[BUG] Vector Search feature missing

[SteveSitekitcare]

Describe the bug I've installed a fresh copy of VSCODE on my Windows 11 PC. All the dependencies go green for the vscode prompt flow plugin setup.

I have downloaded a Prompt Flow from ai.azure.com I made myself. It had a vector db lookup in it. The issue is my local vscode doesnt know how to deal with the vector node in the flow.dag.yml (see screenshot below)

I read I needed to install "python -m pip install promptflow-vectordb[azure]" But I get the following issue

System Info

• Promptflow Package Version: 1.16.1
• Operating System: Windows 11
• Python Version: 3.12.6

Can anyone help?

UPDATE 1

I managed to get all the tooling installed and visible in VS Code. using the following

conda create -n conda311pf python=3.11
conda activate conda311pf

pip install pymongo pyopenssl cryptography --upgrade        
pip install promptflow-vectordb[azure]      
pip install promptflow
pip install promptflow-core
pip install promptflow-tools
pip install promptflow-devkit promptflow-tracing promptflow-azure
pip install azureml-rag

But now I cant seem to use the vectorloolup even though Ive created a connection to the relevant Azure AI Search.

UPDATE 2

I have managed to get further with the vector support by doing the following

1. Create file .azureml/config.json under flow or it's parent folder.
2. Fill the workspace info into it with below format:

{
    "subscription_id": "..",
    "resource_group": "..",
    "workspace_name": ".."
}

3. And running the following

az login --tenant <id>
az configure --defaults group=<groupid> workspace=<workspacename>

My issue is now the index_type fails to populate with the following error

promptflow._utils.tool_utils.RetrieveToolFuncResultError: Unable to retrieve result due to 'Error when calling function promptflow_vectordb.tool.common_index_lookup_utils.list_available_index_types: Invalid URL 'None': No scheme supplied. Perhaps you meant https://None?'. Please contact the tool author/support team for troubleshooting assistance.

UPDATE 3

So I havent found a solution to fixing this vectordb plugin issue, so for the time being I've worked around the issue by writing my own python implementation that does the same thing, Below is a list of resources I used to do this.

• https://learn.microsoft.com/en-us/azure/search/search-get-started-rest?tabs=azure-cli
• https://learn.microsoft.com/en-us/rest/api/searchservice/documents/search-post?view=rest-searchservice-2024-07-01&tabs=HTTP
• https://learn.microsoft.com/en-us/azure/ai-services/openai/how-to/embeddings?tabs=python-new
• https://github.com/Azure/azure-search-vector-samples/blob/main/demo-python/code/integrated-vectorization/azure-search-integrated-vectorization-sample.ipynb
• https://azuresdkdocs.blob.core.windows.net/$web/python/azure-search-documents/latest/index.html
• https://github.com/Azure/azure-sdk-for-python/blob/main/sdk/search/azure-search-documents/samples/sample_vector_search.py#L139
• https://cookbook.openai.com/examples/vector_databases/azuresearch/getting_started_with_azure_ai_search_and_openai
• https://learn.microsoft.com/en-us/azure/search/vector-search-how-to-query?tabs=query-2024-07-01%2Cbuiltin-portal

[tuukkapuranen]

Regarding your update 1, I had the same issue and found the following question asking the same. I used pip install promptflow-vectordb==0.2.10 and pip install promptflow-vectordb[azure]==0.2.10 as a successful workaround.

Perhaps there is an issue with the latest release of vectordb package.