Helper to handle different ways we might choose columns

Tidyselect has a robust way of selecting columns, but we often need to do that selection indirectly- ie pass the selecting in as an argument to an outer function. There are a number of ways to do this, but this is reasonably robust and general, relying on using tidyselect itself. Allows passing character vectors, bare names, or tidyselect syntax, provided this function is called at the right place in the call stack

Usage

selectcreator(selectvals, data, failmissing = TRUE)

Arguments

selectvals

the selection of columns. Can be character, bare names, or tidyselect syntax, e.g. tidyselect::starts_with(). If tidyselect, it can include objects, but they need to be !!. For example, if prefix = 'ABC', to select columns starting with that prefix, should call selectvals = tidyselect::starts_with(!!prefix)

data

a tibble or dataframe to find the columns

failmissing

logical, default TRUE: fail if the requested grouping or aggregation columns not exist. If FALSE, proceed with those that do exist and silently drop those that don't

Value

a character vector of column names. The intention is to use them in dplyr::group_by or select with dplyr::select(dplyr::across({{output}})). The underlying eval_select returns a named integer vector giving column indices, but we return only the names because in use the indices may not be stable throughout the calling function(s)