New linter for `utils::hasName()`

[m-muecke]

Since R 3.4.0 utils::hasName() was added as a convient way of testing the existence of a name i.e. the pattern "name" %in% names(x) or for checking the !is.null(x$abc) pattern without partial matching (taken from the docs):

 x <- list(abc = 1, def = 2)
 !is.null(x$abc) # correct
 !is.null(x$a)   # this is the wrong test!
 hasName(x, "abc")
 hasName(x, "a")

There is also the basically equivalent rlang::has_name() as a common pattern.

[AshesITR]

Note that $ can be overridden. reticulate does this for example. "a" %in% names(x) is probably fine to lint.

[m-muecke]

Note that $ can be overridden. reticulate does this for example. "a" %in% names(x) is probably fine to lint.

As well as active bindings in R6. The name checking seems to be the only relevant case.

[MichaelChirico]

I'm slightly ambivalent on whether hasName(x, nm) is better than nm %in% names(x), seem similar. Happy to host the linter though, it seems reasonable enough as a case of "use the specific function from base".

Since %in% is just match() under the hood, we can also detect the direct match() equivalent, which is much worse:

match(nm, names(x), 0L) > 0L

[MichaelChirico]

WDYT about making this a match_linter(), which could also include the proposed in_linter()? (#2621 / #1875)

Or is it too obscure to expect people to know %in% and match() are so closely linked?