ilr v0.0.3-c

ilr

ilr is like jq but worse! I made it because I'm too dumb dumb to understand jq properly. But I know JavaScript, so I thought why not make something like jq that uses JavaScript instead?

Usage

ilr takes one argument, a string. It's a function which gets called by the input in JSON form.

Here I'm using https://api.github.com/repos/stedolan/jq/commits?per_page=5 as the example.

Identity

curl https://api.github.com/repos/stedolan/jq/commits?per_page=5 | ilr 'x => x'

In addition, the entire Ramda codebase is exported to the global namespace, so you could also do

curl https://api.github.com/repos/stedolan/jq/commits?per_page=5 | ilr 'identity'

Get the first commit

curl https://api.github.com/repos/stedolan/jq/commits?per_page=5 | ilr 'head'

(For the rest of the examples, I'll leave out the curl command.)

Only get interesting fields from the first commit

ilr 'head' | ilr 'paths([["commit","message"],["commit","committer","name"]])' | ilr 'zipObj(["message", "name"])'

Geez that's pretty bad, isn't it? I think ilr is nice for simple transformations that can be expressed using only a few Ramda functions, but anything more complicated like this tends to get messy quickly.

Get parents

ilr 'map(prop("parents"))' | ilr 'map(head)'

Find the first commit from someone who has an email address ending with .jp

ilr 'find(x => endsWith("jp", x.commit.author.email))' | ilr 'prop("commit")'

You could probably do this with entirely Ramda functions but it's cleaner to just use an arrow function.

Todo

Ramda's path and paths APIs are pretty awkward. I think it might be possible to use Babel to create nicer syntax for function application and arrays. In particular:

• space for function application: map prop "parents" becomes map(prop("parents"))
• . for picking, e.g. .commit.author.email becomes x => x.commit.author.email