrieve full texts from state.documents. This tool should be used when a full text from a document is required.
         This tool does not retrieve documents in knowledge base.
@@ -156,7 +151,7 @@ def query_repo(el_query, is_intermediate_step, loop):
 
     @tool(response_format="content_and_artifact")
     def _retrieve_specific_file_knowledge_base(
-        state: Annotated[RedboxState, InjectedState],
+        state: RedboxState,
         uri: str,
     ) -> tuple[str, list[Document]]:
         """
@@ -171,9 +166,7 @@ def _retrieve_specific_file_knowledge_base(
         return query_repo(el_query, is_intermediate_step=False, loop=False)
 
     @tool(response_format="content_and_artifact")
-    def _retrieve_knowledge_base(
-        state: Annotated[RedboxState, InjectedState], is_intermediate_step: bool = False
-    ) -> tuple:
+    def _retrieve_knowledge_base(state: RedboxState, is_intermediate_step: bool = False) -> tuple:
         """
         Retrieve full texts from all knowledge base files.
 
@@ -251,7 +244,7 @@ def search_repo(query, selected_files, permitted_files, ai_settings, start_time=
         return format_documents(sorted_documents), sorted_documents
 
     @tool(response_format="content_and_artifact")
-    def _search_documents(query: str, state: Annotated[RedboxState, InjectedState]) -> tuple[str, list[Document]]:
+    def _search_documents(query: str, state: RedboxState) -> tuple[str, list[Document]]:
         """
         "Searches through state.documents to find and extract relevant information. This tool should be used whenever a query involves finding, searching, or retrieving information from documents that have already been uploaded or provided to the system.
 
@@ -273,7 +266,7 @@ def _search_documents(query: str, state: Annotated[RedboxState, InjectedState])
         )
 
     @tool(response_format="content_and_artifact")
-    def _search_knowledge_base(query: str, state: Annotated[RedboxState, InjectedState]) -> tuple[str, list[Document]]:
+    def _search_knowledge_base(query: str, state: RedboxState) -> tuple[str, list[Document]]:
         """
         "Searches through knowledge base files to find and extract relevant information. This tool should be used whenever a query involves finding, searching, or retrieving information from knowledge base.
 
@@ -396,7 +389,7 @@ def build_query_tabular_file_tool(
     def _query_tabular_file(
         sql_query: str,
         uri: str,
-        state: Annotated[RedboxState, InjectedState],
+        state: RedboxState,
     ) -> tuple[str, list[Document]]:
         """
         Executes the SQL query against tabular files retrieved by the embedded retriever.
@@ -509,7 +502,7 @@ def recalculate_similarity(response, query, num_results):
         return response
 
     @tool(response_format="content_and_artifact")
-    def _search_govuk(query: str, state: Annotated[RedboxState, InjectedState]) -> tuple[str, list[Document]]:
+    def _search_govuk(query: str, state: RedboxState) -> tuple[str, list[Document]]:
         """
         Search for documents on www.gov.uk based on a query string.
         This endpoint is used to search for documents on www.gov.uk. There are many types of documents on www.gov.uk.

redbox/redbox/graph/root.py

@@ -4,8 +4,8 @@
 from langchain_core.messages import AIMessage
 from langchain_core.vectorstores import VectorStoreRetriever
 from langgraph.graph import END, START, StateGraph
-from langgraph.graph.graph import CompiledGraph
-from langgraph.pregel import RetryPolicy
+from langgraph.graph.state import CompiledStateGraph
+from langgraph.types import RetryPolicy
 
 from redbox.chains.components import get_structured_response_with_citations_parser
 from redbox.chains.runnables import build_self_route_output_parser
@@ -134,7 +134,7 @@ def get_search_graph(
     prompt_set: PromptSet = PromptSet.Search,
     debug: bool = False,
     final_sources: bool = True,
-) -> CompiledGraph:
+) -> CompiledStateGraph:
     """Creates a subgraph for retrieval augmented generation (RAG)."""
     citations_output_parser, format_instructions = get_structured_response_with_citations_parser()
 
@@ -376,7 +376,7 @@ def get_summarise_graph(
 
 def get_chat_graph(
     debug: bool = False,
-) -> CompiledGraph:
+) -> CompiledStateGraph:
     """Creates a subgraph for standard chat."""
     builder = StateGraph(RedboxState)
 
@@ -441,7 +441,7 @@ def build_new_route_graph(
     all_chunks_retriever: VectorStoreRetriever,
     agent_configs: Dict[str, AgentConfig],
     debug: bool = False,
-) -> CompiledGraph:
+) -> CompiledStateGraph:
     def update_submission_eval(state: RedboxState):
         state.tasks_evaluator = EVAL_SUBMISSION
         return state

redbox/redbox/models/settings.py

@@ -6,7 +6,7 @@
 
 import boto3
 from elasticsearch import Elasticsearch
-from langchain.globals import set_debug
+from langchain_core.globals import set_debug
 from opensearchpy import OpenSearch, RequestsHttpConnection
 from pydantic import AnyUrl, BaseModel
 from pydantic_settings import BaseSettings, SettingsConfigDict