Add new line with previous line's indentation - sed

Question

I'm struggling since a moment now to update some yaml files by adding an element in a new line with sed.
The sed command (or another linux command) must match a string (image: or - image:), add a new element on a new line with the same indentation as previous line.
The thing is that the new line must be exactly just under the string image: and not under - image:.
Example: When the sed matches the string image, it adds the new line with correct indentation (just above image)
kind: DaemonSet
apiVersion: apps/v1
metadata:
name: calico-node
spec:
template:
spec:
containers:
- name: calico-node
image: myimage
imagePullSecret: mysecret
...
Example 2: when the sed matches the string - image, it adds the new line with correct indentation (just above image and not -):
kind: DaemonSet
apiVersion: apps/v1
metadata:
name: calico-node
spec:
template:
spec:
containers:
- image: myimage
imagePullSecret: mysecret
...
What is not wanted is
kind: DaemonSet
apiVersion: apps/v1
metadata:
name: calico-node
spec:
template:
spec:
containers:
- image: myimage
imagePullSecret: mysecret
I already tried the yq command to deal with this but it's a gigantic pain...
I found this code in another thread but it doesn't work when matching the - image string.
sed '/^ *image: .*/ { G; s/^\( *\)image: .*/&\1imagePullSecret: mysecret/; }' sed.yaml

Answer 1

I don't condone the use of sed to parse yaml, but if you really want to do that, you could try something like:
$ nl='
'
$ printf ' image:\n - image:\n' |
sed -e "s/\(^\( *\)image:\)/\1\\$nl\2inserted text/" \
-e "s/\(^\( *\)- image:\)/\1\\$nl\2 inserted text/"
image:
inserted text
- image:
inserted text

Answer 2

I'd suggest:
sed '
/^( *)image: .*/ {p; s//\1imagePullSecret: mysecret/; }
/^( *)- image: .*/ {p; s//\1 imagePullSecret: mysecret/; }
' sed.yaml
The empty pattern in s//replacement/ reuses the most-recently used pattern.

Answer 3

You could use this awk:
awk '/^[[:space:]-]*image/{ split($0,arr,/image.*/)
gsub(/-/," ", arr[1])
rep=arr[1]
print $0}
rep{ printf "%s%s\n", rep, "imagePullSecret: mysecret"
rep=""
next} 1' file

Answer 4

Using GNU awk to utilise gensub:
awk '/image: /{ str=$0 } str { print;str1=gensub(/(^.*)(image:.*$)/,"\\1imagePullSecret: mysecret",str);sub("-"," ",str1);print str1;del str;next }1' file
Explantion:
awk '/image: /{ # Process the image tag
str=$0 # Set a variable str equal to the line
}
str { # Process when we have a str set
print; # Print the current line
str1=gensub(/(^.*)(image:.*$)/,"\\1imagePullSecret: mysecret",str); # Split the line into two sections and set str1 to the first section (get the leading spaces) and then the text we want to add.
sub("-"," ",str1); # Replace any - in the text with a space
print str1; # Print the new line
del str; # Remove the str variable
next # Skip to the next line
}1' file

Answer 5

Yq v4 doesn't yet support maps for addition, but you could convert to JSON, use jq, and then convert back to YAML:
yq --tojson eval infile.yaml \
| jq '.spec.template.spec.containers[0] += {imagePullSecret: "mysecret"}' \
| yq eval -prettyPrint

Answer 6

This might work for you (GNU sed):
sed -E '/(-( ))?image:.*/{p;s//\2\2imagePullSecret: mySecret/}' file
Match on image:.* with an optional - prepended.
Print the current line.
Replace the match by imagePullSecret: mySecret with the optional space following the - prepended twice.