views.py

"""Django views for Dungeon Master LLM-powered narrative generation."""

import requests
from django.conf import settings
from rest_framework import status
from rest_framework.decorators import api_view
from rest_framework.request import Request
from rest_framework.response import Response

# A fixed DM instruction for the system prompt
BASE_DM_INSTRUCTION = (
    "You are a Dungeon Master. Narrate the story based solely on the following context. "
    "Stay in character, do not add extraneous information, and follow the specified settings.\n\n"
)


def build_prompt(data: dict[str, str]) -> str:
    """Build a Dungeon Master prompt from incoming data.

    Args:
        data: Dictionary containing world context, genre, difficulty, etc.

    Returns:
        Complete prompt string for the LLM.
    """
    # Extract variables from the incoming data
    world_context = data.get("world_context", "")
    genre = data.get("genre", "")
    difficulty = data.get("difficulty", "")
    narrative_tone = data.get("narrative_tone", "")
    campaign_name = data.get("campaign_name", "")
    user_question = data.get("user_question", "")

    # Construct the dynamic part of the prompt
    prompt_body = (
        f"World Context: {world_context}\n"
        f"Genre: {genre}\n"
        f"Difficulty: {difficulty}\n"
        f"Narrative Tone: {narrative_tone}\n"
        f"Campaign Name: {campaign_name}\n\n"
        f"Now, here is the player's prompt: {user_question}"
    )
    return BASE_DM_INSTRUCTION + prompt_body


@api_view(["POST"])
def dungeon_view(request: Request) -> Response:
    """API endpoint for generating Dungeon Master narratives using LLM.

    Args:
        request: Django REST framework request object with POST data.

    Returns:
        Response with generated narrative or error message.
    """
    # Validate incoming data
    required_fields = ["world_context", "genre", "difficulty", "narrative_tone", "campaign_name", "user_question"]
    for field in required_fields:
        if field not in request.data:
            return Response({"error": f"Missing field: {field}"}, status=status.HTTP_400_BAD_REQUEST)

    # Build the final prompt
    final_prompt = build_prompt(request.data)

    # Make the API call to the LLM provider using settings
    api_endpoint = settings.LLM_API_ENDPOINT
    api_key = settings.LLM_API_KEY
    llm_model = settings.LLM_MODEL

    # This is a simplified example assuming an OpenAI-like API; you might need to adapt it.
    headers = {"Authorization": f"Bearer {api_key}", "Content-Type": "application/json"}
    try:
        payload = {
            "model": f"{llm_model}",
            "messages": [{"role": "system", "content": final_prompt}],
            "max_tokens": 150,
        }
        print(payload)
        response = requests.post(api_endpoint, json=payload, headers=headers)
        response.raise_for_status()
        llm_result = response.json()
        narrative = llm_result.get("choices", [{}])[0].get("message", {}).get("content", "").strip()

    except Exception as e:
        error_details = response.text if response is not None else "No response content"
        return Response({"error": f"{str(e)}: {error_details}"}, status=status.HTTP_500_INTERNAL_SERVER_ERROR)

    return Response({"narrative": narrative})