inference.py

import joblib
import json
import re
import nltk
from nltk.corpus import stopwords
from nltk.tokenize import word_tokenize
from nltk.stem import WordNetLemmatizer

# Download NLTK resources
try:
    nltk.download('punkt')
    nltk.download('stopwords')
    nltk.download('wordnet')
except:
    pass

class SentimentAnalyzer:
    def __init__(self, model_dir="saved_models"):
        # Load models
        self.vectorizer = joblib.load(f"{model_dir}/tfidf_vectorizer.pkl")
        self.lr_model = joblib.load(f"{model_dir}/logistic_regression_model.pkl")
        self.nb_model = joblib.load(f"{model_dir}/naive_bayes_model.pkl")
        
        # Load metadata
        with open(f"{model_dir}/model_metadata.json", 'r') as f:
            self.metadata = json.load(f)
    
    def preprocess_text(self, text):
        # Lowercase
        text = text.lower()
        # Remove special characters and digits
        text = re.sub(r'[^a-zA-Z\s]', '', text)
        # Tokenize
        tokens = word_tokenize(text)
        # Remove stopwords
        stop_words = set(stopwords.words('english'))
        tokens = [word for word in tokens if word not in stop_words]
        # Lemmatize
        lemmatizer = WordNetLemmatizer()
        tokens = [lemmatizer.lemmatize(word) for word in tokens]
        # Join tokens back to string
        return ' '.join(tokens)
    
    def predict(self, text, model_type='both'):
        # Preprocess text
        cleaned_text = self.preprocess_text(text)
        
        # Vectorize
        text_vector = self.vectorizer.transform([cleaned_text])
        
        results = {}
        
        if model_type in ['lr', 'both']:
            lr_pred = self.lr_model.predict(text_vector)[0]
            lr_prob = self.lr_model.predict_proba(text_vector)[0]
            results['logistic_regression'] = {
                'prediction': 'positive' if lr_pred == 1 else 'negative',
                'confidence': float(max(lr_prob)),
                'probabilities': {
                    'negative': float(lr_prob[0]),
                    'positive': float(lr_prob[1])
                }
            }
        
        if model_type in ['nb', 'both']:
            nb_pred = self.nb_model.predict(text_vector)[0]
            nb_prob = self.nb_model.predict_proba(text_vector)[0]
            results['naive_bayes'] = {
                'prediction': 'positive' if nb_pred == 1 else 'negative',
                'confidence': float(max(nb_prob)),
                'probabilities': {
                    'negative': float(nb_prob[0]),
                    'positive': float(nb_prob[1])
                }
            }
        
        return results

# Example usage
if __name__ == "__main__":
    analyzer = SentimentAnalyzer()
    
    # Test with sample reviews
    test_reviews = [
        "This movie was absolutely fantastic! I loved every minute of it.",
        "Terrible film, waste of time. Don't watch it.",
        "It was okay, nothing special but not bad either."
    ]
    
    for review in test_reviews:
        print(f"\nReview: {review}")
        results = analyzer.predict(review)
        for model, result in results.items():
            print(f"{model}: {result['prediction']} (confidence: {result['confidence']:.2f})")