[Question]: Does value_choices support delayed_data?

[chlebowa]

What is your question?

value_choices seems to support the concept of delayed_data:

However, var_choices must be a character vector, delayed_data objects are not accepted.

Is this an oversight or am I missing something?

Also, the subset argument can accept a function that only takes a data argument, in keeping with other similar cases throughout the package, but this seems insufficient. In the example given in the documentation:

# functional subset
value_choices(ADRS, "PARAMCD", "PARAM", subset = function(data) {
  levels(data$PARAMCD)[1:2]
})

One must specify the variable manually and, again, delayed data does not seem to be supported.
Wouldn't something like this be more thorough?

value_choices(ADRS, "PARAMCD", "PARAM", subset = function(data, variable) {
  levels(data[[variable]])[1:2]
})

This way one could do

value_choices(
  data = "ADRS", 
  var_choices = variable_choices("ADRS", subset = function(data) names(Filter(is.factor, data))), 
  subset = function(data, variable) levels(data[[variable]])[1:2]
)

Or the function passed to subset could operate on a variable rather than the whole dataset.

Finally, passing choices_labeled to var_choices may run into another issue: [.tbl_df has very strict argument checks, which results in

> data <- cdisc_data(ADSL = rADSL)
> value_choices(
+   data[["ADSL"]],
+   var_choices = variable_choices(data[["ADSL"]], function(data) names(Filter(is.factor, data)))
+ ) |> summary()
Error in `data[var_choices]`:
! Can't subset columns with `var_choices`.
✖ `var_choices` must be logical, numeric, or character, not a <choices_labeled/character> object.
Run `rlang::last_trace()` to see where the error occurred.

but convert the tibble to a data.frame and

> data <- cdisc_data(ADSL = rADSL) |> within(ADSL <- as.data.frame(ADSL))
> value_choices(
+   data[["ADSL"]],
+   var_choices = variable_choices(data[["ADSL"]], function(data) names(Filter(is.factor, data)))
+ ) |> summary()
         Length          Class1          Class2            Mode 
            400 choices_labeled       character       character 

It seems value_choices.data.frame should rather do data[as.vector(var_choices)].

Code of Conduct

• I agree to follow this project's Code of Conduct.

Contribution Guidelines

• I agree to follow this project's Contribution Guidelines.

Security Policy

• I agree to follow this project's Security Policy.