Tracing using RunTree for Generator function

[Abe410]

Hi

I have a chatbot which is distributed across multiple files, and is then exposed with an API endpoint.

I am using RunTree to trace the calls, I am not using Langchain for my agent. RunTree provides the best way to do that considering my code structure.

My question is, how do I pass a generator object to the RunTree object as follows:

def agent_run():
    # Chatbot agent code
    ......
    
    # Patch the generator output to the pipeline
    self.pipeline.end(outputs={"post_content": generator})
    self.pipeline.patch()

    # Generator is sent to API endpoint
    return generator

I want to trace the output from the generator, and then also return the generator to the API endpoint so that I can stream it on the front end.

Any pointers as to how to do it?