I had this old program I called ypath where I thought I’d do XPath sorts of things with YAML. This is an idea from 10 years ago, long before jq showed up. Last week I had a need for something like that again and started to update my ypath. But, I don’t really want to build or support another tool. It’s been a long time, so I looked around:

yq (Go)

yq, a Go thing, is really a program called yaml2json. It’s binary release for darwin-386 doesn’t work on my MacBook Pro, I don’t use homebrew, and the other methods seem quite onerous. It’s not as lightweight as it claims.

shyaml (Python)

There’s shyaml, but you can only redirect input into it:

$ pip --user install shyaml
$ shyaml get-value path.to.value < file.yaml

That’s fine for a pipeline I guess, but it should require it for all cases. Otherwise it seems to work.

yq (Python)

PyPl has its own yq that’s easy to install and use. It works mostly like jq:

$ pip install yq
$ yq .path.to.value file.yml

So far this hasn’t given me problems.

My own alias

I already use jq all over the place, so if I had JSON I’d be all set. I can convert it easily with a one liner, and I even have an alias:

alias yaml2json='perl -MYAML -MMojo::JSON=encode_json -E "binmode STDOUT, q(:raw); say encode_json( YAML::LoadFile(shift) )"'

Now I use a pipeline:

$ yaml2json file.yml | jq .some.path

Giving up

The world has improved since I started App::ypath, so I’ve abandoned it. I keep a GitHub project called CPAN-Adoptable-Modules where I stash old repositories that I no longer want to manage (especially since I have tools that go through all my repos to do various things). I push it over there, make it read only, and delete everything else related to it.