I have come across wanting to do this a couple of times in the past as well.
But before that, I do want to mention that there are two related situations
that have more elegant solutions:
1) You want to create a map or structure where fields that were not
supplied (e.g. by the user request) fall back to a default. This can be
done with patterns like this:
query = %MyRequest{
language: params["language"] || "english",
id: params["id"] || session.id
}
2) You want to pick one of a restricted set of choices; also known as
working with a 'sum type':
shape = case params["shape"] do
"square" ->
%Square{position: params["position"], width: params["width"],
height: params["height"]}
"circle" ->
%Circle{position: params["position"], radius: params["radius"]}
_ ->
raise ArgumentError, "unknown kind of shape provided"
end
If you are able to use either of these approaches in a specific use-case,
then you can reduce the number of input checks you need to do later on. (In
your original example code, all functions operating on `query` will have to
check again if `:language` and/or `:id` exist.)
Having said that, there definitely are cases where you want to normalize a
map with string keys to a map with recognized atom keys (to ensure that (a)
you dont exhaust memory by creating more and more atoms and (b) reject
inputs that you cannot handle).
The pattern I have used in the past was similar to:
def normalize_keys(into \\ %{}, input, accepted_keys) do
string_keys = MapSet.new(accepted_keys, &to_string/1)
Enum.reduce(input, into, fn {string_key, val}, acc ->
if string_key in string_keys do
put_in(acc, [:"#{string_key}"], val)
else
acc
end
end)
end
And use it as:
iex> normalize_keys(%{"language" => "english", "id" => 33, "disregard_me" =>
10}, [:id, :language])
%{id: 33, language: "english"}
Do note that this has a slightly different result from your original code,
since `string_key in string_keys` checks for the existance of a key, while
`params[string_key]` checks for the existance of a key AND the value stored
under key being truthy.
Only after writing out this code I realize its similarity to
Ecto.Changeset.cast/4 <https://hexdocs.pm/ecto/Ecto.Changeset.html#cast/4>
which seems to deal with a variant of the same problem.
A common pattern for sure, interesting.
~Marten / Qqwy
--
You received this message because you are subscribed to the Google Groups
"elixir-lang-core" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to elixir-lang-core+unsubscr...@googlegroups.com.
To view this discussion on the web visit
https://groups.google.com/d/msgid/elixir-lang-core/d307d598-a012-4b41-9700-c31f8fc07faa%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
