Best way to split arrays into multiple small arrays in ruby

Question

What is the simplest way to split arrays into multiple arrays based on some conditions? In my scenario, I need to move the integer and the string values to different arrays.

Answer 1

Here is my solution:

hash = x.group_by { |t| t.kind_of? Fixnum }
# hash  => {true=>[1, 2, 3, 4], false=>["a", "b"]} 
array1 = hash[true] # The array of integers
array2 = hash[false] # The array of strings

Answer 2

You're looking for Enumerable#partition:

x = [1, 2, 3, "a", "b", 4]
numbers, not_numbers = x.partition{|item| item.kind_of?(Fixnum)}
# => [[1, 2, 3, 4], ["a", "b"]]

Answer 3

Try:

x.group_by {|x| x.class}

You can get an array back by then calling to_a on the result which in the example you've given will return:

[[Fixnum, [1, 2, 3, 4]], [String, ["a", "b"]]]

Answer 4

Just to throw some more solutions into the pool:

x = [1,2,3,"a","b",4]

numbers = x.select{ |e| e.is_a?(Fixnum) } # => [1, 2, 3, 4]
letters = x - numbers # => ["a", "b"]

numbers = x.select{ |e| e.kind_of?(Fixnum) } # => [1, 2, 3, 4]
letters = x - numbers # => ["a", "b"]

or

(numbers, letters) = x.group_by {|a| a.class}.values_at(Fixnum, String)
numbers # => [1, 2, 3, 4]
letters # => ["a", "b"]

Along with some benchmarks showing how a subtle change effects speed:

require 'benchmark'

x = [1,2,3,"a","b",4] * 100
n = 10_000
Benchmark.bm do |bench|
  bench.report { n.times {
    numbers = x.select{ |e| e.is_a?(Fixnum) }
    letters = x - numbers
  }}

  bench.report { n.times {
    numbers = x.select{ |e| e.kind_of?(Fixnum) }
    letters = x - numbers
  }}

  bench.report { n.times {
    (numbers, letters) = x.group_by {|a| a.class}.values_at(Fixnum, String)
  }}

  bench.report { n.times {
    numbers, not_numbers = x.partition{|item| item.kind_of? Fixnum}
  }}
end
# >>       user     system      total        real
# >>   4.270000   0.010000   4.280000 (  4.282922)
# >>   4.290000   0.000000   4.290000 (  4.288720)
# >>   5.160000   0.010000   5.170000 (  5.163695)
# >>   3.720000   0.000000   3.720000 (  3.721459)