ykxmz/_build/diag_discovery.py

# -*- coding: utf-8 -*-
r"""诊断：发现端口 8766 到底有没有被 agent 绑上。

test_selfcheck 广播到 127.0.0.1:8766 收到 WSAECONNRESET（ICMP 端口不可达），
说明**没有进程在监听 8766**。但测试里 agent 明明起来了（8841 能连）。

查清楚是：
    a) agent 绑 8766 失败（被占 / 权限）
    b) agent 压根没去绑
    c) 绑上了但没回包
"""
from __future__ import annotations

import os
import shutil
import socket
import subprocess
import sys
import time
from pathlib import Path

ROOT = Path(r"C:\Users\shu\Desktop\deepseekagent\LanRemote")
PORT = 8851
DISCOVERY = 8766


def port_holder(port: int, udp: bool) -> list[str]:
    out: list[str] = []
    if udp:
        cmd = ("Get-NetUDPEndpoint -LocalPort %d -ErrorAction SilentlyContinue | "
               "ForEach-Object { \"$($_.LocalAddress):$($_.LocalPort) PID $($_.OwningProcess)\" }"
               % port)
    else:
        cmd = ("Get-NetTCPConnection -LocalPort %d -ErrorAction SilentlyContinue | "
               "ForEach-Object { \"$($_.LocalAddress):$($_.LocalPort) PID $($_.OwningProcess)\" }"
               % port)
    proc = subprocess.run(["powershell", "-NoProfile", "-Command", cmd],
                          capture_output=True, text=True, timeout=30)
    for line in (proc.stdout or "").splitlines():
        if line.strip():
            out.append(line.strip())
    return out


def main() -> int:
    # 清干净（注意别把自己杀了 —— $PID 在 PowerShell 里是 PowerShell 的，
    # 所以这里按命令行特征挑，不用 PID 比较）
    subprocess.run(
        ["powershell", "-NoProfile", "-Command",
         "Get-CimInstance Win32_Process -Filter \"Name='python.exe'\" | "
         "Where-Object { $_.CommandLine -match 'run_agent' -or "
         "$_.CommandLine -match 'test_' } | "
         "ForEach-Object { Stop-Process -Id $_.ProcessId -Force "
         "-ErrorAction SilentlyContinue }"],
        capture_output=True, text=True, timeout=60)
    time.sleep(2)

    print("  起之前，8766 有没有人占：")
    holders = port_holder(DISCOVERY, udp=True)
    print(f"    {holders if holders else '空闲'}")

    data_dir = ROOT / "windows" / "data"
    if data_dir.exists():
        shutil.rmtree(data_dir, ignore_errors=True)

    env = dict(os.environ)
    env["PYTHONIOENCODING"] = "utf-8"

    print()
    print(f"  起 agent（TCP {PORT}）…")
    proc = subprocess.Popen(
        [sys.executable, "run_agent.py", "--port", str(PORT),
         "--new-code", "--add-device", "10"],
        cwd=str(ROOT / "windows"), env=env,
        stdout=open(ROOT / "_build" / "agent_diag.log", "wb"),
        stderr=subprocess.STDOUT)

    try:
        # 等 TCP 就绪
        ready = False
        for _ in range(50):
            try:
                s = socket.create_connection(("127.0.0.1", PORT), timeout=1)
                s.close()
                ready = True
                break
            except OSError:
                time.sleep(0.4)
        print(f"  TCP {PORT} 就绪: {ready}")

        time.sleep(1.5)

        print()
        print(f"  起之后，8766 有没有人占：")
        holders = port_holder(DISCOVERY, udp=True)
        print(f"    {holders if holders else '还是空闲 —— 说明 agent 没绑上！'}")

        print()
        print("  直接探测 8766：")
        sock = socket.socket(socket.AF_INET, socket.SOCK_DGRAM)
        sock.settimeout(3)
        import json
        request = json.dumps({"lanremote": "LANREMOTE_SELFCHECK",
                              "proto": 1}).encode("utf-8")
        try:
            sock.sendto(request, ("127.0.0.1", DISCOVERY))
            data, addr = sock.recvfrom(2048)
            print(f"    ✓ 收到应答: {data[:160]!r}")
        except ConnectionResetError as exc:
            print(f"    ✗ WSAECONNRESET —— 端口不可达，没人监听（{exc}）")
        except socket.timeout:
            print("    ✗ 超时 —— 有人监听但没回")
        finally:
            sock.close()

        print()
        print("  agent 日志尾部：")
        log = ROOT / "_build" / "agent_diag.log"
        if log.is_file():
            text = log.read_text(encoding="utf-8", errors="replace")
            for line in text.splitlines()[-14:]:
                print(f"    {line}")

    finally:
        proc.terminate()
        try:
            proc.wait(timeout=10)
        except subprocess.TimeoutExpired:
            proc.kill()

    return 0


if __name__ == "__main__":
    sys.exit(main())