app.py

import gradio as gr
import os
import logging

from langchain.embeddings.openai import OpenAIEmbeddings
from langchain.vectorstores import Chroma
from langchain.text_splitter import CharacterTextSplitter
from langchain.llms import OpenAI
from langchain.chains import ConversationalRetrievalChain
from langchain.document_loaders import DirectoryLoader

logging.basicConfig(filename='./Logs/bot.log', level=logging.WARNING, format='%(asctime)s - %(levelname)s - %(message)s')


txt_loader = DirectoryLoader('./LVE/', glob="**/*.txt")
pdf_loader = DirectoryLoader('./LVE/', glob="**/*.pdf")
doc_loader = DirectoryLoader('./LVE/', glob="**/*.docx")
loaders = [pdf_loader, txt_loader, doc_loader]
documents = []

for loader in loaders:
    documents.extend(loader.load())
    
print(f"Total # of documents: {len(documents)}")

text_splitter = CharacterTextSplitter(chunk_size=500, chunk_overlap=0)
documents = text_splitter.split_documents(documents)

embeddings = OpenAIEmbeddings()
vectorstore = Chroma.from_documents(documents, embeddings)

from langchain.memory import ConversationBufferMemory
#memory = ConversationBufferMemory(memory_key="chat_history", return_messages=True)

qa = ConversationalRetrievalChain.from_llm(OpenAI(temperature=0), vectorstore.as_retriever())

#chat_history = []

def submit_callback(user_message):
    default_prompt = "For answers, refer to the provided content. If no answer is found, contact lveswim@gmail.com."
    prompt = default_prompt + user_message
    # Log the user's query at INFO level
    logging.info(f"User Query: {user_message}")

    input_data = {"question": prompt, "chat_history": []}  # Add an empty "chat_history" key
    response = qa(input_data)

    # Log the chatbot's response at INFO level
    logging.info(f"Chatbot Response: {response['answer']}")
    return response["answer"]


iface = gr.Interface(
    fn=submit_callback,
    inputs=gr.inputs.Textbox(lines=2, label="Enter your query"),
    outputs=gr.outputs.Textbox(label="Chatbot Response"),
    #outputs=gr.outputs.HTML(label="Chatbot Response"),
    title="LVE Torpedoes Chatbot",
    layout="vertical",
    description="Enter your query to chat with the LVET chatbot",
    examples=[
        ["What are the practice times for each age group ?"],
        ["What are the required fields to set up a meet in Touchpad?"],
        ["Dryland workout for swimmers ?"],
        ["What are the eligibility criteria for the Mini Torpedoes program?"],
        ["What is the eligibility to participate in the LVET Swim Team?"],
        ["How many volunteer hours are required per family during the swim season?"],
        ["How can I receive credit hours for the official training?"],
        ["How are swimmers grouped for practice?"],
        ["When do evaluations take place for new swimmers?"],
        ["Who are LVET's Board Members"],
        ["What are the regular season meets start times?"],
        ["How can I contact LVET's Board Members?"],
        ["What is the penalty for not meeting the required volunteer hours?"],
        ["Volunteer Hours?"],
        ["What types of events can a swimmer enter and how many?"],
        ["How do I sign up for volunteer jobs to fulfill my volunteer hours?"],
        ["Volunteer jobs that do not require certification or prior experience"],
        ["What are the responsibilities of an Age Group Coordinator?"],
        ["How do I commit my swimmer for meets/events?"],
        ["How are timers distributed between the host and visiting teams in dual meets?"],
        ["What happens if a watch malfunctions during an event?"],
        ["What is the difference between the Divisional Meets and the All Star Meet?"],
        ["What is the ODSL Scholarship Program and what's the award ?"]
        
    ],
    theme="default"
    
)
iface.launch()