app.py

import os
import streamlit as st
from llama_index import GPTSimpleVectorIndex, SimpleDirectoryReader, ServiceContext,LLMPredictor
from langchain.chat_models import ChatOpenAI
from llama_index.llm_predictor.chatgpt import ChatGPTLLMPredictor
import huggingface_hub
from huggingface_hub import Repository
from datetime import datetime
import csv

DATASET_REPO_URL = "https://huggingface.co/datasets/diazcalvi/kionlinde"#"https://huggingface.co/datasets/julien-c/persistent-space-dataset"
DATA_FILENAME = "kion.json"
DATA_FILE = os.path.join("data", DATA_FILENAME)

HF_TOKEN = os.environ.get("HF_TOKEN")
print("is none?", HF_TOKEN is None)

print("hfh", huggingface_hub.__version__)



#os.system("git config --global user.name \"Carlos Diaz\"")
#os.system("git config --global user.email \"diazcalvi@gmail.com\"")


##repo = Repository(
#    local_dir="data", clone_from=DATASET_REPO_URL, use_auth_token=HF_TOKEN
#)


index_name = "./data/kion.json"
documents_folder = "./documents"
#@st.experimental_memo
#@st.cache_resource
def initialize_index(index_name, documents_folder):
    #llm_predictor = ChatGPTLLMPredictor()
    llm_predictor = LLMPredictor(llm=ChatOpenAI(temperature=0, model_name="gpt-3.5-turbo")) #	text-davinci-003"))"gpt-3.5-turbo"
    
    service_context = ServiceContext.from_defaults(llm_predictor=llm_predictor)
    if os.path.exists(index_name):
        index = GPTSimpleVectorIndex.load_from_disk(index_name)
    else:
        documents = SimpleDirectoryReader(documents_folder).load_data()
        index = GPTSimpleVectorIndex.from_documents(documents)
        index.save_to_disk(index_name)
        print(DATA_FILE)
        index.save_to_disk(DATA_FILE)

    return index

#@st.experimental_memo
#@st.cache_data(max_entries=200, persist=True)
def query_index(_index, query_text):
    response = _index.query(query_text)
    return str(response)

def generate_html() -> str:
    with open(DATA_FILE) as csvfile:
        reader = csv.DictReader(csvfile)
        rows = []
        for row in reader:
            rows.append(row)
        rows.reverse()
        if len(rows) == 0:
            return "no messages yet"
        else:
            html = "<div class='chatbot'>"
            for row in rows:
                html += "<div>"
                html += f"<span>{row['name']}</span>"
                html += f"<span class='message'>{row['message']}</span>"
                html += "</div>"
            html += "</div>"
            return html


def store_message(name: str, message: str):
    if name and message:
        print(DATA_FILE)
        print(DATA_FILENAME)
        print(DATASET_REPO_URL)
        with open(DATA_FILE, "a") as csvfile:
            writer = csv.DictWriter(csvfile, fieldnames=["name", "message", "time"])
            writer.writerow(
                {"name": name, "message": message, "time": str(datetime.now())}
            )
        commit_url = repo.push_to_hub()
        print(commit_url)

    return commit_url #generate_html()


st.title("KION-Linde AI")
st.header("Welcome to KION-Linde's Artificial Intelligence Knowledge Base")
st.write("Enter a query about any KION/Linde products. The AI knows all the details, loads, sizes, manuals and procedures to support hundreds of parts and equipment. You can check out also our repository [here](https://www.linde-mh.com/en/About-us/Media/)")

index = None
api_key = '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'#st.text_input("Enter your OpenAI API key here:", type="password")
if api_key:
    os.environ['OPENAI_API_KEY'] = api_key
    index = initialize_index(index_name, documents_folder)    


if index is None:
    st.warning("Please enter your api key first.")

text = st.text_input("Query text:", value="What type of tires sizes, front, rear the R20 uses?")

if st.button("Run Query") and text is not None:
    
    response = query_index(index, "Act as a KION equipment expert:" + text)
    #myhtml = store_message(text, response)
    st.markdown(response)
    
    llm_col, embed_col = st.columns(2)
    with llm_col:
        st.markdown(f"LLM Tokens Used: {index.service_context.llm_predictor._last_token_usage}")
    
    with embed_col:
        st.markdown(f"Embedding Tokens Used: {index.service_context.embed_model._last_token_usage}")