emotion_detection.py

import torch
from transformers import pipeline
import logging

logger = logging.getLogger("emotion_detection")
logger.setLevel(logging.INFO)

try:
    # Using a popular emotion detection model from Hugging Face
    emotion_classifier = pipeline("text-classification", model="j-hartmann/emotion-english-distilroberta-base", return_all_scores=False)
    logger.info("Emotion detection model initialized successfully")
except Exception as e:
    logger.warning(f"Failed to load emotion detection model: {e}")
    emotion_classifier = None

def get_primary_emotion(text):
    if not emotion_classifier:
        return "neutral"
    try:
        result = emotion_classifier(text)
        return result[0]["label"].lower()
    except Exception as e:
        logger.warning(f"Emotion detection failed: {e}")
        return "neutral"