Is it possible to have open WebUI as a chat-interface for Flowise?

[BadlyDrawnBoy]

Hi,
so the reason why I'm still paying for ChatGPT is that it allows me to go back and refine my question after it generated an answer - so because I can have another go (Could have more features but at least you're not left with on-shot question).
There is a chance (not only a chance...) that I'm not fully getting it how to use Flowise in a external application - but couldn't there be a Flowise openAI API-interface like "https://api.openai.com/v1" that could be easily used in existing Applications? Couldn't Flowise and Open WebUI make babies, so that we get the nice UI with the power of Flowise?! :)

Thanks!

Cheers,
Martin

[mdlmarkham]

This sounds fantastic! OpenWebUI has done a great job on the front end, and Flowise is amazing on the back-end... +1!

[taowang1993]

You can use the Pipeline feature of Open WebUI to connect to flowise.

[mdlmarkham]

Thanks @taowang1993 - I haven't used the Pipelines feature yet. Do you have an example of how that could be done?

[taowang1993]

Thanks @taowang1993 - I haven't used the Pipelines feature yet. Do you have an example of how that could be done?

After you deploy Pipeline. Install the following flowise.py:

from typing import List, Union, Generator, Iterator
from pydantic import BaseModel
import requests
import json

class Pipeline:
    class Valves(BaseModel):
        pass  # No API key needed for Flowise API

    def __init__(self):
        self.name = "your flowise chatbot name"
        self.valves = self.Valves()

    async def on_startup(self):
        print(f"on_startup:{__name__}")

    async def on_shutdown(self):
        print(f"on_shutdown:{__name__}")

    def pipe(
        self, user_message: str, model_id: str, messages: List[dict], body: dict
    ) -> Union[str, Generator, Iterator]:
        print(f"pipe:{__name__}")

        print(messages)
        print(user_message)

        API_URL = "your flowise chatbot url"

        headers = {
            "Content-Type": "application/json"
        }

        # Creating the payload based on your example
        payload = {
            "question": user_message
        }

        print("Payload:", payload)

        try:
            r = requests.post(
                url=API_URL,
                json=payload,
                headers=headers,
                stream=True,
            )

            r.raise_for_status()

            if body.get("stream"):
                for line in r.iter_lines():
                    line_data = line.decode('utf-8')
                    print("Line data:", line_data)
                    # Parse the JSON line and extract the text part
                    response_json = json.loads(line_data)
                    if "text" in response_json:
                        print("Streaming text:", response_json["text"])
                        yield response_json["text"]
            else:
                response_json = r.json()
                print("Response JSON:", response_json)
                # Return only the "text" part of the response
                if "text" in response_json:
                    print("Text:", response_json["text"])
                    return response_json["text"]
                else:
                    print("No text in response")
                    return "No text in response"
        except Exception as e:
            print("Error:", e)
            return f"Error: {e}"

[Chakabal]

This code was very useful. Thanks. One question: I want to pass the OpenWebUI Chat ID in the request to flowise via override config. Is there any way to take the Open Web UI ChatID and pass it as a flowise session ID?

[Damsjov]

@taowang1993

Thank you - great help! However, when I run a prompt from Open WebUI over Pipeline to Flowise, it stalls and I get this error message in the pipeline log: Error: Expecting value: line 1 column 1 (char 0). Do you have any idea where it went wrong?

Thanks!

[McNickSistoPro]

Have anyone of you had any issue getting streaming working with Flowise via Pipe/Pipelines ?