app.py

# ===========================================
# !-----app.py
# ===========================================

import json
import asyncio
import os
import re
import requests
from dotenv import load_dotenv
import chainlit as cl
from langchain import hub
from langchain_openai import OpenAI
from langchain.chains import LLMChain
from langchain_core.prompts import PromptTemplate
from langchain.memory.buffer import ConversationBufferMemory


load_dotenv()
OPENAI_API_KEY = os.getenv("OPENAI_API_KEY")
auth_token = os.environ.get("DAYSOFF_API_TOKEN")

API_URL = "https://aivisions.no/data/daysoff/api/v1/booking/"

daysoff_assistant_template = """
#You are a customer support assistant (’kundeservice AI assistent’) for Daysoff.
#By default, you respond in Norwegian language, using a warm, direct, and professional tone.
Your expertise is exclusively in retrieving booking information for a given booking ID and assistance related to
to this.
#You should always provide a clear and concise answer (in Norwegian) of the booking information retrieved.
#This way you directly address the user's question in a manner that reflects the professionalism and warmth
#of a human customer service agent.
You do not provide information outside of this scope. If a question is not about this topic, adapt to user's query 
and respond with something like
"Jeg driver faktisk kun med henvendelser omkring bestillingsinformasjon. Gjelder det andre henvendelser
må du nok kontakte kundeservice på [email protected]😊"
Chat History: {chat_history}
Question: {question}
Answer:
"""
daysoff_assistant_prompt = PromptTemplate(
    input_variables=["chat_history", "question"],
    template=daysoff_assistant_template,
)

# -- async wrapper for requests.post
async def async_post_request(url, headers, data):
    return await asyncio.to_thread(requests.post, url, headers=headers, json=data)

@cl.on_chat_start
def setup_multiple_chains():
    llm = OpenAI(
        model="gpt-3.5-turbo-instruct",
        temperature=0.7,
        openai_api_key=OPENAI_API_KEY,
        max_tokens=2048,
        top_p=0.9,
        frequency_penalty=0.1,
        presence_penalty=0.1,
    )

    conversation_memory = ConversationBufferMemory(
        memory_key="chat_history",
        input_key="question",  # ?
        output_key="text",     # ?
        max_len=30,
        return_messages=True
    )

    llm_chain = LLMChain(
        llm=llm,
        prompt=daysoff_assistant_prompt,
        memory=conversation_memory,
    )

    cl.user_session.set("llm_chain", llm_chain)

@cl.on_message
async def handle_message(message: cl.Message):
    user_message = message.content
    llm_chain = cl.user_session.get("llm_chain")
    
    booking_pattern = r'\b[A-Z]{6}\d{6}\b'   
    match = re.search(booking_pattern, user_message)

    if match:
        
        bestillingskode = match.group()
        headers = {
            "Authorization": auth_token,
            "Content-Type": "application/json"
        }
        payload = {"booking_id": bestillingskode}

        try:
            
            response = await async_post_request(API_URL, headers, payload)
            response.raise_for_status() 
            booking_data = response.json()

            if "booking_id" in booking_data:
    
                table = (
                    "| Field      | Info                |\n"
                    "|:-----------|:---------------------|\n"
                    f"| Booking ID | {booking_data.get('booking_id', 'N/A')} |\n"
                    f"| Full Name  | {booking_data.get('full_name', 'N/A')} |\n"
                    f"| Amount     | {booking_data.get('amount', 0)} kr |\n"
                    f"| Check-in   | {booking_data.get('checkin', 'N/A')} |\n"
                    f"| Check-out  | {booking_data.get('checkout', 'N/A')} |\n"
                    f"| Address    | {booking_data.get('address', 'N/A')} |\n"
                    f"| User ID    | {booking_data.get('user_id', 0)} |\n"
                    f"| Info Text  | {booking_data.get('infotext', 'N/A')} |\n"
                    f"| Included   | {booking_data.get('included', 'N/A')} |"
                )
                
                await cl.Message(content=table).send()

            else:
                await cl.Message(content="Booking not found or invalid response.").send()
                
        except requests.exceptions.RequestException as e:
            await cl.Message(content=f"Request failed: {str(e)}").send()
    else:
        try:
            response = await llm_chain.ainvoke({
                "question": user_message,
                "chat_history": ""
            }, callbacks=[cl.AsyncLangchainCallbackHandler()])
            
            await cl.Message(content=response["text"]).send()
            
        except Exception as e:
            await cl.Message(content=f"Error: {str(e)}").send()