QUERY_CONFIG_TYPE is no longer supported in llama_index.

[91d906h4]

Checked other resources

• I added a very descriptive title to this issue.
• I searched the LangChain documentation with the integrated search.
• I used the GitHub search to find a similar question and didn't find it.
• I am sure that this is a bug in LangChain rather than my code.
• The bug is not resolved by updating to the latest stable version of LangChain (or the specific integration package).

Example Code

import networkx as nx
from langchain_community.llms.ollama import Ollama
from langchain.chains.retrieval_qa.base import RetrievalQA
from langchain_community.retrievers import LlamaIndexGraphRetriever

# Initialize the LLM
llm = Ollama(model="llama3.2")

# Create a knowledge graph
graph = nx.Graph()

# Add nodes (countries)
graph.add_node("United States", capital="Washington D.C.")
graph.add_node("Canada", capital="Ottawa")
graph.add_node("France", capital="Paris")

# Add relationships (borders)
graph.add_edge("United States", "Canada")
graph.add_edge("France", "United States", relation="allies")

# Initialize the graph retriever
graph_retriever = LlamaIndexGraphRetriever(graph=graph, max_retriever_size=3)

graph_retriever.get_relevant_documents("what is the relationship between france and United States?")

Error Message and Stack Trace (if applicable)

gchain-ollama package and should be used instead. To use it run pip install -U :class:~langchain-ollamaand import asfrom :class:~langchain_ollama import OllamaLLM``. llm = Ollama(model="llama3.2") E:\Project\.Share\2024-12-11 Graph RAG\test\index.py:31: LangChainDeprecationWarning: The method BaseRetriever.get_relevant_documents was deprecated in langchain-core 0.1.46 and will be removed in 1.0. Use :meth:~invoke` instead.
graph_retriever.get_relevant_documents("what is the relationship between france and United States?")
Traceback (most recent call last):
File "D:\python\Lib\site-packages\langchain_community\retrievers\llama_index.py", line 62, in _get_relevant_documents
from llama_index.core.composability.base import (
ImportError: cannot import name 'QUERY_CONFIG_TYPE' from 'llama_index.core.composability.base' (D:\python\Lib\site-packages\llama_index\core\composability\base.py)

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
File "E:\Project.Share\2024-12-11 Graph RAG\test\index.py", line 31, in
graph_retriever.get_relevant_documents("what is the relationship between france and United States?")
File "D:\python\Lib\site-packages\langchain_core_api\deprecation.py", line 182, in warning_emitting_wrapper
return wrapped(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\python\Lib\site-packages\langchain_core\retrievers.py", line 409, in get_relevant_documents
return self.invoke(query, config, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\python\Lib\site-packages\langchain_core\retrievers.py", line 266, in invoke
raise e
File "D:\python\Lib\site-packages\langchain_core\retrievers.py", line 259, in invoke
result = self._get_relevant_documents(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\python\Lib\site-packages\langchain_community\retrievers\llama_index.py", line 67, in _get_relevant_documents
raise ImportError(
ImportError: You need to install pip install llama-index to use this retriever.

---

In addition, here is the code in D:\python\Lib\site-packages\langchain_community\retrievers\llama_index.py

class LlamaIndexGraphRetriever(BaseRetriever):
    """`LlamaIndex` graph data structure retriever.

    It is used for question-answering with sources over an LlamaIndex
    graph data structure."""

    graph: Any = None
    """LlamaIndex graph to query."""
    query_configs: List[Dict] = Field(default_factory=list)
    """List of query configs to pass to the query method."""

    def _get_relevant_documents(
        self, query: str, *, run_manager: CallbackManagerForRetrieverRun
    ) -> List[Document]:
        """Get documents relevant for a query."""
        try:
            from llama_index.core.base.response.schema import Response
            from llama_index.core.composability.base import (
                QUERY_CONFIG_TYPE,
                ComposableGraph,
            )
        except ImportError:
            raise ImportError(
                "You need to install `pip install llama-index` to use this retriever."
            )
        graph = cast(ComposableGraph, self.graph)

        # for now, inject response_mode="no_text" into query configs
        for query_config in self.query_configs:
            query_config["response_mode"] = "no_text"
        query_configs = cast(List[QUERY_CONFIG_TYPE], self.query_configs)
        response = graph.query(query, query_configs=query_configs)
        response = cast(Response, response)

        # parse source nodes
        docs = []
        for source_node in response.source_nodes:
            metadata = source_node.metadata or {}
            docs.append(
                Document(page_content=source_node.get_content(), metadata=metadata)
            )
        return docs

Description

1. I'm trying to build the GraphRAG with langchain_community.
2. I imported LlamaIndexGraphRetriever class from langchain_community.retrievers
3. I used function get_relevant_documents to retrieval the graph but got error.
4. I install the latest version and several old versions of llama-index, but couldn't find the version with QUERY_CONFIG_TYPE.

System Info

networkx==3.2.1
langchain_community==0.3.11
langchain==0.3.11