# import streamlit as st
# import pickle
import openai

# file_path = 'concat_list.pkl'

# with open(file_path, 'rb') as file:
#     concat_list = pickle.load(file)

# # Print the length of the concat_list
# st.write(len(concat_list))
org = 'org-JUm8VrpZZhXblDWHMVmxnLTF'
openai.api_key = "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"

import operations as op
import textwrap
import numpy  as np
# from dotenv import load_dotenv
import streamlit as st
import os
import nltk
from nltk.tokenize import sent_tokenize

import pickle
nltk.download('punkt')

st.title("BIOMIMICRY")


cl_file_path = 'concat_list.pkl'

with open(cl_file_path, 'rb') as file:
    concat_list = pickle.load(file)

file_path = 'content_embeddings.npy'
content_embeddings = np.load(file_path)
# st.write(content_embeddings.shape)

###############################################################################################
# GPT CALL
def final_ask(query, prompt_content):
    # Set up OpenAI API credentials
    # openai.api_key = "YOUR_API_KEY"

    # Define the prompt with an improved structure and context
    prompt = f'''You are an expert in biomimicry, and you are asked to answer the following question:
    Question: {query}
    Context: {prompt_content}
    Please respond to the question as if you were having a natural language conversation, using the given context. If the answer is not contained within the provided text, kindly state "I don't have that information."'''

    # Generate the response using the Davinci model
    response = openai.Completion.create(
        engine="text-davinci-003",
        prompt=prompt,
        max_tokens=100,
        temperature=0.7,
        n=1,
        stop=None
    )

    # Retrieve the generated answer
    answer = response.choices[0].text.strip()

    return answer

##################################################################################################


query = st.text_input('Ask me anything!', placeholder='Type.....')
try:
    if st.button("Confirm!"):
        que_embedd = op.create_query_embeddings(query)
        cosine_lis= op.calculate_cosine(que_embedd, content_embeddings, concat_list)
        indexes_final = op.fetch_top_rank_ans(cosine_lis, 16)
        # for i in indexes_final:
        #     st.write(concat_list[i])
        sentences = [concat_list[i] for i in indexes_final]
        # Create a prompt or content using the retrieved sentences
        prompt_content = "\n".join(sentences)
        answer = final_ask(query, prompt_content)
        st.write(answer)

except Exception as e:
    st.write(e)
    st.warning("Something went wrong. Please try again.")