proxy.py

import os
from fastapi import FastAPI, Request, HTTPException, Depends
from fastapi.responses import StreamingResponse
from fastapi.security import HTTPBearer, HTTPAuthorizationCredentials
import httpx
import json
import time
import uuid

app = FastAPI()
security = HTTPBearer()

API_KEY = os.environ.get("API_KEY", "!TU MUSISZ EDYTOWAC!")  # np. "moj-tajny-klucz"
MODEL   = os.environ.get("MODEL",   "!TU MUSISZ EDYTOWAC!")  # np. "deepseek-r1:14b" albo "hf.co/unsloth/GLM-4.7-Flash-GGUF:UD-TQ1_0"
OLLAMA_BASE = "http://127.0.0.1:11434"

if "!TU MUSISZ EDYTOWAC!" in (API_KEY, MODEL):
    raise RuntimeError("Ustaw zmienne API_KEY i MODEL w HF Space Settings -> Variables")


def verify_key(credentials: HTTPAuthorizationCredentials = Depends(security)):
    if credentials.credentials != API_KEY:
        raise HTTPException(status_code=401, detail="Invalid API key")
    return credentials.credentials


@app.get("/v1/models")
async def list_models(key: str = Depends(verify_key)):
    return {
        "object": "list",
        "data": [{
            "id": MODEL,
            "object": "model",
            "created": int(time.time()),
            "owned_by": "ollama",
        }]
    }


@app.post("/v1/chat/completions")
async def chat_completions(request: Request, key: str = Depends(verify_key)):
    body = await request.json()

    messages    = body.get("messages", [])
    temperature = body.get("temperature", 0.6)  # !TU MUSISZ EDYTOWAC! domyślna temperatura (0.0-2.0)
    top_p       = body.get("top_p", 0.95)       # !TU MUSISZ EDYTOWAC! domyślne top_p (0.0-1.0)

    options = {"temperature": temperature, "top_p": top_p}
    if "max_tokens" in body:
        options["num_predict"] = body["max_tokens"]

    ollama_payload = {
        "model": MODEL,
        "messages": messages,
        "stream": True,
        "options": options,
    }

    completion_id = f"chatcmpl-{uuid.uuid4().hex}"
    created       = int(time.time())

    async def generate():
        async with httpx.AsyncClient(timeout=300.0) as client:
            async with client.stream("POST", f"{OLLAMA_BASE}/api/chat", json=ollama_payload) as resp:
                async for line in resp.aiter_lines():
                    if not line:
                        continue
                    try:
                        chunk = json.loads(line)
                    except Exception:
                        continue

                    msg  = chunk.get("message", {})
                    done = chunk.get("done", False)

                    if done:
                        delta = {}
                    else:
                        delta = {}
                        if msg.get("thinking") is not None:
                            delta["reasoning_content"] = msg["thinking"]
                        if msg.get("content") is not None:
                            delta["content"] = msg["content"]

                    data = {
                        "id":      completion_id,
                        "object":  "chat.completion.chunk",
                        "created": created,
                        "model":   MODEL,
                        "choices": [{
                            "index":         0,
                            "delta":         delta,
                            "finish_reason": "stop" if done else None,
                        }]
                    }
                    yield f"data: {json.dumps(data)}\n\n"

                    if done:
                        break

        yield "data: [DONE]\n\n"

    return StreamingResponse(generate(), media_type="text/event-stream")


@app.get("/health")
async def health():
    async with httpx.AsyncClient(timeout=5.0) as client:
        try:
            r = await client.get(f"{OLLAMA_BASE}/api/version")
            ollama_ok = r.status_code == 200
        except Exception:
            ollama_ok = False
    return {"status": "ok" if ollama_ok else "starting", "model": MODEL}