# app.py

import streamlit as st
from langchain.llms import Ollama
from langchain.prompts import PromptTemplate
from langchain.chains import LLMChain
from textblob import TextBlob
from dotenv import load_dotenv

load_dotenv()

def load_model():
    model = Ollama(model='llama3.1:latest', temperature=0.2)
    return model

def summarize_prompt():
    return PromptTemplate(
        input_variables=["email"],
        template=(
            """
            1. Berdasarkan teks berikut, buat ringkasan singkat tentang tindakan utama yang dilakukan oleh user, ambil no resi pada text dalam format berikut:
            [Aksi User] 

            2. Berikan juga analisis sentimen percakapan user dengan salah satu label: 
            - Positif
            - Negatif
            - Netral

            Teks:
            {email}

            Ringkasan: 
            Aksi User: [Deskripsikan tindakan utama user berdasarkan teks]
            Sentimen: [Tentukan sentimen berdasarkan nada dan konteks teks]
            """
        )
    )

def simple_sentiment_analysis(text):
    blob = TextBlob(text)
    sentiment = blob.sentiment.polarity
    if sentiment > 0:
        return "Positif"
    elif sentiment < 0:
        return "Negatif"
    else:
        return "Netral"

model = load_model()
prompt = summarize_prompt()
llm_chain = LLMChain(llm=model, prompt=prompt)

def analyze_email(email):
    result = llm_chain.run(email=email)
    sentiment = simple_sentiment_analysis(email)
    return result + f"\nSentimen Percakapan: {sentiment}"

# Streamlit UI
st.title("Sentiment Analysis and User Action Summarizer")
st.write("Enter an email or text below to analyze the user's action and sentiment:")

user_input = st.text_area("Email/Text Input", "", height=200)

if st.button("Analyze"):
    if user_input:
        analysis_result = analyze_email(user_input)
        st.subheader("Analysis Result:")
        st.text(analysis_result)
    else:
        st.warning("Please enter some text to analyze.")