rrrr
How to "merge" / "map" two sparse maps into one?
What is Elixir way to “merge” / “map” two sparse maps onto one? I want to apply some function on both or single value from these maps. By “sparse” i mean that some keys are missing in another map. Like
a = %{1=>1, 2=>2, 4=>4}
b = %{0=>0, 1=>11, 2=>22, 5=>55}
looking for c=%{0=>f(0), 1=>f(1,11), …}
So far I have come with obtaining list of common keys and moving with it as index for both maps.
Closest built-in solution was Map.merge(map1, map2, fun), but it does not cover custom behavior when key is missing in one map.
Can solution be Stream’able?
Should I look for some non-default library?
Marked As Solved
LostKobrakai
Map.new(Map.keys(a) ++ Map.keys(b), fn key ->
case {Map.fetch(a, key), Map.fetch(b, key)} do
{:error, {:ok, val}} -> {key, f(val)}
{{:ok, val}, :error} -> {key, f(val)}
{{:ok, val_a}, {:ok, val_b}} -> {key, f(val_a, val_b)}
# {:error, :error} not possible
end
end)
Also Liked
Sorc96
Can you merge the maps first and then apply the function?
Map.merge(a, b, fn _key, value1, value2 -> {value1, value2} end)
|> Map.new(fn
{key, {value1, value2}} -> {key, f(value1, value2)}
{key, value} -> {key, f(value)}
end)
It could get a bit cleaner if Elixir had an equivalent of Ruby’s transform_values, then it wouldn’t be necessary to repeat the key manually.
I’m guessing you want to use Stream because of t hef function? In that case I think you could first use Stream.map and then convert the result back into a map.
hlx
map1 = %{a: 1, b: 2}
map2 = %{b: 1, c: 1}
fun = fn x, y -> x + y end
for {k1, v1} <- map1, {k2, v2} <- map2, reduce: %{} do
acc ->
if k1 == k2 do
Map.put(acc, k1, fun.(v1, v2))
else
acc
|> Map.put(k1, v1)
|> Map.put(k2, v2)
end
end
edit: after reviewing it myself this is not a great solution.
LostKobrakai
This one iterates map2 for each key/value pair in map1. That can become expensive rather quickly.
BradS2S
a = %{1=>1, 2=>2, 4=>4}
b = %{0=>0, 1=>11, 2=>22, 5=>55}
c = %{}
keys = Map.merge(a, b) |> Map.keys
for key ← keys do
case {Map.take(a, [key]), Map.take(b, [key])} do
{a_value, b_value} when map_size(b_value) == 0 → Map.put(c, key, f.(a_value))
{a_value, b_value} when map_size(a_value) == 0 → Map.put(c, key, f.(b_value))
{a_value, b_value} → Map.put(c, key, ff.(a_value, b_value))
end
end
Would something like this work?
LostKobrakai
I think you’ll get (k1+k2) *2 only if you can make sure non of the map values is whatever you store in the map as an intermediate representation for “there were matching keys with two separate values”.







