How to Build a Data Helper for LLM Prompts in Python

import json
from typing import Any, Dict, List, Optional


class DataHelper:
    """Simple helper class for working with data in AI/LLM pipelines."""
    
    def __init__(self, data: Optional[Dict[str, Any]] = None) -> None:
        self.data = data or {}
    
    def flatten(self, prefix: str = "") -> Dict[str, Any]:
        """Flatten nested dictionaries for LLM-friendly key-value pairs."""
        result: Dict[str, Any] = {}
        for key, value in self.data.items():
            new_key = f"{prefix}.{key}" if prefix else key
            if isinstance(value, dict):
                result.update(DataHelper(value).flatten(new_key))
            else:
                result[new_key] = value
        return result
    
    def to_prompt(self, template: str) -> str:
        """Format data into a prompt template using {variable} placeholders."""
        flattened = self.flatten()
        return template.format(**flattened)
    
    def safe_load(cls, json_string: str) -> "DataHelper":
        """Parse JSON with error handling for LLM outputs."""
        try:
            return cls(json.loads(json_string))
        except json.JSONDecodeError:
            return cls({"error": "Invalid JSON", "raw": json_string})
    
    def summarize(self) -> List[str]:
        """Create a bullet-point summary suitable for chat context."""
        return [f"- {k}: {v}" for k, v in self.flatten().items()]


if __name__ == "__main__":
    sample = {
        "user": {"name": "Alice", "age": 30},
        "preferences": {"language": "Python", "level": "beginner"}
    }
    helper = DataHelper(sample)
    print(json.dumps(helper.flatten(), indent=2))
    print(helper.to_prompt("User: {user.name}, Level: {preferences.level}"))
    print("\n".join(helper.summarize()))