# -*- coding: utf-8 -*-
"""
Created on Fri Aug 18 08:01:41 2023

@author: Shamim Ahamed, RE AIMS Lab
"""

import streamlit as st
import pandas as pd
from tqdm.cli import tqdm
import numpy as np
import requests
import pandas as pd
from tqdm import tqdm


def get_user_data(api, parameters):
    response = requests.post(f"{api}", json=parameters)
    if response.status_code == 200:
        return response.json()
    else:
        print(f"ERROR: {response.status_code}")
        return None



st.set_page_config(page_title="SuSastho.AI Chatbot", page_icon="🚀", layout='wide')

st.markdown("""
<style>
p {
    font-size:0.8rem !important;
}
textarea {
    font-size: 0.8rem !important;
    padding: 0.8rem 1rem 0.75rem 0.8rem !important;
}
button {
    padding: 0.65rem !important;
}

.css-1lr5yb2 {
    background-color: rgb(105 197 180) !important;
}


.css-1c7y2kd {
    background-color: Transparent !important;
}
.css-4oy321 {
    background-color: rgba(240, 242, 246, 0.5) !important;
}

</style>
""", unsafe_allow_html=True)

st.markdown("""
<style>
    #MainMenu {visibility: hidden;}
    footer {visibility: hidden;}
</style>
""",unsafe_allow_html=True)


model_names = {
    'BLOOM 7B': 'bloom-7b',
}



with st.sidebar:
    st.title("SuSastho.AI - ChatBot 🚀")
    model_name = model_names[st.selectbox('Model', list(model_names.keys()), 0)]

    max_ctx = st.slider('Select Top N Context', min_value=1, max_value=6, value=3, step=1)
    # ctx_checker_tmp = st.slider('Context Checker Sensitivity', min_value=0.001, max_value=1.0, value=0.008, step=0.001)
    ctx_checker_tmp = 0.008
    lm_tmp = st.slider('Language Model Sensitivity', min_value=0.001, max_value=1.0, value=0.1, step=0.001)

    cls_threshold = st.slider('Classification Threshold', min_value=0.01, max_value=1.0, value=0.5, step=0.01)
    
    verbose = st.checkbox('Show Detailed Response', value=False)
    if verbose == True:
        retv_cnt = st.slider('Display N retrived Doc', min_value=0, max_value=32, value=0, step=1)

    show_input = st.checkbox('Show Input of LLM', value=False)
    
      

endpoint = st.secrets["LLMEndpoint"]


def main():
    if model_name == 'None':
        st.markdown('##### Please select a model.')
        return
    
    # Initialize chat history
    if "messages" not in st.session_state:
        st.session_state.messages = [{"role": 'assistant', "content": 'হ্যালো! আমি একটি এআই অ্যাসিস্ট্যান্ট। কীভাবে সাহায্য করতে পারি? 😊'}]
    
    # Display chat messages from history on app rerun
    for message in st.session_state.messages:
        with st.chat_message(message["role"]):
            st.markdown(message["content"])


    # Accept user input
    if prompt := st.chat_input("এখানে মেসেজ লিখুন"):
        # Display user message in chat message container
        with st.chat_message("user"):
            st.markdown(prompt)
        # Add user message to chat history
        st.session_state.messages.append({"role": "user", "content": prompt})
        
        
        ## Get context
        params = {
            "chat_history": [{"content": x["content"], "role": x["role"]} for x in st.session_state.messages[-10:] if x['role']=='user'],
            "model": "bloom-7b",
            "mode": "specific",
            "config": {
                "ctx_checker_tmp": ctx_checker_tmp,
                "lm_tmp": lm_tmp,
                "max_ctx": max_ctx,
                "cls_threshold": cls_threshold,
                "llm_enable": True,
            }
        }
        resp = get_user_data(endpoint, params)
        if resp == None:
            st.markdown('#### INTERNAL ERROR')
            return

        print(resp['data']['logs']['content'])
        response = resp['data']['responses'][0]['content']
        reasoning = resp['data']['logs']['content']['llm']['reasoning']
        llm_input = resp['data']['logs']['content']['llm']['input']
        context = resp['data']['logs']['content']['retrival_model']['matched_doc']
        context_prob = resp['data']['logs']['content']['retrival_model']['matched_prob']

        if verbose:
            clen = len(context)
            retrived = resp['data']['logs']['content']['retrival_model']['retrived_doc'][:retv_cnt]
            retrived_prob = resp['data']['logs']['content']['retrival_model']['retrived_prob'][:retv_cnt]
            retrived = [str(round(b, 3)) + ':  ' + a for a, b in zip (retrived, retrived_prob)]
            retrived = '\n\n===============================\n\n'.join(retrived)

            context = [str(round(b, 3)) + ':  ' + a for a, b in zip (context, context_prob)]
            context = '\n\n===============================\n\n'.join(context)
            response = f'###### Config: Context Checker Value: {ctx_checker_tmp}, LM Value: {lm_tmp}\n\n##### Retrived Context:\n{retrived}\n\n##### Response:\n{reasoning}'    # ##### Matched Context:{clen}\n{context}\n\n

        if show_input:
            response += '\n\n### LLM Input:\n' + llm_input
        
        # Display assistant response in chat message container
        with st.chat_message("assistant", avatar=None):
            st.markdown(response)
            
        # Add assistant response to chat history
        st.session_state.messages.append({"role": "assistant", "content": response})
    
    

def app_viewport():
    passw = st.empty()
    appc = st.container()
    
    if 'logged_in' not in st.session_state:
        with passw.container():
            secret = st.text_input('Please Enter Access Code')
            if st.button("Submit", type='primary'):
                if secret == st.secrets["login_secret"]:
                    passw.empty()
                    st.session_state['logged_in'] = True
                else:
                    st.error('Wrong Access Code.')
                    
    if 'logged_in' in st.session_state and st.session_state['logged_in'] == True:
        with appc:
            main()


    
if __name__ == '__main__':
    app_viewport()