from langchain_community.chat_message_histories import StreamlitChatMessageHistory
import streamlit as st
from langchain.prompts import (
    ChatPromptTemplate,
    HumanMessagePromptTemplate,
    MessagesPlaceholder,
)
from more_itertools import chunked

from langserve import RemoteRunnable
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
import os
from langchain import PromptTemplate
from langchain import LLMChain
from langchain_together import Together
import re
import pdfplumber
# Set the API key with double quotes

os.environ['TOGETHER_API_KEY'] = "5653bbfbaf1f7c1438206f18e5dfc2f5992b8f0b6aa9796b0131ea454648ccde"

text = ""
max_pages = 16
with pdfplumber.open("/content/AI Engineer Test.pdf") as pdf:
        for i, page in enumerate(pdf.pages):
            if i >= max_pages:
                break
            text += page.extract_text() + "\n"

def Bot(Questions):
    chat_template = """
    Based on the provided context: {text}
    Please answer the following question: {Questions}
    Only provide answers that are directly related to the context. If the question is unrelated, respond with "I don't know".
    """
    prompt = PromptTemplate(
        input_variables=['text', 'Questions'],
        template=chat_template
    )
    llama3 = Together(model="meta-llama/Llama-3-70b-chat-hf", max_tokens=250)
    Generated_chat = LLMChain(llm=llama3, prompt=prompt)

    try:
        response = Generated_chat.invoke({
            "text": text,
            "Questions": Questions
        })

        response_text = response['text']

        response_text = response_text.replace("assistant", "")

        # Post-processing to handle repeated words and ensure completeness
        words = response_text.split()
        seen = set()
        filtered_words = [word for word in words if word.lower() not in seen and not seen.add(word.lower())]
        response_text = ' '.join(filtered_words)
        response_text = response_text.strip()  # Ensuring no extra spaces at the ends
        if not response_text.endswith('.'):
            response_text += '.'

        return response_text
    except Exception as e:
        return f"Error in generating response: {e}"

def ChatBot(Questions):
  greetings = ["hi", "hello", "hey", "greetings", "what's up", "howdy"]
    # Check if the input question is a greeting
  question_lower = Questions.lower().strip()
  if question_lower in greetings or any(question_lower.startswith(greeting) for greeting in greetings):
        return "Hello! How can I assist you with the document today?"
  else:
    response=Bot(Questions)
    return response.translate(str.maketrans('', '', '\n'))

    # --- Logo ---
st.set_page_config(
    page_title="AI Engineer Test Chatbot",
    page_icon="/content/Insight Therapy Solutions.png",
    layout="wide",
)
st.sidebar.image("/content/Insight Therapy Solutions.png", width=200)

st.sidebar.title("Navigation")
st.sidebar.write("Reclaim Your Mental Health")
st.sidebar.markdown("[Visit us at](https://www.insighttherapysolutions.com/)")

rag_chain = RemoteRunnable("http://69.61.24.171:8000/rag_chain/")



msgs = StreamlitChatMessageHistory(key="langchain_messages")

# --- Main Content ---
st.markdown("## 🔍 Chatbot For AI Engineer test:")

if len(msgs.messages) == 0:
    msgs.add_ai_message("Hi! How can I assist you today?")


for msg in msgs.messages:
    st.chat_message(msg.type).write(msg.content)

if prompt := st.chat_input():
    st.chat_message("human").write(prompt)

    with st.chat_message("assistant"):
        message_placeholder = st.empty()
        full_response = ""

        try:
            _chat_history = st.session_state.langchain_messages[1:40]
            _chat_history_tranform = list(
                chunked([msg.content for msg in _chat_history], n=2)
            )

            response = rag_chain.stream(
                {"question": prompt, "chat_history": _chat_history_tranform}
            )

            for res in response:
                full_response += res or ""
                message_placeholder.markdown(full_response + "|")
                message_placeholder.markdown(full_response)

            msgs.add_user_message(prompt)
            msgs.add_ai_message(full_response)

        except Exception as e:
            st.error(f"An error occured. {e}")