Help combining two maps based on map name

[glitchcrab]

I have the following data structure, whereby each entry under .nodePools represents a basic VM configuration, and each VM's configuration is extended by referencing a named class under .nodeClasses:

nodePools:
  - class: default
    replicas: 2
    name: worker
  - class: default
    replicas: 2
    name: anotherworker
nodeClasses:
  default:
    memoryMiB: 16896
    diskGiB: 50

I have tried and failed at the following; I would like to write a command to iterate over each nodePool, look up its class and then duplicate the keys from the class into the nodePool. The aim is to end up with the following:

nodePools:
  - class: default
    replicas: 2
    name: worker
    memoryMiB: 16896
    diskGiB: 50
  - class: default
    replicas: 2
    name: anotherworker
    memoryMiB: 16896
    diskGiB: 50
nodeClasses:
  default:
    memoryMiB: 16896
    diskGiB: 50

Could anyone help me out with this please?

[mikefarah]

Short Answer:

yq '.nodeClasses as $nodeClasses | .nodePools[] *=n $nodeClasses[.class]' file.yaml

• First we make a variable called $nodeClasses. We'll need this because we are going to update each of the nodePools relative to themselves (and so the . operator will refer to the specific node pool).
• then we want to update all of the nodePools by merging in their current value with the nodeClass[.class].
  • * is the merge operator, we could do `.nodePools[] |= . * ($nodeClasses | .[.class])' but we can instead use the update shorthand:
  • *=n doc is a relative merge operator we can use to do exactly that.
  • the extra n tells it only to merge in "new" values (so if someone specifies a "diskGiB" for a particular nodePool, it won't get clobbered by the default. You can drop the n if that's not the desired thing

[glitchcrab]

Amazing, thanks Mike 🙏 this makes me realise that I was massively overcomplicating the task - your answer is much more concise than I was coming up with.