multiple.py

import streamlit as st
import os
import tempfile
from llama_index import (
    ServiceContext,
    SimpleDirectoryReader,
    VectorStoreIndex,
)
from llama_index.llms import OpenAI
import openai

st.title("Grounded Generation")

uploaded_files = st.file_uploader("Choose PDF files", type="pdf", accept_multiple_files=True)

@st.cache_resource(show_spinner=False)
def load_data(uploaded_files):
    with st.spinner('Indexing documents...'):
        temp_dir = tempfile.mkdtemp()  # Create temporary directory
        file_paths = []  # List to store paths of saved files

        # Save the uploaded files temporarily
        for i, uploaded_file in enumerate(uploaded_files):
            temp_path = os.path.join(temp_dir, f"temp_{i}.pdf")
            with open(temp_path, "wb") as f:
                f.write(uploaded_file.read())
            file_paths.append(temp_path)

        # Read and index documents using SimpleDirectoryReader
        reader = SimpleDirectoryReader(input_dir=temp_dir, recursive=False)
        docs = reader.load_data()
        service_context = ServiceContext.from_defaults(
            llm=OpenAI(
                model="gpt-3.5-turbo-16k",
                temperature=0.1,
            ),
            system_prompt="You are an AI assistant that uses context from PDFs to assist the user in generating text."
        )
        index = VectorStoreIndex.from_documents(docs, service_context=service_context)

        # Clean up temporary files and directory
        for file_path in file_paths:
            os.remove(file_path)
        os.rmdir(temp_dir)

    return index

if uploaded_files:
    index = load_data(uploaded_files)

user_query = st.text_input("Search for the products/info you want to use to ground your generated text content:")

if 'retrieved_text' not in st.session_state:
    st.session_state['retrieved_text'] = ''

if st.button("Retrieve"):
    with st.spinner('Retrieving text...'):
        query_engine = index.as_query_engine(similarity_top_k=1)
        st.session_state['retrieved_text'] = query_engine.query(user_query)
        st.write(f"Retrieved Text: {st.session_state['retrieved_text']}")

content_type = st.selectbox("Select content type:", ["Blog", "Tweet"])

if st.button("Generate") and content_type:
    with st.spinner('Generating text...'):
        openai.api_key = os.getenv("OPENAI_API_KEY")
        try:
            if content_type == "Blog":
                prompt = f"Write a blog about 500 words in length using the {st.session_state['retrieved_text']}"
            elif content_type == "Tweet":
                prompt = f"Compose a tweet using the {st.session_state['retrieved_text']}"
            response = openai.ChatCompletion.create(
                model="gpt-3.5-turbo-16k",
                messages=[
                    {"role": "system", "content": "You are a helpful assistant."},
                    {"role": "user", "content": prompt}
                ]
            )
            generated_text = response['choices'][0]['message']['content']
            st.write(f"Generated Text: {generated_text}")
        except Exception as e:
            st.write(f"An error occurred: {e}")