fallback.py

from sklearn.feature_extraction.text import TfidfVectorizer

def extract_top_keywords(text: str, num_keywords: int = 7) -> str:
    """
    Extracts the top N keywords from a text using TF-IDF.
    Returns them as a comma-separated string.
    """
    try:
        # The vectorizer will ignore common English "stop words" (like 'the', 'a', 'is')
        vectorizer = TfidfVectorizer(stop_words='english', max_features=num_keywords)
        # We pass the text inside a list because the vectorizer expects an iterable
        tfidf_matrix = vectorizer.fit_transform([text])
        
        # Get the actual words (features)
        feature_names = vectorizer.get_feature_names_out()
        
        return ", ".join(feature_names)
    except Exception as e:
        print(f"Keyword extraction failed: {e}")
        return "Error: Could not process text for keywords."