In my application I am executing curl.exe as external command to get list of files from an FTP folder. I need to scan multiple folders to the list so I am calling this command multiple times output the file listing into local directory.
curl ftp://my-ip:port/path/1234/ --user username:password --list-only -o C:\localfolder\1234.txt
curl ftp://my-ip:port/path/8910/ --user username:password --list-only -o C:\localfolder\8910.txt
So I will have 1 file per each directory. We can't use wild card pattern for listing files from all required directories as they have no pattern, just random integer numbers.
Most of the cases I will need to scan 200 directors but also in some cases a couple of thousands upto 9000 (which is maximum). I have no problem for it as it is working perfect without any issues. Except in some extreme cases I won't get list from 1 or 2 directories due to network glitch.
Now I am thinking of optimization. Two optimizations can be done here
My application level optimization: Instead of calling curl.exe so many times, I will create a dynamic bat file with multiple calls in it so my application will call it only once and batch file will have the actual curl calls.
Curl connection pooling: This is what I am looking if it is possible with curl.exe I read somewhere sometime ago that curl library has connection pooling support like in PHP but I am not sure if curl executable has that capability. It could be good performance improvement if I can utilize this feature if it has one.
答案 0 :(得分：0)
根据cURL手册（在 PERSISTENT CONNECTIONS 一节中）：
在单个命令行上指定多个文件将进行卷曲传输 所有这些，按照指定的顺序一个接一个地。
libcurl将尝试使用持久连接进行传输 第二次传输到同一主机可以使用相同的连接 已经启动并在之前的转移中保持开放状态。这很大 除了第一次转移之外，所有人都减少了连接时间 更好地利用网络。
来源：https://curl.haxx.se/docs/manual.html
像
这样的东西curl --user username:password --list-only ftp://my-ip:port/path/1234/ -o C:\localfolder\1234.txt ftp://my-ip:port/path/8910/ -o C:\localfolder\8910.txt
应该做你想要的（2个或更多的URL）。