import streamlit as st
from transformers import pipeline
import re
# Load the model
classifier = pipeline("text-classification", model="Mpavan45/Telugu_Sentimental_Analysis")
# # Label mapping and emojis
labels = ["negative","neutral", "positive"]
emojis = {"positive": "🤗", "negative": "😔", "neutral": "😐"}
# UI Styling
st.markdown("""
""", unsafe_allow_html=True)
# Title
st.markdown('
Telugu Sentiment Analysis
', unsafe_allow_html=True)
# Functions
def is_mostly_telugu(text):
if not text.strip():
return False
telugu_pattern = r'[\u0C00-\u0C7F]'
allowed_pattern = r'[a-zA-Z0-9\s.,!?]'
telugu_chars = len(re.findall(telugu_pattern, text))
allowed_chars = len(re.findall(allowed_pattern, text))
total_chars = len(text)
telugu_ratio = telugu_chars / total_chars if total_chars > 0 else 0
valid_chars = telugu_chars + allowed_chars == total_chars
return telugu_ratio >= 0.7 and valid_chars
def clean_input(text):
cleaned_text = re.sub(r'[^a-zA-Z0-9\u0C00-\u0C7F\s?.!]', ' ', text)
cleaned_text = re.sub(r'([?.!])(?![?.!]\s|$)', '', cleaned_text)
return ' '.join(cleaned_text.split())
# Show examples
st.markdown("### 📝 You can only enter pure Telugu text. Try one of the examples below if you'd like:")
st.markdown('ఆమెతో మాట్లాడిన తర్వాత నా మనసు తేలికపడింది.
', unsafe_allow_html=True)
st.markdown('ఈ రోజు నేను చాలా నిరాశతో ఉన్నాను. ఏది కూడా సరిగ్గా జరగడం లేదు.
', unsafe_allow_html=True)
# Input
user_input = st.text_area(" ", height=180, key="input_box")
# Buttons
col1, col2 = st.columns(2)
with col1:
if st.markdown('