Allow for keyword arguments

[pdeffebach]

This just adds keyword arguments for calls of the form @subset(df, :a; view = true), rather than the multi-line version

@subset df begin 
    :a 
end

I have added this for @subset but not others yet.

[pdeffebach]

@jkrumbiegel can I get a quick check on the implementation? It borrows from DataFrameMacros.jl.

With your approval I will add the same structure to everything else.

[jkrumbiegel]

Yeah looks ok from a quick glance. I still haven't decided how keyword arguments could work for non-parentheses style.

[pdeffebach]

I guess the current code isn't very DRY, but each 4-line *_helper function is unique and it seems overly abstract to make an outer function right now.

Will plug away with tests.

[pdeffebach]

I think I have the pipeline figured out for how to process arguments. I've written it for @rtransform but need to port it to the rest of the macros.

[bkamins]

Nice!

[pdeffebach]

Okay! I think this is all done @bkamins @nalimilan @jkrumbiegel

Please take a look!

[bkamins]

is passing skipmissing explicitly tested? Why does skipmissing need to be in a tuple and later ...?

[pdeffebach]

It is explicitely tested for throwing an error. Every keyword argument is tested.

I do not know why exactly I need a tuple and splatting, but by omitting that I get

julia> @subset(df, :x .== 1; skipmissing = false)
ERROR: syntax: keyword argument "skipmissing" repeated in call to "DataFrames.subset" around /home/peterwd/Documents/Development/DataFramesMeta/src/macros.jl:765
Stacktrace:
 [1] top-level scope
   @ REPL[8]:1

so it seems necessary.

[bkamins]

ah - OK. So that is why it is needed. If you pass it explicitly uniqueness is tested, if you pass collections they are merged. This is strange, as this is not required in other contexts:

julia> (;a=1, [:a => 2]...)
(a = 2,)

(and you get merging without having to wrap the first in a tuple.

[pdeffebach]

Thanks!

[pdeffebach]

Okay I'm ready to merge this. There will always be more docs to add but I think this is now pretty comprehensive.