Bash for loop prints unexpected output, why is that? - regex

Question

I wrote a script which queries a dns zone on AWS Route53 and returns IP's of specific servers.
The aim is to run this script after starting a CloudFormation stack and that the script will update my local ~/.ssh/config file with the updated servers IP's.
I want to "catch" the line in the config which starts with HostName but related to the hostname of the server I would like to edit and so I've ran the following command:
$ grep -A2 dev-api2.company-private ~/.ssh/config
Host dev-api2.company-private
User ec2-user
HostName 5.5.5.5
Next, I've ran:
$ for line in $(grep -A2 dev-api2.company-private ~/.ssh/config); do echo $line ; done
Host
dev-api2.company-private
User
ec2-user
HostName
5.5.5.5
For some reason the output is printed not as I expected, it broke each word to a separate line, but why does it happen?
Edit #1:
My idea was to do something like that:
for line in $(grep -A2 dev-api2.company-private ~/.ssh/config)
do
if [[ $line == "^HostName.*" ]]
then
sed -i 's!$line!HostName 1.1.1.1!g' ~/.ssh/config
fi
done
Edit #2:
Ok, so the relevant portion of the ~/.ssh/config file looks like that:
Host as-bastion
User ec2-user
HostName ec2-34.us-west-2.compute.amazonaws.com
IdentityFile ~/.ssh/company/bastion.pem
Host dev-api1.company-private
User ec2-user
HostName 172.51.24.171
IdentityFile ~/.ssh/company/company.pem
ProxyCommand ssh ec2-user#as-bastion -W %h:%p
Host dev-api2.company-private
User ec2-user
HostName 172.51.25.248
IdentityFile ~/.ssh/company/company.pem
ProxyCommand ssh ec2-user#as-bastion -W %h:%p
The file contains many blocks like the ones above, but I want to sed only the "HostName" lines of blocks which contain the string I've used in the grep (dev-api2.company-private).
#anubhava: The code in your answer changed lines containing the string "HostName" even in blocks which are not related to "dev-api2.company-private".
How can it be done?

Answer 1

You should be using this while read loop using process substitution instead of a for loop using command substitution:
while IFS= read -r line; do
echo "$line"
done < <(grep -A2 dev-api2.company-private ~/.ssh/config)
Based on your comments I suggest it is better to use awk like this:
awk -v ORS='\n\n' -v RS= -v api='^Host dev-api2' -v ip='1.1.1.1' '
$0 ~ api "\\.company-private$" {sub(/\nHostName [^\n]+/, "\nHostName " ip)} 1' ~/.ssh/config
To save changes use this awk command:
awk -v ORS='\n\n' -v RS= -v api='^Host dev-api2' -v ip='1.1.1.1' '
$0 ~ api "\\.company-private$" {sub(/\nHostName [^\n]+/, "\nHostName " ip)} 1
' ~/.ssh/config > $$.tmp && mv $$.tmp ~/.ssh/config
Here is one liner sed to do the same job:
sed -i.bak '/dev-api2\.company-private/{N;p;N;s/.*/HostName 1.1.1.1/;}' ~/.ssh/config