Sed: execute alternative command when output from previous one is empty - bash

Question

I wanted to modify dig command to automatically apply a reverse lookup to any A record I receive in output.
Therefore, I've created the following function:
dt ()
{
remove=$(echo $# | sed 's^https*^^' | sed 's^[,/*:]^^g' );
dig any +trace +nocl +nodnssec $remove | sed "s/\($remove.*A\s*\([0-9]\{1,3\}\.[0-9]\{1,3\}\.[0-9]\{1,3\}\.[0-9]\{1,3\}\)\)/dig -x \2 +short | xargs echo \"\1 ->\"/e"
}
With this code I have the following output (only A record part is shown so to avoid the question getting bigger):
domain.com. 1200 A 198.54.115.174 -> server224-3.web-hosting.com.
However, now I also need to make a whois lookup using the IP I receive from dig output, but only in case dig -x \2 +short doesn't give any result (stackoverflow.com can be a good example of a domain with A records that do not have PTR).
I tried something like this to check exit code of regular host command (since dig implies that output is successful even if it's empty) and execute proper command depending on the result:
dig any +trace +nocl +nodnssec "$remove" | sed -e "s/\($remove.*A\s*\([0-9]\{1,3\}\.[0-9]\{1,3\}\.[0-9]\{1,3\}\.[0-9]\{1,3\}\)\)/echo -n \"\1 -> \"; host \2 | grep -q ''; if [ ${PIPESTATUS[0]} -eq 0 ]; then dig -x \2 +short; else whois \2 | grep 'network:Network-Name:';fi; /e"
But it seems that sed somehow affects the value of ${PIPESTATUS[0]} array.
I wanted to do these modifications in sed because I needed something that will print lines on the go. If I use variables and modify the output from them, the function will work slower, at least visually.
Maybe awk can be of use here, but I am not sure how I should write the code using this command.
Is there a way around this problem? Can sed be used for this purpose or should I use some other tool? Thanks.

Answer 1

The good old' bash gives you variety of tools grep, awk, xargs, while read and so on. Using sed with e command with inside checking PIPESTATUS and executing xargs.... is just unreadable and very long too read. It's not clear to me what do you want to do, as too much happens on one line.
The dig command results zero status on success - you can check it's output if it has zero length when it "fails".
Consider the following script:
dt () {
remove=$(
<<<$1 \
sed 's#^http[s]?://\([^/]*\).*#\1#'
)
dig any +trace +nocl +nodnssec "$remove" |
tr -s '\t' |
awk -v "remove=${remove}." '{
if ($1 == remove && $3 == "A") {
print
}
}' |
while IFS= read -r line; do
IFS=$'\t' read -r _ _ _ ip _ <<<"$line"
if ! dig_result=$(dig -x "$ip" +short); then
echo "Dig error" >&2
exit 1
fi
if [ -z "$dig_result" ]; then
printf "%s -> no-result-from-dig\n" "$line"
else
printf "%s -> %s\n" "$line" "$dig_result"
fi
done
}
First dig is run. Notice the "$remove" is quoted.
Just a precaution - I squeze the tabulators in the output.
Then I filter - the first column should have "$remove". with a dot and the third column should be an "A"
Then for each line (as there are meny)
I get the ip address (ok, maybe ip=$(<<<"$line" cut -f4) would be cleaner and simpler).
I get the result from the dig -x using the ip address
If the result is empty I print "no-result-from-dig" if it not, I print the result.
Running:
dt stackoverflow.com
dt google.pl
Outputs:
stackoverflow.com. 300 A 151.101.65.69 -> no-result-from-dig
stackoverflow.com. 300 A 151.101.193.69 -> no-result-from-dig
stackoverflow.com. 300 A 151.101.129.69 -> no-result-from-dig
stackoverflow.com. 300 A 151.101.1.69 -> no-result-from-dig
google.pl. 300 A 216.58.215.67 -> waw02s16-in-f3.1e100.net.
The reason your command did not work is that $PIPESTATUS was quoted inside " quotes thus expanded before running the shell. You should escape the $ so that's \$PIPESTATUS or pass it inside nonexpading single quotes " ... "'$PIPESTATUS'" ... ".