We're building a website using the Pentaho CTools library, which has a graphical dashboard editor which writes out JSON-format files for part of the dashboard.
I'd like to apply a transform to these files before check-in to git in order to sort them by key and then by the value of certain keys. The purpose is to make diffs easier, since the editor has a habit of rearranging all of the json fields.
For example, we might have something like this:
{
"components": {
"rows": [
{
"id": "CHARTS",
"name": "Charts",
"parent": "UnIqEiD",
"properties": [
{
"name": "Group",
"type": "Label",
"value": "Charts"
}
],
"type": "Label",
"typeDesc": "<i>Group</i>"
},
{
"id": "kjalajsdjf",
"meta_cdwSupport": "true",
"parent": "CHARTS",
"properties": [
{
"name": "name",
"type": "Id",
"value": "Value1"
},
{
"name": "title",
"type": "String",
"value": "Value2"
},
{
"name": "listeners",
"type": "Listeners",
"value": "[]"
},
...
We are able to jq --sort-keys
(http://stedolan.github.io/jq/) to sort all of the keys, but I'm struggling to find out how to use the sort_by
function to then sort certain specific elements by the value of certain keys (so, in the example above, sorting by properties.name
for example. Any ideas?
Ok with some assistance on the IRC channel I've found an answer.
Basically, it looks like this:
jq '.components.rows|=sort_by(.id)|.components.rows[].properties|=sort_by(.name)' file.json > out.json
So you do the select of the right object, walking into arrays if needed, and then sort_by just takes a single value. I was trying sort_by(.components.rows.id)
which failed.
The |=
instead of the |
passes the values along instead of stripping them.