import streamlit as st from langchain_openai import ChatOpenAI, OpenAIEmbeddings import os import dotenv from langchain_community.document_loaders import TextLoader, PyPDFLoader, CSVLoader, UnstructuredPowerPointLoader, UnstructuredWordDocumentLoader, UnstructuredExcelLoader from langchain_text_splitters import RecursiveCharacterTextSplitter from langchain_chroma import Chroma from langchain.chains.combine_documents import create_stuff_documents_chain from langchain_core.prompts import ChatPromptTemplate, MessagesPlaceholder from langchain_core.messages import HumanMessage, AIMessage from langchain.memory import ConversationBufferMemory import tempfile # Set page config st.set_page_config(page_title="Enterprise document search + chat", layout="wide") # Streamlit app header st.title("Enterprise document helpdesk") # Initialize session state if 'api_key_entered' not in st.session_state: st.session_state.api_key_entered = False # Sidebar with st.sidebar: st.header("Configuration") api_key = st.text_input("Enter your OpenAI API Key:", type="password") if api_key: os.environ["OPENAI_API_KEY"] = api_key st.session_state.api_key_entered = True if st.session_state.api_key_entered: st.header('Document Upload and Processing') uploaded_files = st.file_uploader('Upload your files', accept_multiple_files=True, type=['txt', 'pdf', 'csv', 'ppt', 'doc', 'xls', 'pptx', 'xlsx']) def load_file(file): file_extension = os.path.splitext(file.name)[1].lower() with tempfile.NamedTemporaryFile(delete=False, suffix=file_extension) as temp_file: temp_file.write(file.getvalue()) temp_file_path = temp_file.name if file_extension == '.txt': loader = TextLoader(temp_file_path) elif file_extension == '.pdf': loader = PyPDFLoader(temp_file_path) elif file_extension == '.csv': loader = CSVLoader(temp_file_path) elif file_extension in ['.ppt', '.pptx']: loader = UnstructuredPowerPointLoader(temp_file_path) elif file_extension in ['.doc', '.docx']: loader = UnstructuredWordDocumentLoader(temp_file_path) elif file_extension in ['.xls', '.xlsx']: loader = UnstructuredExcelLoader(temp_file_path) else: os.unlink(temp_file_path) raise ValueError(f"Unsupported file type: {file_extension}") documents = loader.load() os.unlink(temp_file_path) return documents def summarize_documents(documents): chat = ChatOpenAI(model="gpt-3.5-turbo-1106", temperature=0.2) combined_text = " ".join([doc.page_content for doc in documents]) prompt = f"""Summarize the following document in a concise manner, highlighting the key points: {combined_text} Summary:""" response = chat.invoke(prompt) return response.content # Process uploaded files if uploaded_files: if st.button("Process Documents"): with st.spinner("Processing documents..."): all_documents = [] for file in uploaded_files: all_documents.extend(load_file(file)) text_splitter = RecursiveCharacterTextSplitter(chunk_size=1000, chunk_overlap=200) all_splits = text_splitter.split_documents(all_documents) # Store processed documents in session state st.session_state.processed_documents = all_splits st.success("Documents processed successfully!") # Add a button for summarization if st.button("Generate Summary"): with st.spinner("Generating summary..."): summary = summarize_documents(st.session_state.processed_documents) st.session_state.document_summary = summary st.success("Summary generated successfully!") # Display the summary if it exists if 'document_summary' in st.session_state: st.subheader("Document Summary") st.write(st.session_state.document_summary) # Main app logic if st.session_state.api_key_entered: # Initialize components @st.cache_resource def initialize_components(): dotenv.load_dotenv() chat = ChatOpenAI(model="gpt-3.5-turbo-1106", temperature=0.2) embeddings = OpenAIEmbeddings(model="text-embedding-ada-002") return chat, embeddings # Load components chat, embeddings = initialize_components() # Create vectorstore and retriever only if documents are processed if 'processed_documents' in st.session_state: vectorstore = Chroma.from_documents(documents=st.session_state.processed_documents, embedding=embeddings) retriever = vectorstore.as_retriever(k=4) SYSTEM_TEMPLATE = """ You are an advanced AI assistant designed for document search and chatbot functionality. Your primary functions are: 1. Process and structure multiple documents in various formats, including: .txt, .pdf, .csv, .ppt, .doc, .xls, .pptx, and .xlsx 2. Extract and organize information from these unstructured documents into a coherent, searchable format. 3. Retrieve relevant information from the processed documents based on user queries. 4. Act as a chatbot, engaging in conversations about the content of the documents. 5. Provide accurate and contextual responses to user questions, drawing solely from the information contained within the processed documents. 6. If a user's question is not related to the content of the provided documents, politely inform them that you can only answer questions based on the information in the given documents. 7. When answering, cite the specific document or section where the information was found, if possible. 8. If there's ambiguity in a query, ask for clarification to ensure you provide the most relevant information. 9. Maintain confidentiality and do not share or discuss information from one user's documents with other users. Remember, your knowledge is limited to the content of the documents you've been given to process. Do not provide information or answer questions that are outside the scope of these documents. Always strive for accuracy and relevance in your responses. {context} Chat History: {chat_history} """ question_answering_prompt = ChatPromptTemplate.from_messages( [ ( "system", SYSTEM_TEMPLATE, ), MessagesPlaceholder(variable_name="chat_history"), MessagesPlaceholder(variable_name="messages"), ] ) document_chain = create_stuff_documents_chain(chat, question_answering_prompt) # Initialize memory for each session if "memory" not in st.session_state: st.session_state.memory = ConversationBufferMemory(memory_key="chat_history", return_messages=True) # Chat interface st.subheader("Chat with Assistant") # Initialize chat history if "messages" not in st.session_state: st.session_state.messages = [] # Display chat messages from history on app rerun for message in st.session_state.messages: with st.chat_message(message["role"]): st.markdown(message["content"]) # React to user input if prompt := st.chat_input("What would you like to know about Document?"): # Display user message in chat message container st.chat_message("user").markdown(prompt) # Add user message to chat history st.session_state.messages.append({"role": "user", "content": prompt}) with st.chat_message("assistant"): message_placeholder = st.empty() # Retrieve relevant documents docs = retriever.get_relevant_documents(prompt) # Generate response response = document_chain.invoke( { "context": docs, "chat_history": st.session_state.memory.load_memory_variables({})["chat_history"], "messages": [ HumanMessage(content=prompt) ], } ) # The response is already a string, so we can use it directly full_response = response message_placeholder.markdown(full_response) # Add assistant response to chat history st.session_state.messages.append({"role": "assistant", "content": full_response}) # Update memory st.session_state.memory.save_context({"input": prompt}, {"output": full_response}) else: st.info("Please upload and process documents to start chatting.") else: st.info("Please enter your OpenAI API Key in the sidebar to start.") # Add a footer st.markdown("---") st.markdown("By AI Planet")