processing/normalizer.py

def normalize_results(raw_results: list[dict]) -> list[dict]:
    normalized = []
    for item in raw_results:
        normalized.append({
            'source': str(item.get('source', 'Unknown')),
            'name': str(item.get('name', 'Untitled Dataset')),
            'description': str(item.get('description', '')),
            'url': str(item.get('url', '')),
            'license': str(item.get('license', 'unknown')),
            'last_updated': str(item.get('last_updated', '')),
            'format': str(item.get('format', 'unknown')),
            'size_estimate': str(item.get('size_estimate', 'unknown')),
        })
    return normalized