In Ruby, how do I go from this:
[
1,
["green", "blue", "black"],
[ ["1", "2"], ["3"], ["4", "5"] ]
]
to this?
[
[1, "green", "1"],
[1, "green", "2"],
[1, "blue", "3"],
[1, "black", "4"],
[1, "black", "5"],
]
I tried .zip but with no luck. Any help is greatly appreciated, and of course I'm looking for a performant solution.
The pattern is not completely clear to me, but this gets the expected output for the example you provided:
data[1].
zip(data[2]).
flat_map { |x, ys| [x].product(ys) }.
map { |zs| [data[0], *zs] }
#=> [[1, "green", "1"], [1, "green", "2"], [1, "blue", "3"],
# [1, "black", "4"], [1, "black", "5"]]
We are given
arr = [1, ["green", "blue", "black"], [ ["1", "2"], ["3"], ["4", "5"] ]]
Here are a couple of ways to obtain the desired result.
#1
arr[1].flat_map.with_index { |color,i| [arr[0]].product([color], arr[2][i]) }
#=> [[1, "green", "1"], [1, "green", "2"], [1, "blue", "3"],
# [1, "black", "4"], [1, "black", "5"]]
The steps are as follows.
enum0 = arr[1].flat_map
#=> #<Enumerator: ["green", "blue", "black"]:flat_map>
enum1 = enum0.with_index
#=> #<Enumerator: #<Enumerator: ["green", "blue", "black"]:flat_map>:with_index>
enum1 can be thought of as a compound enumerator. We can see the values that will be generated by enum1 and passed to the block by converting it to an array.
enum1.to_a
#=> [["green", 0], ["blue", 1], ["black", 2]]
The first value is generated and passed to the block, the block variables are assigned and the block calculation is performed.
color, i = enum1.next
#=> ["green", 0]
color
#=> "green"
i #=> 0
[arr[0]].product([color], arr[2][i])
#=> [1].product(["green"], )
#=> [[1, "green", "1"], [1, "green", "2"]]
The calculations are similar for the two remaining elements generated by enum1.
An alternative to this is to dup arr[2] and shift elements of the dup:
a2 = arr[2].dup
arr[1].flat_map { |color,i| [arr[0]].product([color], a2.shift) }
#2
arr[1].zip(arr[2]).flat_map { |color, a| [arr[0]].product([color], a) }
#=> [[1, "green", "1"], [1, "green", "2"], [1, "blue", "3"],
# [1, "black", "4"], [1, "black", "5"]]
The steps are as follows.
b = arr[1].zip(arr[2])
#=> [["green", ["1", "2"]], ["blue", ["3"]], ["black", ["4", "5"]]]
b[0] is passed to flat_map and the block variables are assigned and the block calculation is performed.
color, a = b[0]
#=> ["green", ["1", "2"]]
color
#=> "green"
a #=> ["1", "2"]
[arr[0]].product([color], a)
#=> [["1"]].product(["green"], ["1", "2"])
#=> [[1, "green", "1"], [1, "green", "2"]]
After the remaining elements of b are passed to map the desired array is returned by Enumerable#flat_map.
flat_map believer, I should have seen that. I edited.I needed a more generic and flexible solution as compared to the ones proposed (my bad, I should have been more clear about the requirements), so I came up with the following:
class Array
def transpose_rows
arys = self.select{|el| el.is_a?(Array)}
if arys.size == 0
[self]
else
result = []
(arys.map(&:size).max || 1).times.map{ |i|
self.map { |r|
r.is_a?(Array) ? r[i] : r
}.transpose_rows.map{|r| result << r}
}
result
end
end
end
The initial spec is that every element in the array is either a value or another array of varying depth. Each subarray "explodes" the values of the subarrays of depth-1 into an arbitrary number of "sub-values". The result should be a set of rows listing all combinations deriving from the original array.
The other solutions proposed do work for the original array I posted, which was just an example, while this one works for more complex scenarios such as the following:
[
2,
[3,4,5],
6,
[7,8,9],
[ [11,22], [33], [44,55] ],
[0, 1, 2],
[ [66], [77], [nil,99] ],
4
].transpose_rows
# => [
# [2, 3, 6, 7, 11, 0, 66, 4],
# [2, 3, 6, 7, 22, 0, nil, 4],
# [2, 4, 6, 8, 33, 1, 77, 4],
# [2, 5, 6, 9, 44, 2, nil, 4],
# [2, 5, 6, 9, 55, 2, 99, 4]
# ]
arr = [1, ["green", "blue", "black"], [ ["1", "2"], ["3"], ["4", "5"] ]]. That way, readers can refer to the variables in answers and comments without having define them, and all readers will use the same variable names.