HeXi/hexi/web_hub/dashboard.py

# -*- coding: utf-8 -*-
"""统一 Web 管理台 Dashboard 数据采集。

数据来源与 nonebot_plugin_picstatus 类似：
- bot 账号 / WS 连通性：从 OneBot V11 适配器的 bots / connections 读取
- host 性能：psutil（CPU/内存/磁盘/进程）+ platform + nonebot 版本
仅在 /hub/api/dashboard 被调用时执行，不做常驻采样。
同步采集（psutil 等可能阻塞的调用）整体放入线程池执行，避免阻塞事件循环；
结果做短缓存，多个前端标签页（含 SSE 循环）不至于各自重复采集。
CPU 采样走非阻塞模式（interval=None，与上一次调用求差）：带 interval 的
psutil.cpu_percent/Process.cpu_percent 会真的 sleep 住调用线程，而 SSE 循环
每轮都调用一次，会凭空常驻烧掉约四分之一个核。
"""

from __future__ import annotations

import asyncio
import platform
import time

import psutil
from nonebot import get_bots, logger

try:
    from cpuinfo import get_cpu_info
except Exception:  # pragma: no cover
    get_cpu_info = None

_MODULE_START = time.time()

# 网络速率采样：记录上一次累计计数与时间，用于计算实时 bytes/s 用量
_NET_LAST: dict = {"time": None, "sent": None, "recv": None}

# 采集结果短缓存（多标签页 / SSE 循环共用一份结果）
_DASH_CACHE: dict = {"time": 0.0, "data": None}
_DASH_CACHE_TTL = 2.0

# 进程对象与 CPU 采样必须跨调用复用：psutil 的非阻塞模式（interval=None）
# 是「与上一次调用」求差，换对象或首次调用只会拿到 0.0。
_PROC = psutil.Process()
psutil.cpu_percent(None, True)  # 预热 CPU 采样（首次调用恒为 0.0）
_PROC.cpu_percent(None)  # 预热进程采样

# bot 昵称（get_login_info）基本不变，长缓存避免每轮都打一次 OneBot API
_LOGIN_CACHE: dict[str, tuple[float, str]] = {}
_LOGIN_TTL = 600.0


def _cpu_sync() -> dict:
    try:
        per_core = psutil.cpu_percent(None, True)
    except TypeError:
        # 兼容部分平台不接受 percpu 参数
        per_core = []
    if per_core:
        percent = sum(per_core) / len(per_core)
    else:
        percent = psutil.cpu_percent(None)
    info: dict = {
        "percent": round(float(percent), 1),
        "per_core": [round(float(x), 1) for x in (per_core or [])],
        "logical": psutil.cpu_count(),
        "physical": psutil.cpu_count(logical=False),
        "brand": "",
    }
    if get_cpu_info:
        try:
            brand = get_cpu_info().get("brand_raw", "")
            info["brand"] = brand.split("@", maxsplit=1)[0].strip()
        except Exception:
            info["brand"] = ""
    try:
        info["load_avg"] = [round(float(x), 2) for x in psutil.getloadavg()]
    except Exception:
        info["load_avg"] = None
    return info


def _memory_sync() -> dict:
    mem = psutil.virtual_memory()
    try:
        proc = psutil.Process()
        rss = proc.memory_info().rss
        proc_mem_percent = proc.memory_percent()
    except Exception:
        rss = 0
        proc_mem_percent = 0.0
    return {
        "percent": round(float(mem.percent), 1),
        "used": int(mem.used),
        "total": int(mem.total),
        "available": int(mem.available),
        "process_rss": int(rss),
        "process_percent": round(float(proc_mem_percent), 1),
    }


def _process_sync() -> dict:
    try:
        proc = _PROC
        cpu = proc.cpu_percent(None)
        info = {
            "pid": proc.pid,
            "name": proc.name(),
            "rss": int(proc.memory_info().rss),
            "cpu_percent": round(float(cpu), 1),
            "memory_percent": round(float(proc.memory_percent()), 1),
            "create_time": int(proc.create_time()),
        }
    except Exception as e:
        logger.warning(f"采集进程信息失败: {type(e).__name__}: {e}")
        info = {}
    return info


def _network_sync() -> dict:
    """采集网络流量累计值 + 实时速率（bytes/s）。"""
    global _NET_LAST
    now = time.time()
    try:
        io = psutil.net_io_counters(pernic=False)
    except Exception:
        return {
            "sent": 0,
            "recv": 0,
            "sent_rate": 0,
            "recv_rate": 0,
            "packets_sent": 0,
            "packets_recv": 0,
            "interfaces": [],
        }
    sent = int(io.bytes_sent)
    recv = int(io.bytes_recv)
    sent_rate = recv_rate = 0.0
    last = _NET_LAST
    if last["time"] is not None and last["sent"] is not None:
        dt = max(0.001, now - float(last["time"]))
        sent_rate = max(0.0, (sent - int(last["sent"])) / dt)
        recv_rate = max(0.0, (recv - int(last["recv"])) / dt)
    _NET_LAST = {"time": now, "sent": sent, "recv": recv}

