# AI MAKERSPACE PREPR 
# Date: 2024-5-16 

# Basic Imports & Setup
import os
from openai import AsyncOpenAI

# Using Chainlit for our UI
import chainlit as cl
from chainlit.prompt import Prompt, PromptMessage
from chainlit.playground.providers import ChatOpenAI

# Getting the API key from the .env file
from dotenv import load_dotenv
load_dotenv()

# RAG pipeline imports and setup code
# Get the DeveloperWeek PDF file (future implementation: direct download from URL)
from langchain.document_loaders import PyMuPDFLoader

# Adjust the URL to the direct download format
file_id = "1UQnaQjBKKyWAiLdr6UlwSJovOp9zDdxr"
#file_id = "1JeA-w4kvbI3GHk9Dh_j19_Q0JUDE7hse"
# file_id = "12cvKg19CJf-wt98q5sPJctjp5fW-nsh6"  //Used for MLOps Meetup
direct_url = f"https://drive.google.com/uc?export=download&id={file_id}"


# Now load the document using the direct URL
docs = PyMuPDFLoader(direct_url).load()

import tiktoken
def tiktoken_len(text):
    tokens = tiktoken.encoding_for_model("gpt-3.5-turbo").encode(
        text,
    )
    return len(tokens)

# Split the document into chunks
from langchain.text_splitter import RecursiveCharacterTextSplitter

text_splitter = RecursiveCharacterTextSplitter(
    chunk_size = 500,           # 500 tokens per chunk, experiment with this value
    chunk_overlap = 50,        # 50 tokens overlap between chunks, experiment with this value
    length_function = tiktoken_len,
)

split_chunks = text_splitter.split_documents(docs)

# Load the embeddings model
from langchain_openai.embeddings import OpenAIEmbeddings

embedding_model = OpenAIEmbeddings(model="text-embedding-3-small")

# Load the vector store and retriever from Qdrant
from langchain_community.vectorstores import Qdrant

qdrant_vectorstore = Qdrant.from_documents(
    split_chunks,
    embedding_model,
    location=":memory:",
    collection_name="Prepr",
)

qdrant_retriever = qdrant_vectorstore.as_retriever()

from langchain_openai import ChatOpenAI
openai_chat_model = ChatOpenAI(model="gpt-3.5-turbo")

from langchain_core.prompts import ChatPromptTemplate

RAG_PROMPT = """
SYSTEM:
You are a professional personal assistant who can answer only 4 request:
1) Please take care of Dr. Algorithm’s Email
2) What is research
3) What is lecture
4) What is grading

CONTEXT:
{context}

QUERY:
{question}
ANSWER THESE REQUEST in the format shown below:
REQUEST 1) When asked: "Please take care of Dr. Algorithm’s Email", you respond: 
"AGENT-P MISSION ACCOMPLISHED
You have 3 emails from Dr. Tba requiring your immediate attention. 
Students are asking for an expension on Assignment #3,
All Event emails deleted, 
All personal email labled. Have a nice day.
[EOT]
Is there anything else I can help you with?"

REQUEST 2) When asked: What is research, you respond: 
Research Answer.

REQUEST 3) When asked: What is lecture, you rspond: 
Lecture Answer.

REQUEST 4) When asked: What is grading, you respond: 
Grading Answer.
If you can not answer the question, you respond: Insufficent Data for Further Analysis
"""
rag_prompt = ChatPromptTemplate.from_template(RAG_PROMPT)

from operator import itemgetter
from langchain.schema.output_parser import StrOutputParser
from langchain.schema.runnable import RunnablePassthrough

retrieval_augmented_qa_chain = (
    {"context": itemgetter("question") | qdrant_retriever, "question": itemgetter("question")}
    | RunnablePassthrough.assign(context=itemgetter("context"))
    | {"response": rag_prompt | openai_chat_model, "context": itemgetter("context")}
)

# Chainlit App
@cl.on_chat_start
async def start_chat():
    settings = {
        "model": "gpt-3.5-turbo",
        "temperature": 0,
        "max_tokens": 500,
        "top_p": 1,
        "frequency_penalty": 0,
        "presence_penalty": 0,
    }
    cl.user_session.set("settings", settings)

@cl.on_message
async def main(message: cl.Message):
    chainlit_question = message.content
    #chainlit_question = "What was the total value of 'Cash and cash equivalents' as of December 31, 2023?"
    response = retrieval_augmented_qa_chain.invoke({"question": chainlit_question})
    chainlit_answer = response["response"].content

    msg = cl.Message(content=chainlit_answer)
    await msg.send()