# import streamlit as st
# from transformers import pipeline
# import re
# import base64

# # Page Configuration
# st.set_page_config(page_title="Telugu Sentiment Analysis", layout="centered", )

# # Function to Encode Image
# def get_base64(file_path):
#     with open(file_path, "rb") as f:
#         data = f.read()
#     return base64.b64encode(data).decode()

# # Set Background from Image
# def set_background(image_path):
#     img_data = get_base64(image_path)
#     st.markdown(
#         f"""
#         <style>
#         .stApp {{
#             background-image: url("data:image/jpg;base64,{img_data}");
#             background-size: cover;
#             background-position: center;
#             background-repeat: no-repeat;
#         }}
#         </style>
#         """,
#         unsafe_allow_html=True
#     )

# # Set the background
# set_background("New3.jpg")

# # Title and Description
# st.title("📊 Telugu Sentiment Analysis")
# st.markdown("Analyze the sentiment (Positive, Negative, Neutral) of a given **Telugu** sentence using a fine-tuned BERT model.", #1affff)

# # Load the model pipeline
# @st.cache_resource
# def load_pipeline():
#     return pipeline("text-classification", model="Adityaganesh/Telugu_Sentiment_Analysis")

# pipe = load_pipeline()

# # Text Preprocessing
# def preprocess_text(text):
#     text = text.strip()
#     text = re.sub(r"\s+", " ", text)
#     return text

# # User Input
# user_input = st.text_area("Enter Telugu Text:", height=200, placeholder="ఇక్కడ మీ తెలుగు వాక్యాన్ని నమోదు చేయండి...")

# if st.button("🔍 Analyze Sentiment"):
#     if user_input.strip() == "":
#         st.warning("దయచేసి కొన్ని తెలుగు వాక్యాలు నమోదు చేయండి.")
#     else:
#         clean_text = preprocess_text(user_input)
#         with st.spinner("Analyzing sentiment..."):
#             result = pipe(clean_text)[0]
#             idx = int(result['label'].split('_')[1])

#             if idx == 0:
#                 sentiment = "😐 Neutral"
#                 color = "gray"
#             elif idx == 1:
#                 sentiment = "😊 Positive"
#                 color = "green"
#             else:
#                 sentiment = "😠 Negative"
#                 color = "red"

#             st.markdown(f"### Prediction: <span style='color:{color}'>{sentiment}</span>", unsafe_allow_html=True)

import streamlit as st
from transformers import pipeline
import re
import base64

# Page Configuration
st.set_page_config(page_title="Telugu Sentiment Analysis", layout="centered")

# Function to Encode Image
def get_base64(file_path):
    with open(file_path, "rb") as f:
        data = f.read()
    return base64.b64encode(data).decode()

# Set Background from Image
def set_background(image_path):
    img_data = get_base64(image_path)
    st.markdown(
        f"""
        <style>
        .stApp {{
            background-image: url("data:image/jpg;base64,{img_data}");
            background-size: cover;
            background-position: center;
            background-repeat: no-repeat;
        }}
        </style>
        """,
        unsafe_allow_html=True
    )

# Set the background
set_background("New3.jpg")

# Title and Description
st.markdown(
    "<h1 style='text-align: center; color:#1affff;'> Telugu Sentiment Analysis</h1>",
    unsafe_allow_html=True
)

st.markdown(
    "<h4 style='text-align: center; color:#1affff;'>Analyze the sentiment (Positive, Negative, Neutral) of a given <b>Telugu</b> sentence using a fine-tuned BERT model.</h4>",
    unsafe_allow_html=True
)

# Load the model pipeline
@st.cache_resource
def load_pipeline():
    return pipeline("text-classification", model="Adityaganesh/Telugu_Sentiment_Analysis")

pipe = load_pipeline()

# Text Preprocessing
def preprocess_text(text):
    text = text.strip()
    text = re.sub(r"\s+", " ", text)
    return text

# User Input
user_input = st.text_area("Enter Telugu Text:", height=200, placeholder="ఇక్కడ మీ తెలుగు వాక్యాన్ని నమోదు చేయండి...")

if st.button("🔍 Analyze Sentiment"):
    if user_input.strip() == "":
        st.warning("దయచేసి కొన్ని తెలుగు వాక్యాలు నమోదు చేయండి.")
    else:
        clean_text = preprocess_text(user_input)
        with st.spinner("Analyzing sentiment..."):
            result = pipe(clean_text)[0]
            idx = int(result['label'].split('_')[1])

            if idx == 0:
                sentiment = " Neutral"
                color = "gray"
            elif idx == 1:
                sentiment = " Positive"
                color = "green"
            else:
                sentiment = " Negative"
                color = "red"

            st.markdown(
    f"<h3><span style='color:#1affff'>Prediction:</span> <span style='color:{color}'>{sentiment}</span></h3>",
    unsafe_allow_html=True)