Extracting git url's domain with POSIX

I'm trying to build the most robust way to extract the domain from a git repo.
For urls like:

ssh://git@gitlab.com:22411/usage/project_100.git
git://example.org/path/to/repo.git
https://github.com/example/foobar.git
http://github.com/example/foobar.git
ssh://user@host.com/path/to/repo.git
git://host.com/path/to/repo.git

I can use:

echo $url | awk -F[/:] '{print $4}'

But for repos like:

"git@github.com:User/UserRepo.git"

It won't work. But the following does:

echo $url | awk -v FS="(@|:)" '{print $2}'

Is there some robust way I could always exctract the domain in POSIX?

If the URL contains ://, you know to drop the protocol, then drop everything from the first / onwards. Otherwise, if it contains @, assume it is your second case, and drop everything up to and including the @, then everything from the : onwards. Other cases can be added as necessary.

url="..."
case $url in
  *://*)
    domain=${url#*://}
    domain=${domain#*@}
    domain=${domain%%/*}
    ;;
  *@*:*)
    domain=${url#*@}
    domain=${domain%%:*}
    ;;
 esac

With sed. I switched from s/// to s|||.

sed 's|.*//||; s|.*@||; s|/.*||; s|:.*||' file

output:
<pre>
gitlab.com
example.org
github.com
github.com
host.com
host.com
</pre>

Perl version :

perl -pe 's{.*//([^/]+@)?([^:/]+).*}{$2}' input-file

You can do that with sed easily.

echo $url | sed -E 's/.*\:\/\/(.*)@?.*\:.*/\1 /' | awk -F@ '{print $1}'