perf(123pan): optimize rate limiting

[seiuneko]

perf(123pan): optimize rate limiting

• eliminating fixed 200 ms delay in getFiles to prevent thread starvation
  Under a concurrency of 5 requests, performance improvement is as follows:

Command	Mean [s]	Min [s]	Max [s]	Relative
before	205.099 ± 14.682	188.500	227.832	1.13 ± 0.08
after	182.253 ± 0.188	182.112	182.488	1.00

Benchmark Commands

31812 (seiuneko@R9000P) /w
❯❯❯ cat benchmark
#!/usr/bin/zsh

export user=admin pass='D8@Sj&VDXKCM%47s'
# seq 0 255 | xargs printf "%02x\n" > dir_list
xargs -a dir_list -I _ -P 5 curl -u $user:$pass -X PROPFIND -H 'Depth: 1' -o /dev/null -s http://127.0.233.1:5244/dav/123pan/restic-repo/data/_

31804 (seiuneko@R9000P) /w
❯❯❯ hyperfine --shell zsh --runs 5 --prepare './alist-before --data /w/alist-run server 2>/dev/null &; sleep 1' ./benchmark --conclude 'pkill alist; sleep 60' --prepare './alist-after --data /w/alist-run server 2>/dev/null &; sleep 1' ./benchmark --conclude 'pkill alist; sleep 60' --export-markdown result.md --command-name before --command-name after
Benchmark 1: before
  Time (mean ± σ):     205.099 s ± 14.682 s    [User: 0.574 s, System: 0.650 s]
  Range (min … max):   188.500 s … 227.832 s    5 runs

Benchmark 2: after
  Time (mean ± σ):     182.253 s ±  0.188 s    [User: 0.568 s, System: 0.666 s]
  Range (min … max):   182.112 s … 182.488 s    5 runs

  Warning: Statistical outliers were detected. Consider re-running this benchmark on a quiet system without any interferences from other programs. It might help to use the '--warmup' or '--prepare' options.

Summary
  after ran
    1.13 ± 0.08 times faster than before

• allowing cancellation via context to mitigate potential DoS attacks by immediately cancelling excessive requests

[xrgzs]

I have encountered performance issues these days. After applying these changes, it works well.

[xrgzs]

用 #6859 导致号被封禁了，估计是修改后调用接口超速了，请回滚到之前的版本

PS：之前的代码及限速部分为123官方人员提交，根据客服回应看，应该不会出现问题

另外之前本人在使用时出现的性能问题为DNS服务器导致，现已查明

@seiuneko @xhofe

[seiuneko]

@xrgzs 经过对代码逻辑的确认，接口限流间隔依然是之前 123 云盘官方提交的 700 ms，下面是实际调试截图，可以看到是按 700 ms 间隔请求的：

之前在测试时一天内多次触发 123 云盘风控，但账号能正常登录，只是 IP 被封禁，且光猫重新拨号后能正常使用：

猜测账号封禁原因可能与你截图里显示的异地登录有关？

[xrgzs]

@xrgzs 经过对代码逻辑的确认，接口限流间隔依然是之前 123 云盘官方提交的 700 ms，下面是实际调试截图，可以看到是按 700 ms 间隔请求的：

之前在测试时一天内多次触发 123 云盘风控，但账号能正常登录，只是 IP 被封禁，且光猫重新拨号后能正常使用：

猜测账号封禁原因可能与你截图里显示的异地登录有关？

异地登录倒是没有，我这边的所有登录设备都是处于同一IP的。之前用老版本没有出现问题，更换新版以后跑两天就出事了，现在回到老版本，再测试几天。