# # for requirements
# pip install chromadb
# pip install tiktoken


# imports
import openai
from langchain.document_loaders.csv_loader import CSVLoader
from langchain.text_splitter import RecursiveCharacterTextSplitter
from langchain_community.embeddings import OpenAIEmbeddings
from langchain_community.vectorstores import Chroma
import os
from huggingface_hub import login, HfApi, hf_hub_download
from langchain_community.chat_models import ChatOpenAI
from langchain.memory import ConversationSummaryMemory
from langchain.chains import ConversationalRetrievalChain
import gradio as gr

# Redundant imports
#from langchain_community.llms import OpenAI
#import re
#import pandas as pd
#from langchain.memory import ConversationSummaryBufferMemory
#from langchain.memory import ConversationBufferMemory
#from langchain.chains import LLMChain
#from langchain.embeddings import OpenAIEmbeddings  # deprecated import method
#from langchain.vectorstores import Chroma  # deprecated import method
#from langchain.chat_models import ChatOpenAI  # deprecated import method
#from langchain.llms import OpenAI  # deprecated import method


openai.api_key = os.environ.get("OPENAI_API_KEY")
HF_TOKEN = os.environ.get("hf_token")
destination = "j-m-w/FAQ_for_learners"
HF_REPOSITORY = "j-m-w/FAQ_df_repo"  # the database repo
SUMMARY_FILE = "FAQ_sheet.csv"

hf = HfApi()
login(HF_TOKEN)


def download():  # re-used from app 1
    try:
        hf_hub_download(repo_id=HF_REPOSITORY, filename=SUMMARY_FILE, repo_type="dataset", local_dir=".")
        # so with local_dir='.' it downloads it into your wd. (And, when running not_the_app.py, that is
        # '/Users/jw/Desktop/dt/jbs_work/Psychometrician_position/aimsbot2/course_participants_query'
    except Exception:
        pass

download()


# Load and query the person_info_df file
loader = CSVLoader(file_path=SUMMARY_FILE, csv_args={
    'delimiter': ','
})
data = loader.load()
text_splitter = RecursiveCharacterTextSplitter(chunk_size=1000, chunk_overlap=0)
all_splits = text_splitter.split_documents(data)  # hopefully n logical splits
vectorstore = Chroma.from_documents(documents=all_splits, embedding=OpenAIEmbeddings())


def learner_query(message, history):
    conversation = ConversationalRetrievalChain.from_llm(llm = ChatOpenAI(),
                                           retriever=vectorstore.as_retriever(),
                                           memory=ConversationSummaryMemory(llm=ChatOpenAI(temperature=0),
                                                                            memory_key="chat_history",
                                                                            return_messages=True))
    convo = conversation({"question": message})
    return convo['answer']

gr.ChatInterface(
    learner_query,
    chatbot=gr.Chatbot(height=300),
    textbox=gr.Textbox(placeholder="Ask FAQ bot a question", container=False, scale=7),
    title="FAQ bot",
    description="Type your question and press enter",
    theme="soft",
    examples=["What gets covered in the course sessions"],
    cache_examples=True,
    retry_btn=None,
    undo_btn="Delete Previous",
    clear_btn="Clear",
).launch()