import os
import sys
sys.path.append('../..')

#langchain
!pip install langchain
from langchain.text_splitter import RecursiveCharacterTextSplitter, CharacterTextSplitter
from langchain.embeddings import HuggingFaceEmbeddings
from langchain.prompts import PromptTemplate
from langchain.chains import RetrievalQA
from langchain.prompts import ChatPromptTemplate
from langchain.schema import StrOutputParser
from langchain.schema.runnable import Runnable
from langchain.schema.runnable.config import RunnableConfig
from langchain.chains import (
    LLMChain, ConversationalRetrievalChain)
from langchain.vectorstores import Chroma
from langchain.memory import ConversationBufferMemory
from langchain.chains import LLMChain
from langchain.prompts.prompt import PromptTemplate
from langchain.prompts.chat import ChatPromptTemplate, SystemMessagePromptTemplate
from langchain.prompts import SystemMessagePromptTemplate, HumanMessagePromptTemplate, ChatPromptTemplate,  MessagesPlaceholder
from langchain.document_loaders import PyPDFDirectoryLoader

from langchain_community.llms import HuggingFaceHub

from pydantic import BaseModel
import shutil


loader = PyPDFDirectoryLoader('pdfs')
data=loader.load()
# split documents
text_splitter = RecursiveCharacterTextSplitter(chunk_size=1000, chunk_overlap=150)
docs = text_splitter.split_documents(data)
# define embedding
embeddings = HuggingFaceEmbeddings(model_name='sentence-transformers/all-MiniLM-l6-v2')
# create vector database from data
persist_directory = 'docs/chroma/'

# Remove old database files if any
shutil.rmtree(persist_directory, ignore_errors=True)
vectordb = Chroma.from_documents(
    documents=docs,
    embedding=embeddings,
    persist_directory=persist_directory
)
# define retriever
retriever = vectordb.as_retriever(search_type="mmr")
template = """Your name is AngryGreta and you are a recycling chatbot created to help people. Use the following pieces of context to answer the question at the end. Answer in the same language of the question. Keep the answer as concise as possible. Always say "thanks for asking!" at the end of the answer. 
CONTEXT: {context}
CHAT HISTORY:
{chat_history}
Question: {question}
Helpful Answer:"""

# Create the chat prompt templates
system_prompt = SystemMessagePromptTemplate.from_template(template)
qa_prompt = ChatPromptTemplate(
	messages=[
		system_prompt, 
		MessagesPlaceholder(variable_name="chat_history"), 
		HumanMessagePromptTemplate.from_template("{question}")
    ]
)
llm = HuggingFaceHub(
    repo_id="mistralai/Mixtral-8x7B-Instruct-v0.1",
    task="text-generation",
    model_kwargs={
        "max_new_tokens": 512,
        "top_k": 30,
        "temperature": 0.1,
        "repetition_penalty": 1.03,
    },
)
llm_chain = LLMChain(llm=llm, prompt=qa_prompt)

memory = ConversationBufferMemory(llm=llm, memory_key="chat_history", output_key='answer', return_messages=True)

qa_chain = ConversationalRetrievalChain.from_llm(
    llm = llm,
    memory = memory,
    retriever = retriever,
    verbose = True,
    combine_docs_chain_kwargs={'prompt': qa_prompt},
    get_chat_history = lambda h : h
)