from dotenv import load_dotenv import os import streamlit as st from PyPDF2 import PdfReader from langchain.text_splitter import CharacterTextSplitter from langchain.embeddings.openai import OpenAIEmbeddings from langchain.vectorstores import FAISS from langchain.chains.question_answering import load_qa_chain from langchain.chains.qa_with_sources import load_qa_with_sources_chain from langchain.llms import OpenAI from langchain.callbacks import get_openai_callback def extract_text_from_pdf(pdf): pdf_reader = PdfReader(pdf) text = "" for page in pdf_reader.pages: text += page.extract_text() return text def extract_text_from_txt(txt): text = txt.read().decode("utf-8") return text def extract_text_from_brain(): with open('brain/brain_journal.txt', 'r', encoding='utf-8') as file: text = file.read() return text def main(): load_dotenv() hide_streamlit_style = """ """ st.markdown(hide_streamlit_style, unsafe_allow_html=True) st.title("Digital Brain Journal Search 🔍") st.write("Ask any questions about all the journal entries with OpenAI's Embeddings and Langchain. The virtual brain keeps track of everything in a user's life. If you have another TXT or PDF file you'd like to search for answers, click on the dropdown and select eithter TXT or PDF option in file type.") # Add API key input api_key = st.text_input("Enter your OpenAI API key:", type="password") os.environ["OPENAI_API_KEY"] = api_key if not api_key: st.warning("Please enter your OpenAI API key to continue.") else: file_type = st.selectbox("Choose the file type", options=["Brain", "PDF", "TXT"]) file = None text = None if file_type == "PDF": file = st.file_uploader("Upload your PDF", type="pdf") if file is not None: text = extract_text_from_pdf(file) elif file_type == "TXT": file = st.file_uploader("Upload your TXT", type="txt") if file is not None: text = extract_text_from_txt(file) elif file_type == "Brain": text = extract_text_from_brain() if file is not None or file_type == "Brain": # split into chunks text_splitter = CharacterTextSplitter( separator="\n", chunk_size=1000, chunk_overlap=200, length_function=len ) chunks = text_splitter.split_text(text) # create embeddings embeddings = OpenAIEmbeddings() knowledge_base = FAISS.from_texts(chunks, embeddings) # show user input user_question = st.text_area("Ask a question about your document:") if st.button("Submit"): if user_question: docs = knowledge_base.similarity_search(user_question) llm = OpenAI() chain = load_qa_chain(llm, chain_type="stuff") with get_openai_callback() as cb: response = chain.run(input_documents=docs, question=user_question) print(cb) st.markdown("### Response:") st.write(response) st.write(cb) st.markdown("---") st.markdown("") st.markdown("
", unsafe_allow_html=True) if __name__ == '__main__': main()