Error: 1 validation error for OpenAIEmbeddings root Did not find openai_api_key

Using Streamlit
1

I am running the app in the cloud and locally as well. I am getting the following error:

Error: 1 validation error for OpenAIEmbeddings root Did not find openai_api_key, please add an environment variable OPENAI_API_KEY which contains it, or pass openai_api_key as a named parameter. (type=value_error)

My app link is: https://webpage-chatbot.streamlit.app/
My apps’s Github repo’s link is: GitHub - ziayounasch/Webpage-Chatbot

I am using the latest version of streamlit and python 3.11.

def get_vectorstore(document_chunks):
    embeddings = OpenAIEmbeddings(openai_api_key=api_key_input)
    # embeddings = HuggingFaceInstructEmbeddings(model_name="hkunlp/instructor-xl")
    vectorstore = FAISS.from_documents(documents=document_chunks, embedding=embeddings)
    return vectorstore

st.sidebar.image("Logo for Webpage Chatbot Final.png")
api_key_input=st.sidebar.text_input("OpenAI API Key:", type="password", placeholder="Paste your OpenAI API Key here (sk-...)", help="You can get your API key from https://platform.openai.com/account/api-keys.")
if api_key_input and api_key_input.startswith('sk-'):
    llm = ChatOpenAI(openai_api_key=api_key_input, model_name="gpt-3.5-turbo-1106")
else:
    st.warning("Please enter a valid OpenAI API Key starting with 'sk-'.")
2

@Zia_Younas
By opening the application it’s directly printing else statement. Instead you can use submit button for checking whether the api key is provided or not.

3

By pressing the enter button the api key is passed on… In my other apps the same code is working fine on the streamlit cloud (https://ai-pdf-chatbot.streamlit.app/) but when I created this app it didn’t work… I dont know what went wrong… I have checked the code many time but found no mistake…

4

@Zia_Younas

Screenshot_2024-03-09-18-37-08-514_com.android.chrome
Screenshot_2024-03-09-18-37-08-514_com.android.chrome1080×2400 144 KB

5

Yes it is working ok… when I insert api key in this app it processes and generates the response… but the webpage app gives the error…

6

May i know the error.

7

The error is as under:

8

Try to define a key for the text input to receive the openai api key. we can use it to get the user openai api key input as a session variable. This way the value will persist for the given session.

key='openaik',

in

st.sidebar.text_input(
    "OpenAI API Key:",
    type="password",
    key='openaik',
    placeholder="Paste your OpenAI API Key here (sk-...)",
    help="You can get your API key from https://platform.openai.com/account/api-keys."
)

You can use it like this.

if st.session_state.openaik and st.session_state['openaik'].startswith('sk-'):
    llm = ChatOpenAI(openai_api_key=st.session_state.openaik, model_name="gpt-3.5-turbo-1106")
else:
    st.warning("Please enter a valid OpenAI API Key starting with 'sk-'.")

And for the function.

def get_vectorstore(document_chunks):
    embeddings = OpenAIEmbeddings(openai_api_key=st.session_state.openaik)
    vectorstore = FAISS.from_documents(documents=document_chunks, embedding=embeddings)
    return vectorstore
9

I have made these changes and ran the code again but it is still giving the same error… please guide…

10

After serious :slightly_smiling_face: debugging, found out the error is caused by:

def get_website_text(url):
    loader = WebBaseLoader(url)
    index = VectorstoreIndexCreator().from_loaders([loader])
    webDocument = loader.load()
    return webDocument

The class VectorstoreIndexCreator needs an embedding. So to fix this we need to define the embeddings, right here, you can optimize the code later.

def get_website_text(url):
    embeddings = OpenAIEmbeddings(openai_api_key=api_key_input)  # this
    loader = WebBaseLoader(url)
    index = VectorstoreIndexCreator(embedding=embeddings).from_loaders([loader])  # argument
    webDocument = loader.load()
    return webDocument

It runs fine from my test after this fix.

This code can be improve, to capture the error properly.

        # Process input
        if st.button("Process"):
            if url:
                with st.spinner("Processing"):
                    # Validate the URL format
                    if url.startswith("http://") or url.startswith("https://"):
                    # Load the documents from the URL using WebBaseLoader
                        try:
                            # get link text
                            raw_text = get_website_text(url)

                            # get the text chunks
                            text_chunks = get_text_chunks(raw_text)

                            # create vector store
                            vectorstore = get_vectorstore(text_chunks)

                            # Update the processing status in the sidebar
                            st.sidebar.info("Processing completed.")

                            # create conversation chain
                            st.session_state.conversation = get_conversation_chain(
                                vectorstore)
                        except Exception as e:
                            st.error(f"Error: {str(e)}")
                    else:
                        st.error("Error: Invalid URL format.")
            else:
                st.error("Please enter URL.")

Nice app by the way :heartpulse:

11

The other way to fix which in not safe is by using the environment.

import os

oakey = st.text_input('openai api key')
os.environ['OPENAI_API_KEY'] = oakey

With that being defined, the VectorstoreIndexCreator() would now work too without defining the embeddings.

12

Dear Ferdy,
It was really helpful talking to you. I really really appreciate your sincere efforts. Your proposed solution worked. You really made my day.
Thank you sooooo much. You are a star.
Please feel free to contact in case you need my help anytime. I ll be happy to assist.
My LinkedIn : www.linkedin.com/in/zia-younas

Regards,
Zia