Rails serialization of sibling objects into groups by name (active model serializers) - ruby-on-rails

Question

I'm not sure the best way of approaching this; I've got a generic system for requesting data objects from an api and returning json by using ActiveModelSerializers and it's been fantasticly simple.
For one class I need to return data formatted differently than the normal class serilization and unsure of the best approach. No matter what it seems that I'm going to have to have an 'elsif' for this one class in the return (which pains me a little bit).
Currently I've got a class named 'curve' which a user has_many of, and so requesting a user's curve gets me something like this:
[{
"id": 7,
"name": "A",
"angle": 30,
"date": "2017-05-23T01:52:00.589-04:00",
"direction": "left",
"top": "C1",
"bottom": "C4",
"risser": 3,
"sanders": 8
}, {
"id": 8,
"name": "B",
"angle": 0,
"date": "2017-05-23T01:52:56.107-04:00",
"direction": "right",
"top": "C5",
"bottom": "C6",
"risser": null,
"sanders": null
}, {
"id": 9,
"name": "A",
"angle": 22,
"date": "2017-05-25T01:56:00.656-04:00",
"direction": "right",
"top": "C3",
"bottom": "C5",
"risser": null,
"sanders": null
}, {
"id": 11,
"name": "C",
"angle": 3,
"date": "2017-05-26T01:57:08.078-04:00",
"direction": "right",
"top": "C4",
"bottom": "C7",
"risser": null,
"sanders": null
}]
But I actually need each to be grouped by the name like so:
[{
"name": "A",
"points": [{
"id": 7,
"name": "A",
"angle": 30,
"date": "2017-05-23T01:52:00.589-04:00",
"direction": "left",
"top": "C1",
"bottom": "C4",
"risser": 3,
"sanders": 8
}, {
"id": 9,
"name": "A",
"angle": 22,
"date": "2017-05-25T01:56:00.656-04:00",
"direction": "right",
"top": "C3",
"bottom": "C5",
"risser": null,
"sanders": null
}]
},
{
"name": "B",
"points": [{
"id": 8,
"name": "B",
"angle": 0,
"date": "2017-05-23T01:52:56.107-04:00",
"direction": "right",
"top": "C5",
"bottom": "C6",
"risser": null,
"sanders": null
}]
},
{
"name": "C",
"points": [{
"id": 11,
"name": "C",
"angle": 3,
"date": "2017-05-26T01:57:08.078-04:00",
"direction": "right",
"top": "C4",
"bottom": "C7",
"risser": null,
"sanders": null
}]
}
]
Now I know I can do group_by and fiddle with the response - but even then it won't be using the default Curve serializer. I could also create a custom GroupedCurve serializer and possibly process each curve with a CurveSerializer - but at that point isn't it just like getting the default array and doing a map, and constructing my own hashes?
What's the best / cleanest way of processing data from top to the bottom format?
UPDATE:
Something like the following 'does the job' but the job is dirty:
def self.reformat_curves(curves)
arr = []
curves.each do |i|
alpha = i[:name]
entry = arr.find{|chunk| chunk[:name]==alpha}
if entry.nil?
arr << {name: alpha, points: [i]}
else
entry[:points] << i
end
end
return arr
end