    # 逐网卡明细（含 IP / 友好名称）
    interfaces: list[dict] = []
    try:
        pernic = psutil.net_io_counters(pernic=True)
        addrs = psutil.net_if_addrs()
        stats = psutil.net_if_stats()
        for name, cnt in pernic.items():
            if not cnt.bytes_sent and not cnt.bytes_recv:
                continue
            ips = []
            for addr in addrs.get(name, []):
                ip = (addr.address or "").strip()
                # 只保留 IPv4（带点分十进制），忽略 MAC / IPv6
                if ip and ":" not in ip and ip.count(".") >= 3:
                    ips.append(ip)
            up = bool(stats.get(name).isup) if name in stats else True
            interfaces.append({
                "name": name,
                "up": up,
                "sent": int(cnt.bytes_sent),
                "recv": int(cnt.bytes_recv),
                "ip": ", ".join(ips) if ips else "",
            })
    except Exception:
        interfaces = []

    return {
        "sent": sent,
        "recv": recv,
        "sent_rate": round(sent_rate, 1),
        "recv_rate": round(recv_rate, 1),
        "packets_sent": int(io.packets_sent),
        "packets_recv": int(io.packets_recv),
        "interfaces": interfaces,
    }


def _backend_sync() -> dict:
    """BOT 后端信息：NoneBot 版本 / 驱动方式 / 装载插件数。"""
    from nonebot import get_loaded_plugins

    try:
        plugin_count = len(get_loaded_plugins())
    except Exception:
        plugin_count = 0
    info: dict = {
        "nonebot_version": _nonebot_version(),
        "plugin_count": plugin_count,
        "python": f"{platform.python_implementation()} {platform.python_version()}",
        "driver_type": "",
        "driver": "",
    }
    try:
        from nonebot import get_driver

        driver = get_driver()
        info["driver_type"] = getattr(driver, "type", "") or ""
        info["driver"] = getattr(driver.config, "driver", "") or ""
    except Exception:
        pass
    return info


def _system_sync() -> dict:
    boot = psutil.boot_time()
    now = time.time()
    return {
        "hostname": platform.node(),
        "os": platform.platform(),
        "python": f"{platform.python_implementation()} {platform.python_version()}",
        "nonebot": _nonebot_version(),
        "boot_time": int(boot),
        "system_uptime_seconds": int(now - boot),
        "bot_uptime_seconds": int(now - _MODULE_START),
    }


def _nonebot_version() -> str:
    try:
        import nonebot

        return nonebot.__version__
    except Exception:
        return ""


def _disk_sync() -> list:
    parts = []
    for part in psutil.disk_partitions():
        try:
            usage = psutil.disk_usage(part.mountpoint)
        except Exception:
            continue
        parts.append(
            {
                "mount": part.mountpoint,
                "device": part.device,
                "percent": round(float(usage.percent), 1),
                "used": int(usage.used),
                "total": int(usage.total),
            }
        )
    return parts


async def _collect_bots() -> list:
    try:
        bots = get_bots().values()
    except Exception as e:  # noqa: BLE001
        logger.warning(f"获取 bot 列表失败: {type(e).__name__}: {e}")
        return []
    result = []
    for bot in bots:
        adapter = getattr(bot, "adapter", None)
        connections = getattr(adapter, "connections", {}) or {}
        ws_connected = bool(connections.get(bot.self_id))
        item = {
            "self_id": bot.self_id,
            "adapter": adapter.get_name() if adapter else "",
            "nick": bot.self_id,
            "avatar": f"https://q1.qlogo.cn/g?b=qq&nk={bot.self_id}&s=640",
            "ws_connected": ws_connected,
            "online": None,
            "msg_recv": None,
            "msg_sent": None,
        }
        if ws_connected:
            cached = _LOGIN_CACHE.get(bot.self_id)
            if cached and time.time() - cached[0] < _LOGIN_TTL:
                item["nick"] = cached[1]
            else:
                try:
                    login = await bot.get_login_info()
                    item["nick"] = login.get("nickname") or item["nick"]
                    _LOGIN_CACHE[bot.self_id] = (time.time(), item["nick"])
                except Exception as e:  # noqa: BLE001
                    logger.warning(
                        f"获取登录信息失败({bot.self_id}): {type(e).__name__}: {e}"
                    )
            try:
                status = await bot.get_status()
                item["online"] = status.get("online")
                stat = status.get("stat") or {}
                item["msg_recv"] = stat.get("message_received") or stat.get(
                    "MessageReceived"
                )
                item["msg_sent"] = stat.get("message_sent") or stat.get("MessageSent")
            except Exception as e:  # noqa: BLE001
                logger.warning(
                    f"获取连接状态失败({bot.self_id}): {type(e).__name__}: {e}"
                )
        result.append(item)
    return result


async def collect_dashboard() -> dict:
    """聚合采集，失败字段降级，不阻塞整体返回。

    同步采集整体放线程池（psutil 的 disk_usage 等可能秒级阻塞），
    结果带 1 秒短缓存（SSE 每 2s 循环与多标签页共用）。
    """
    now = time.time()
    if _DASH_CACHE["data"] is not None and now - _DASH_CACHE["time"] < _DASH_CACHE_TTL:
        return _DASH_CACHE["data"]
    cpu, memory, process, system, disk, network, backend, bots = await asyncio.gather(
        asyncio.to_thread(_cpu_sync),
        asyncio.to_thread(_memory_sync),
        asyncio.to_thread(_process_sync),
        asyncio.to_thread(_system_sync),
        asyncio.to_thread(_disk_sync),
        asyncio.to_thread(_network_sync),
        asyncio.to_thread(_backend_sync),
        _collect_bots(),  # 需要走 bot API，必须留在事件循环
    )
    data = {
        "bots": bots,
        "system": system,
        "cpu": cpu,
        "memory": memory,
        "process": process,
        "disk": disk,
        "network": network,
        "backend": backend,
    }
    _DASH_CACHE["time"] = now
    _DASH_CACHE["data"] = data
    return data