On Thu, Sep 17, 2009 at 11:15:12AM +0430, Mohsen Alimomeni wrote:
> when I do "ssh ad...@host", I can get the username "admin", by the
> command "who -m", since there is tty which the username is assigned to
> it.
You also know it's "admin" because you typed "admin" in the ssh command.
Don't be too quick to discount client-side knowledge... though clearly
it's up to you to determine whether the client can be trusted.
> But suppose I want to execute a remote command "ssh ad...@host
> myprog", I want to get the username inside the myprog. The command
> "who -m", doesn't work because no tty is created for the user.
You could create a pseudoterminal by running "ssh -t ad...@host myprog"
but I suspect this is a red herring. I think what you're really asking
is "How does a program determine the name of {a,the} user that maps to
the program's {effective,real} UID, apart from running some shell command
like 'whoami'?".
> How can I get the username in this case?
If myprog has access to libc, and host is a Unix-like system, then I
believe the standard approach is:
1) Call geteuid() to get the effective UID, or getuid() to get the "real"
UID -- whichever you actually want.
2) Call getpwuid() to map the UID to a human-readable name.
This really has nothing to do with ssh per se. It's just standard
Unix/libc programming.
