R functions that shorten/filter stuff: less is more

good practice / code style / useful functions
View source 

My sticky note is full! And luckily all functions on it can be squeezed into a similar topic: making things smaller!

Make lists smaller with purrr::compact(), purrr::keep(), purrr::discard()

Once upon a time there was a list (isn’t this the beginning of all R scripts?!)

my_list <- list(
  name = "Maëlle",
  city = "Nancy",
  r_problems_encountered = Inf,
  python_skills = NULL
)

Imagine you want to get rid of NULL elements. That’s what purrr::compact() does!

purrr::compact(my_list)
#> $name
#> [1] "Maëlle"
#> 
#> $city
#> [1] "Nancy"
#> 
#> $r_problems_encountered
#> [1] Inf

Imagine you want to only keep elements that are infinite numbers. For some unknown reason is.infinite() works on strings (but not on NULL) so the following works.

my_list |>
  purrr::compact() |>
  purrr::keep(is.infinite)
#> $r_problems_encountered
#> [1] Inf

Similarly, imagine you want to discard all elements that are character.

purrr::discard(my_list, is.character)
#> $r_problems_encountered
#> [1] Inf
#> 
#> $python_skills
#> NULL

We’ve only used built-in functions as predicate functions (the second argument of all these calls). You can really unleash the power of these three functions when pairing them with your own predicate functions!

Related to these functions, don’t miss Jonathan Carroll’s great post “Four Filters for Functional (Programming) Friends”.

Make string (vectors) smaller with stringr::str_squish(), stringr::str_subset()

This week when removing stringr from the dependencies of the glitter package I discovered stringr::str_subset(). It will help you keep, in purrr’s parlance, only the elements of a string vector that have a match with a pattern.

string <- c("R", "Python", "Julia", "Ruby")
stringr::str_subset(string, "R")
#> [1] "R"    "Ruby"

It will also help you discard them if you use the negate arguments.

stringr::str_subset(string, "R", negate = TRUE)
#> [1] "Python" "Julia"

In glitter, where I chose to write similar-looking functions for the stringr functions that we used, I replaced stringr::str_subset() with

str_subset <- function(x, pattern, negate = FALSE) {
  grep(pattern, x, value = TRUE, invert = negate)
}

since its docs mentioned this base R’s equivalence.

Another stringr function that makes things smaller is stringr::str_squish() that removes useless whitespace from a string vector: whitespace at the start and end, and repeated whitespace in the middle.

text <- " So much\n\n useless    whitespace    "
stringr::str_squish(text)
#> [1] "So much useless whitespace"

It’s documented on the same page as stringr::str_trim() which is stringr’s own trimws() (a base R function) that only removes whitespace from the start and end. 💇

Make your code shorter with purrr::partial()

I am working on a vignette for glitter where I write (well, copy-paste) the same function call over and over again.

blablabla %>%
  spq_perform(
    endpoint = "https://ld.admin.ch/query", 
    request_type = "body-form"
  )

From the developer’s perspective it makes me wonder whether we should add some options as default. From an user’s perspective it reminded me of purrr::partial()! With purrr::partial one easily creates a function that corresponds to a function with some arguments pre-filled.

So, as a glitter user, I could use

my_spq_perform <- purrr::partial(
    spq_perform,
    endpoint = "https://ld.admin.ch/query", 
    request_type = "body-form"
)

blablabla %>%
  my_spq_perform

Quite neat!

Conclusion

In this short (of course) post I presented functions for making lists, string vectors and code shorter by filtering or squishing them. 🔨