How to replace a text in yml file with sed making sure that the indentation is preserved

Question

I have a YAML file with the following content:

apiVersion: v1
kind: ConfigMap
metadata:
  name: castlereport-cfg
  namespace: qa
data:
  application.properties: |
    BODY

I need to replace the BODY with the application.properties file content. I've used sed to do that like this:

sed -i -e '/BODY/r target/classes/application-dev.properties' -e 's///' -e '/^ *\$/d'  target/classes/$kubeConfigFile

It does what I need but loses the indentation and I what I get is this:

apiVersion: v1
kind: ConfigMap
metadata:
  name: castlereport-cfg
  namespace: qa
data:
  application.properties: |

server.port=8080
server.context-path=/

spring.main.banner-mode=off
logbook.format.style=http

What should I do to keep the indentation?

And one more thing that -e '/^ *\$/d' part in the sed command is there to remove blank lines, but it doesn't seem to work either.

The Expected output is the following:

apiVersion: v1
kind: ConfigMap
metadata:
  name: castlereport-cfg
  namespace: qa
data:
  application.properties: |
    server.port=8080
    server.context-path=/
    spring.main.banner-mode=off
    logbook.format.style=http   

UPDATE: I have a Jenkins pipeline script where I am able to achieve what I want with the following three lines:

   sh ( script : "sed -i -e 's/^[ \t\$]*/   /' -e \"/^ *\$/d\"  ./target/classes/"+configFile, returnStdout: true).trim() //remove trailing spaces and blank lines
   sh ( script : "sed -i -e \"/BODY/r ./target/classes/"+configFile+"\" -e \"s///\" ./target/classes/" + configMapKubernetes, returnStdout: true).trim() // insert content of application-dev.properties to YAML file using BODY
   sh ( script : "sed -i -e \"/^ *\$/d\"  ./target/classes/" + configMapKubernetes, returnStdout: true).trim() //remove  blank lines

The pipeline script is in groovy so the whole command is inside double-quotes.

Answer 1

sed is for doing simple s/old/new on individual strings that is all. This awk script is probably what you want:

$cat tst.awk
NR==FNR {
    rec[++numLines] = $0
    next
}
s = index($0,"BODY") {
    indent = sprintf("%*s",s-1,"")
    for (lineNr=1; lineNr<=numLines; lineNr++) {
        print indent rec[lineNr]
    }
    next
}
{ print }

For example, given these input files:

$ cat foo.yaml
apiVersion: v1
kind: ConfigMap
metadata:
  name: castlereport-cfg
  namespace: qa
data:
  application.properties: |
    BODY

$ cat props
here is some text
    split across
  a few lines

we can do this which uses the BODY indentation from the yaml file but also retains any additional indentation from the props file:

$ awk -f tst.awk props foo.yaml
apiVersion: v1
kind: ConfigMap
metadata:
  name: castlereport-cfg
  namespace: qa
data:
  application.properties: |
    here is some text
        split across
      a few lines

If you wanted instead to ignore the indentation from the props file and line all the props text up where BODY began that's a simple tweak:

$ cat tst.awk
NR==FNR {
    sub(/^[[:space:]]+/,"")
    rec[++numLines] = $0
    next
}
s = index($0,"BODY") {
    indent = sprintf("%*s",s-1,"")
    for (lineNr=1; lineNr<=numLines; lineNr++) {
        print indent rec[lineNr]
    }
    next
}
{ print }

$ awk -f tst.awk props foo.yaml
apiVersion: v1
kind: ConfigMap
metadata:
  name: castlereport-cfg
  namespace: qa
data:
  application.properties: |
    here is some text
    split across
    a few lines