Add Access.values/0

[xxdavid]

Hi, I am adding Access.values/0 as discussed here, and also Access.keys/0 for parity with keys. If you think Access.keys/0 is not useful and should not be added, I can remove it from the PR.

I am not adding any explicit unit tests as I find the coverage by doctests enough, but I can write some if you wish so.

Thank you for reviewing the changes!

[sabiwara]

Suggested change

	@spec values() :: Access.access_fun(data :: map(), current_value :: list())
	@doc since: "1.19.0"
	@spec values() :: Access.access_fun(data :: map(), current_value :: list())

[xxdavid]

Added in f7705f4. Thanks for noticing!

[sabiwara]

Suggested change

	@spec keys() :: Access.access_fun(data :: map(), current_value :: list())
	@doc since: "1.19.0"
	@spec keys() :: Access.access_fun(data :: map(), current_value :: list())

[sabiwara]

LGTM, thanks @xxdavid ! 💜

[josevalim]

Let's not add keys for now but I'd suggest to make it so values works both with keyword lists and maps. WDYT?

[xxdavid]

Okay, I'll drop keys/0. Supporting also keyword lists sounds reasonable. I'll add it to the PR.

[xxdavid]

I added the support for keyword lists in the last commit. I was thinking about what to allow as a keyword list and I've come to a pragmatic decision to check whether the first element of a list is a 2-tuple and the first element of that tuple is an atom. If not, we'll raise with Access.values/0 expected a map or a keyword list, got: …. It may fail with FunctionClauseError if given for example [{:a, 1}, {:b, 2, :haha}] but I think that's acceptable as for example Keyword.filter/1 would fail as well.

I've also added unit tests as it got a bit more complex. And I've also found that some doctests for values/0 were relying on a particular or of keys in a map, which I fixed also in the last commit.

[sabiwara]

Perhaps we should still have guards to be sure it's a keyword?

Suggested change

	Enum.map(data, fn {_key, value} -> next.(value) end)
	end
	defp values_keyword(:get_and_update, data, next) do
	{reverse_gets, reverse_updated_data} =
	Enum.reduce(data, {[], []}, fn {key, value}, {gets, data_acc} ->
	Enum.map(data, fn {key, value} when is_atom(key) -> next.(value) end)
	end
	defp values_keyword(:get_and_update, data, next) do
	{reverse_gets, reverse_updated_data} =
	Enum.reduce(data, {[], []}, fn {key, value}, {gets, data_acc} when is_atom(key) ->

[xxdavid]

I looked at Keyword.filter/2 and Keyword.values/1 and they don't seem to be that strict. But why not. 🙂 Added in 8b4aa0f.

[josevalim]

💚 💙 💜 💛 ❤️