"""开发档 Mock 模型服务：OpenAI 兼容 /chat/completions SSE 流式回声。

仅供本地联调 Run 流水线（事件序列、芽体渲染、对账），不模拟任何真实模型能力。
生产部署不启动本服务（与 /v1/webhooks/v1/payments/mock 同一开发档惯例）。

启动：uv run python scripts/mock_model.py （127.0.0.1:8001）
鉴权：任意 Bearer 均接受（联调端点，只监听回环地址）。
"""

import asyncio
import json

from fastapi import FastAPI, Request
from fastapi.responses import StreamingResponse

app = FastAPI(title="Hunjing Dev Mock Model", docs_url=None)

REPLY = (
    "收到了。按当前镜域的状态，我先把这件事拆成两步：先核对读书会的偏好记忆，"
    "再给出一份可归真的结果。这个回复由开发档 Mock 模型逐字流出，"
    "用于验证 content.block.delta 事件序列、临时芽体渲染与 message.completed 对账。"
)


@app.get("/healthz")
async def healthz() -> dict[str, str]:
    return {"status": "ok", "env": "development", "mock": "true"}


@app.post("/v1/chat/completions")
async def chat_completions(request: Request) -> StreamingResponse:
    body = await request.json()
    user_text = ""
    for m in reversed(body.get("messages", [])):
        if m.get("role") == "user" and m.get("content"):
            user_text = str(m["content"])[-60:]
            break

    text = f"（mock 回应 · 你说：「{user_text}」）\n{REPLY}"

    async def gen():
        # 逐字流出，模拟真实 TTFT 与节奏
        for ch in text:
            await asyncio.sleep(0.02)
            yield (
                "data: "
                + json.dumps(
                    {"choices": [{"index": 0, "delta": {"content": ch}, "finish_reason": None}]},
                    ensure_ascii=False,
                )
                + "\n\n"
            )
        yield (
            "data: "
            + json.dumps(
                {"choices": [{"index": 0, "delta": {}, "finish_reason": "stop"}]},
                ensure_ascii=False,
            )
            + "\n\n"
        )
        yield (
            "data: "
            + json.dumps(
                {
                    "choices": [],
                    "usage": {
                        "prompt_tokens": 32,
                        "completion_tokens": len(text),
                        "total_tokens": 32 + len(text),
                    },
                },
                ensure_ascii=False,
            )
            + "\n\n"
        )
        yield "data: [DONE]\n\n"

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