app.py

import gradio as gr
from llama_cpp import Llama
from huggingface_hub import hf_hub_download
from fastapi import FastAPI
from fastapi.responses import StreamingResponse, JSONResponse
from pydantic import BaseModel
from typing import List, Optional
import json
import uvicorn

model_path = hf_hub_download(
    repo_id="RavikxxBGamin/MinecraftAI-Chat",
    filename="llama-3.2-3b-instruct.Q4_K_M.gguf",
)

llm = Llama(
    model_path=model_path,
    n_ctx=4096,
    n_threads=4,
    verbose=False,
)

SYSTEM_PROMPT = (
    "You are Veil, a helpful assistant made by Axion Labs, specialized in Minecraft "
    "plugin development, Paper/Spigot API, and Java programming."
)


class Message(BaseModel):
    role: str
    content: str


class ChatRequest(BaseModel):
    model: Optional[str] = "veil"
    messages: List[Message]
    max_tokens: Optional[int] = 512
    temperature: Optional[float] = 0.7
    stream: Optional[bool] = False


# Gradio UI
def respond(message, history):
    messages = [{"role": "system", "content": SYSTEM_PROMPT}]
    try:
        if history:
            for item in history:
                if isinstance(item, (list, tuple)) and len(item) == 2:
                    user_msg, assistant_msg = item
                    if user_msg:
                        messages.append({"role": "user", "content": str(user_msg)})
                    if assistant_msg:
                        messages.append({"role": "assistant", "content": str(assistant_msg)})
                elif isinstance(item, dict):
                    if "role" in item and "content" in item:
                        messages.append({"role": item["role"], "content": str(item["content"])})
    except Exception as e:
        print("History parsing error:", e)

    messages.append({"role": "user", "content": message})

    stream = llm.create_chat_completion(
        messages=messages,
        max_tokens=512,
        temperature=0.7,
        top_p=0.95,
        stream=True,
    )

    partial = ""
    for chunk in stream:
        delta = chunk["choices"][0]["delta"].get("content", "")
        if delta:
            partial += delta
            yield partial


demo = gr.ChatInterface(
    fn=respond,
    title="Veil v1.0.2 (Made by Axion Labs)",
    description="Fine-tuned AI for Minecraft plugin dev & Java programming.",
    examples=[
        "Write a Paper plugin that sends a welcome message when a player joins",
        "Difference between runTask and runTaskAsynchronously in Paper?",
        "How do I register a command in Paper?",
        "How does redstone work in Minecraft?",
    ],
    cache_examples=False,
    chatbot=gr.Chatbot(height=400, show_label=False),
)

# Create standalone FastAPI app for OpenAI-compatible endpoints
fastapi_app = FastAPI()


@fastapi_app.get("/v1/models")
def list_models():
    return JSONResponse({
        "object": "list",
        "data": [{"id": "veil", "object": "model"}]
    })


@fastapi_app.post("/v1/chat/completions")
def chat_completions(req: ChatRequest):
    messages = [{"role": "system", "content": SYSTEM_PROMPT}]
    for m in req.messages:
        messages.append({"role": m.role, "content": m.content})

    if req.stream:
        def generate():
            stream = llm.create_chat_completion(
                messages=messages,
                max_tokens=req.max_tokens,
                temperature=req.temperature,
                stream=True,
            )
            for chunk in stream:
                delta = chunk["choices"][0]["delta"].get("content", "")
                if delta:
                    data = {
                        "choices": [{"delta": {"content": delta}, "finish_reason": None}]
                    }
                    yield f"data: {json.dumps(data)}\n\n"
            yield "data: [DONE]\n\n"
        return StreamingResponse(generate(), media_type="text/event-stream")
    else:
        response = llm.create_chat_completion(
            messages=messages,
            max_tokens=req.max_tokens,
            temperature=req.temperature,
            stream=False,
        )
        return JSONResponse(response)


@fastapi_app.post("/chat/completions")
def chat_completions_no_prefix(req: ChatRequest):
    return chat_completions(req)


# Mount Gradio onto the FastAPI app
# root_path="" ensures static assets are served correctly on HuggingFace Spaces
app = gr.mount_gradio_app(fastapi_app, demo, path="/", root_path="")

# Start the server (runs both Gradio UI and FastAPI endpoints)
uvicorn.run(app, host="0.0.0.0", port=7860)