Module ‘langchain.vectorstores.pinecone’ has no attribute ‘init’

When I attempted to install ‘pinecone’ I received the error “could not find a version that satisfies the requirement pinecone”

So instead, I installed pinecone-client

Here is my code:

from langchain.vectorstores import pinecone
pinecone.init(api_key=“API_KEY”, environment=“asia-northeast1-gcp”)

I receive the following error:

File “C:\Users\Luke\PycharmProjects\Langchainv1\Langchain_Pinecone.py”, line 2, in
pinecone.init(api_key=“API_KEY”, environment=“asia-northeast1-gcp”)
AttributeError: module ‘langchain.vectorstores.pinecone’ has no attribute ‘init’

Does this error have anything to do with the fact that I was unable to install pinecone but installed pinecone-client? That doesnt appear to be the case using Pinecones documentation: https://docs.pinecone.io/docs/python-client

I appreciate any help. I have reinstalled pinecone-client and am a bit stumped.

I am attempting to embed text to pinecone to use for queries using Langchain. Unfortunately, I cannot even get to the point where pinecone is loading properly.

from langchain.vectorstores import pinecone

The correct import here is import pinecone.

There are two steps to getting Pinecone set up with LangChain: (1) connect to Pinecone client with the pinecone module and authenticate, then (2) use the Pinecone interface that LangChain provides.

You already have done some of the steps, and @NickODell noted the right way to import the Pinecone client. This answer demonstrates a quick walkthrough of both steps in this process.

Step 1: Initialize/Authenticate Pinecone client

#!pip install pinecone-client
import pinecone 

pinecone.init(
    api_key=os.environ["PINECONE_API_KEY"], 
    environment=os.environ["PINECONE_ENV"]
)

pc_index_name = "put-your-pinecone-index-name-here"
pc_namespace = "if-you-are-using-a-namespace-put-it-here"

# note: these 2 lines are not strictly necessary, just testing client connection
pc_index = pinecone.Index(pc_index_name)
pc_index.describe_index_stats() 

Step 2: Create LangChain Pinecone interface

from langchain.vectorstores import Pinecone

pc_interface = Pinecone.from_existing_index(
    pc_index_name, 
    embedding=OpenAIEmbeddings(), 
    namespace=pc_namespace
)

from langchain.chains.qa_with_sources import load_qa_with_sources_chain
from langchain.chains import RetrievalQAWithSourcesChain

# now you can do QA with your Pinecone DB or whatever else you want
my_qa_chain = RetrievalQAWithSourcesChain(
    combine_documents_chain=load_qa_with_sources_chain(...),
    retriever=pc_interface.as_retriever()
)

So pinecone (lowercase) is the Pinecone client module, made by Pinecone. And Pinecone (uppercase) is the LangChain interface to a Pinecone vectorstore, made by LangChain (which needs an authenticated pinecone.init() in order to function).