Parse yml file with yq when multiple keys are not nested (i.e abc.def)

Question

I have a yml file that looks like:

abc.def:
  key: value

and I am trying to get the value of key by doing the following:

yq '.abc.def.key' file.yml

but it is returning null. If I format the yml file as:

abc:
  def:
    key: value

the yq command returns value. I need to be able to parse both versions and I cant just manually update the yml files. Is there something I am missing? Is there a format command I can preprocess the file with so yq will work? Thanks

Answer 1

Your two YAML documents are NOT two "versions" or representations of the same data, they are actually two very different files. One has a key that happens to include a dot, the other one has one level more of nesting. Again, this is not another version of the same document.

To address a key with special characters (like a dot), use the universal syntax for addressing keys: .["abc.def"] or ."abc.def". Dropping (brackets and) quotes is just a shortcut for keys that do not contain special characters (as with .key): .["abc.def"].key or ."abc.def".key

I need to be able to parse both versions

You can use the alternative operator // to address another path if the first one evaluates to null or false:

yq '(.["abc.def"] // .abc.def).key'

This works with both kislyuk/yq and mikefarah/yq. (Tested with kislyuk/yq 3.2.3 and mikefarah/yq v4.35.1).

---

Alternatively, if you want to regard both documents as equivalent, you could just transform one into the other, i.e. expand paths with dots in a key name to the nested variant (note that collisions might occur), and then just query the unified version using the simple syntax, e.g. .abc.def.key.

Here's how you'd convert with kislyuk/yq using tostream to recursively get paths and values, and reduce to iteratively re-build the output object:

reduce (tostream | select(has(1)) | first |= map(split(".")[]))
  as [$p,$v] (null; setpath($p; $v))

And here's the same with mikefarah/yq using .. to traverse, path to get the path, and ireduce for the build-up:

[ ..
  | select(tag != "!!map" and tag != "!!seq")
  | [[path | .[] | split(".") | .[]], .]
] | .[] as $p ireduce (null; setpath($p[0]; $p[1]))