How can I get my bash script to remove the first n and last n lines from a variable?

Question

I'm making a script to preform "dig ns google.com" and cut off the all of the result except for the answers section.

So far I have:

#!/bin/bash

echo -n "Please enter the domain: "
read d
echo "You entered: $d"
dr="$(dig ns $d)"
sr="$(sed -i 1,10d $dr)"
tr="$(head -n -6 $sr)"
echo "$tr"

Theoretically, this should work. The sed and head commands work individually outside of the script to cut off the first 10 and last 6 respectively, but when I put them inside my script sed comes back with an error and it looks like it's trying to read the variable as part of the command rather than the input. The error is:

sed: invalid option -- '>'

So far I haven't been able to find a way for it to read the variable as input. I've tried surrounding it in "" and '' but that doesn't work. I'm new to this whole bash scripting thing obviously, any help would be great!

Answer 1

sed takes its input from stdin, not from the command line, so your script won't work either theoretically or practically. sed -i 1,10d $dr does not do what you think it does...sed will treat the value of "$dr" as a list of filenames to process.

Try echo "$dr" | sed -e '1,10d' or sed -e '1,10d' <<<"$dr".

BTW, you must use double-quotes around "$dr" here, otherwise sed will not get multiple lines of input separated by \n, it will only get one input line.

Or better yet, to get only the NS records, replace all of your sed commands with just this one command:

tr=$(echo "$dr" | sed -n -e '/IN[[:space:]]\+NS[[:space:]]/p')

Alternatively, eliminate all the intermediate steps and just run this:

tr=$(dig ns "$d" | sed -n -e '/IN[[:space:]]\+NS[[:space:]]/p')

Or you can get just the nameservers' hostnames with:

tr=$(dig ns "$d" | awk '/IN[[:space:]]+NS[[:space:]]/ {print $5}')

BTW, the output of host -t ns domain.example.com may be easier to parse than the output of dig.