I have an array of hashes:
[{:a => "63-aaa",:b =>1}, {:a => "90-bbb",:b => 89}, {:a => "63-aaa", :b => 1},{:a => "45-hhh", :b => 44}, {:a => "32-kkk", :b => 67}, {: => "90-bbb", :b => 89}]
a key has value 63-aaa, 90-bbb, 45-hhh and 32-kkk.
I want to get the unique elements of the array on the basis of ids of values of a like:
[{:a => "63-aaa",:b =>1}, {:a => "90-bbb",:b => 89}, {:a => "45-hhh", :b => 44}, {:a => "32-kkk", :b => 67}]
You can pass a block to Array#uniq in this case:
arr.uniq { |hash| hash[:a] }
You don't appear to care which one of each set of duplicates is kept. Array#uniq, with a block, is purpose-built for that. It will keep the one with the smallest index. If you want the one with the largest index, just apply Array#reverse before and after uniq.
If, however, you wanted to use other information in each hash to determine which one to keep, you could do that by using the form of Hash#update (a.k.a. merge!) that takes a block:
arr = [{:a=>"63-aaa", :b=> 1}, {:a=>"90-bbb", :b=>89}, {:a=>"63-aaa", :b=>22},
{:a=>"45-hhh", :b=>44}, {:a=>"32-kkk", :b=>67}, {:a=>"90-bbb", :b=>14}]
arr.each_with_object({}) do |g,h|
h.update({ g[:a]=>g }) { |k,oh,nh| <code to return oh or nh> }
end.values
Suppose, for example, you want to keep the hash h for which h[:b] is largest:
arr.each_with_object({}) do |g,h|
h.update({ g[:a]=>g }) { |k,oh,nh| nh[:b] > oh[:b] ? nh : oh }
end.values
#=> [{:a=>"63-aaa", :b=>22}, {:a=>"90-bbb", :b=>89},
# {:a=>"45-hhh", :b=>44}, {:a=>"32-kkk", :b=>67}]
I have created an empty hash (block variable h), and then for each hash g in arr, update h with the hash f = { g[:a]=>g }. If both h and f have the key g[:a], the block
{ |k,oh,nh| (nh[:b] > oh[:b]) ? nh : oh }
is called upon to determine the value of the key g[:a] in h (that is, which of the two hashes to keep). The block variables are:
k = g[:a]
oh = h[g[:a]]
nh = g
(Note that k is not used in the block, so we might write the block variables as |_,oh,nh|.)
each_with_object returns
h = {"63-aaa"=>{:a=>"63-aaa", :b=>22}, "90-bbb"=>{:a=>"90-bbb", :b=>89},
"45-hhh"=>{:a=>"45-hhh", :b=>44}, "32-kkk"=>{:a=>"32-kkk", :b=>67}}
so we merely have to extract the values:
h.values
#=> [{:a=>"63-aaa", :b=>22}, {:a=>"90-bbb", :b=>89},
# {:a=>"45-hhh", :b=>44}, {:a=>"32-kkk", :b=>67}]
Another way is to first execute:
arr.group_by { |h| h[:a] }.values
#=> [[{:a=>"63-aaa", :b=> 1}, {:a=>"63-aaa", :b=>22}],
# [{:a=>"90-bbb", :b=>89}, {:a=>"90-bbb", :b=>14}],
# [{:a=>"45-hhh", :b=>44}],
# [{:a=>"32-kkk", :b=>67}]]
and then use whatever criteria you like to select one hash from each of the four arrays.
