You should be able to pipe your output just into that
sed command.
infile is just the filename I saved your example on my harddisk and I feed
sed not with a pipe in front of it but just with this infile.
If the output you want to parse varies a lot, you can't go like that. You have to know how your output looks as close as possible.
Code:
sed -n # Print only the stuff we want to see; else it prints all
# the stuff it is processing additionally
' # Here starts the regex and commands
/^$/,/^$/ # The / starts and ends a pattern; ^ is the start of a
# line, $ the end of a line and since nothing is specified
# between it, it must be an empty line. So writing two
# patterns separated with a comma is like giving a
# range "from" "to"
{/^[^ ]/p} # When the range is separated, do the command in curled
# braces. The command inside the braces says to get a
# pattern, that does not start with a blank. Inside square brackets the ^ means "not". The p simply stands for print.
' infile
