Reputation: 125
Essentially I'm trying to take data and use Enum.group_by to create a dictionary but I'd like to continue to group the same data by sub categories.
data = [
%{company: "company_one", state: "LA", size: 100},
%{company: "company_one", state: "LA", size: 200},
%{company: "company_two", state: "TX", size: 200},
%{company: "company_two", state: "LA", size: 300},
%{company: "company_three", state: "LA", size: 400},
%{company: "company_four", state: "TX", size: 500}
]
I want to first group by company then within that map group those now by state and once again, within that nested map group them by size. Essentially a map with two nested maps and an array of the corresponding data.
%{"company_one" => %{
"LA" => %{
"100" => [
%{company: company_one, state: LA, size: 100}
],
"200" => [
%{company: company_one, state: LA, size: 200}
]
}
}
}
My attempt looks something like this
list =
data
|> Enum.group_by(fn x -> x.company end)
keys = Map.keys(data)
updated_list =
for key <- keys do
list[key]
|> Enum.group_by(fn x -> x.state end)
end
I've used Enum.group_by/1 for the initial format but anything i've tried after that really messes up the data structure. Any help would be great. Thanks.
Upvotes: 1
Views: 499
Reputation: 3215
I found this question interesting, and ended making a generic way to do this - just pass your structure and the keys you want to group by.
defmodule NestedGroup do
defp access_keys([key]), do: [Access.key(key, [])]
defp access_keys([key | rest]), do: [Access.key(key, %{}) | access_keys(rest)]
def nested_group(data, keys) do
data
|> Enum.reduce(%{}, fn elt, acc ->
keys =
Enum.map(keys, &Map.get(elt, &1))
|> access_keys()
update_in(acc, keys, fn list -> [elt | list] end)
end)
end
end
We need the access_key
helper to make sure the last key is a list, instead of a map.
iex(2)> NestedGroup.nested_group(data, [:company, :state, :size])
%{
"company_four" => %{
"TX" => %{500 => [%{company: "company_four", size: 500, state: "TX"}]}
},
"company_one" => %{
"LA" => %{
100 => [%{company: "company_one", size: 100, state: "LA"}],
200 => [%{company: "company_one", size: 200, state: "LA"}]
}
},
"company_three" => %{
"LA" => %{400 => [%{company: "company_three", size: 400, state: "LA"}]}
},
"company_two" => %{
"LA" => %{300 => [%{company: "company_two", size: 300, state: "LA"}]},
"TX" => %{200 => [%{company: "company_two", size: 200, state: "TX"}]}
}
}
Upvotes: 2
Reputation: 23091
It's not very pretty, but you could do it like this:
Enum.group_by(data, & &1.company)
|> Map.new(fn {k, v} ->
{k,
Enum.group_by(v, & &1.state)
|> Map.new(fn {k, v} -> {k, Enum.group_by(v, & &1.size)} end)}
end)
Result:
%{
"company_four" => %{
"TX" => %{500 => [%{company: "company_four", size: 500, state: "TX"}]}
},
"company_one" => %{
"LA" => %{
100 => [%{company: "company_one", size: 100, state: "LA"}],
200 => [%{company: "company_one", size: 200, state: "LA"}]
}
},
"company_three" => %{
"LA" => %{400 => [%{company: "company_three", size: 400, state: "LA"}]}
},
"company_two" => %{
"LA" => %{300 => [%{company: "company_two", size: 300, state: "LA"}]},
"TX" => %{200 => [%{company: "company_two", size: 200, state: "TX"}]}
}
}
Upvotes: 3