audit use of ... in package

[EmilHvitfeldt]

The ... are used throughout the package for different things.

Many of the steps define them themselves where it would probably be better if they were to be inherited.

Other functions such as prep() says ... aren't used, and should probably be checked as I suspect that some people are using prep(data = ddd) instead of prep(training = ddd)

[EmilHvitfeldt]

library(tidyverse)
library(recipes)

steps <- ls(getNamespace("recipes")) |>
  stringr::str_subset("^step_") |>
  stringr::str_subset("_new$", negate = TRUE) |>
  setdiff("step_testthat_helper")


get_arg <- function(x) {
  res <- getFromNamespace(x, getNamespace("recipes"))
  res <- formals(res)
  res$keep_original_cols
}

get_doc <- function(x) {
  readLines(paste0("man/", x, ".Rd")) |>
    paste(collapse = " ") |>
    str_extract("\\item\\{\\.\\.\\.\\}\\{.*?\\item") |>
    str_remove("\\item\\{\\.\\.\\.\\}\\{") |>
    str_remove("\\} *\\\\item")
}

dots <- tibble(
  name = steps
) |>
  mutate(doc = map_chr(name, get_doc)) |>
  filter(str_detect(name, "impute$", negate = TRUE)) 

dots |>
  count(doc) |>
  pull(doc)

dots |>
  filter(is.na(doc))