pages/manual_comparisor.py

import os
import random
import datetime as dt
import streamlit as st
from streamlit.logger import get_logger
from langchain.schema.messages import HumanMessage
from mongo_utils import get_db_client, new_battle_result, get_non_assesed_comparison, new_completion_error
from app_config import ISSUES, SOURCES

logger = get_logger(__name__)
openai_api_key = os.environ['OPENAI_API_KEY'] 
if 'db_client' not in st.session_state:
    st.session_state["db_client"] = get_db_client()

def disable_buttons():
    return len(comparison) == 0

def replaceA():
    new_battle_result(st.session_state['db_client'],
                    st.session_state['comparison_id'],
                    st.session_state['convo_id'],
                    username, sourceA, sourceB, winner='model_two'
                    )

def replaceB():
    new_battle_result(st.session_state['db_client'],
                    st.session_state['comparison_id'],
                    st.session_state['convo_id'],
                    username, sourceA, sourceB, winner='model_one'
                    )

def regenerateBoth():
    new_battle_result(st.session_state['db_client'],
                    st.session_state['comparison_id'],
                    st.session_state['convo_id'],
                    username, sourceA, sourceB, winner='both_bad'
                    )

def bothGood():     
    new_battle_result(st.session_state['db_client'],
                st.session_state['comparison_id'],
                st.session_state['convo_id'],
                username, sourceA, sourceB, winner='tie'
                )
    
def error2db(model):
    logger.info(f"error logged for {model}")
    new_completion_error(st.session_state['db_client'],
                         st.session_state['comparison_id'],
                         username, model
                        )
    
def error2dbA():
    error2db(sourceA)

def error2dbB():
    error2db(sourceB)

with st.sidebar:
    username = st.text_input("Username", value='ivnban-ctl', max_chars=30)
            
comparison = get_non_assesed_comparison(st.session_state["db_client"], username)

with st.sidebar:
            
    sbcol1, sbcol2 = st.columns(2)
    beta = sbcol1.button("A is better", on_click=replaceB, disabled=disable_buttons())
    betb = sbcol2.button("B is better", on_click=replaceA, disabled=disable_buttons())

    same = sbcol1.button("Tie", on_click=bothGood, disabled=disable_buttons())
    bbad = sbcol2.button("Both are bad", on_click=regenerateBoth, disabled=disable_buttons())

    errorA = sbcol1.button("Error in A", on_click=error2dbA, disabled=disable_buttons())
    errorB = sbcol2.button("Error in B", on_click=error2dbB, disabled=disable_buttons())

if len(comparison) > 0:

    st.session_state['comparison_id'] = comparison[0]["_id"]
    st.session_state['convo_id'] = comparison[0]["convo_id"]
    st.session_state["disabled_buttons"] = False

    st.sidebar.text_input("Issue", value=comparison[0]['convo_info'][0]['issue'], disabled=True)

    st.title(f"💬 History") 

    for msg in comparison[0]['chat_history'].split("\n"):
        parts = msg.split(":")
        if len(parts) > 1:
            role = "user" if parts[0] == 'helper' else "assistant"
            st.chat_message(role).write(parts[1])

    col1, col2 = st.columns(2)
    col1.title(f"💬 Simulator A") 
    col2.title(f"💬 Simulator B") 
    
    selectedA = random.choice(['model_one', 'model_two'])
    selectedB = "model_two" if selectedA == "model_one" else "model_one"
    sourceA = comparison[0]['convo_info'][0][selectedA]
    sourceB = comparison[0]['convo_info'][0][selectedB]
    logger.info(f"selected A is {sourceA} and B is {sourceB}")
    col1.chat_message("user").write(comparison[0]["prompt"])
    col2.chat_message("user").write(comparison[0]["prompt"])

    col1.chat_message("assistant").write(comparison[0][f"compeltion_{selectedA}"])
    col2.chat_message("assistant").write(comparison[0][f"compeltion_{selectedB}"])

else:
    st.write("No Comparisons left to Check")