CODEWITHSUNDEEP
ruby
vladiim
vladiim

Reputation: 1960

Group By an Array of Hashes

For the array of hashes:

array = [{id: 1, name: 'name', count: 2},
         {id: 1, name: 'name', count: 1},
         {id: 1, name: 'new name', count: 1}]

I'd like the result:

[{id: 1, name: 'name', count: 3},
 {id: 1, name: 'new name', count: 1}]

I've achieved this with:

grouped_data = array.each_with_object(Hash.new(0)) do |row, sum|
  sum["#{row.fetch(:id)} #{row.fetch(:name)}"] += row.fetch(:count).to_i
end

result = array.uniq.each do |row|
  row[:count] = grouped_data["#{row.fetch(:id)} #{row.fetch(:name)}"]
  row
end

Is there a more elegant way to achieve the desired result?

Upvotes: 1

Views: 74

Answers (4)

iGian
iGian

Reputation: 11183

Another option, where you initialise the hash in map_with_object:

array
  .group_by { |h| h[:name] }.values
  .map{ |e| e.map.with_object({name: e[0][:name], id: e[0][:id], count: 0}){ |h, nh| nh[:count] += h[:count] } }
#=> [{:name=>"name", :id=>1, :count=>3}, {:name=>"new name", :id=>1, :count=>1}]

Upvotes: 0

Cary Swoveland
Cary Swoveland

Reputation: 110675

This is one way it would commonly be done.

 array.each_with_object({}) {|g,h| h.update(g[:name]=>g) {|_,o,n|
   o.merge(count: o[:count] + n[:count]) } }.values
  #=> [{:id=>1, :name=>"name", :count=>1},
  #    {:id=>1, :name=>"new name", :count=>1}]

This uses the form of Hash#update (aka merge!) that employs a block to determine the values of keys that are present in both hashes being merged. See the doc for definitions of the the three block variables, _, o and n. The first of these holds the the common key. It's common practice to use _ for block variables that are not used in the block calculation.

Note that the receiver of values is as follows.

array.each_with_object({}) {|g,h| h.update(g[:name]=>g) {|_,o,n|
  o.merge(count: o[:count] + n[:count]) } }
  #=> {"name"=>{:id=>1, :name=>"name", :count=>2},
  #    "new name"=>{:id=>1, :name=>"new name", :count=>1}}

Upvotes: 3

moveson
moveson

Reputation: 5213

Here's a slightly different take, which as a bonus produces the expected result if the count attribute for any element is greater than 1:

> array.group_by { |e| [e[:id], e[:name]] }.values
    .map { |group| group.first.merge(count: group.sum { |e| e[:count] }) }
#> [{:id=>1, :name=>"name", :count=>2}, {:id=>1, :name=>"new name", :count=>1}]

> array = [{:id=>1, :name=>"name", :count=>2}, 
           {:id=>1, :name=>"name", :count=>3}, 
           {:id=>1, :name=>"new name", :count=>1}]
> array.group_by { |e| [e[:id], e[:name]] }.values
    .map { |group| group.first.merge(count: group.sum { |e| e[:count] }) }
#> [{:id=>1, :name=>"name", :count=>5}, {:id=>1, :name=>"new name", :count=>1}]

Upvotes: 2

yzalavin
yzalavin

Reputation: 1836

You can group and map:

array
  .group_by { |el| [el[:id], el[:name]] }
  .map { |k,v| v.first.merge(count: v.length) }

Upvotes: 2

Related Questions