R/c.R

In testarguments: Test (Multiple) Arguments of a User-Defined Prediction Algorithm

#' @title Combine objects of class \code{'testargs'}
#'
#' @description Combines an arbitrary number of \code{'testargs'} objects
#'
#' @details  If the argument and diagnostic names are inconsistent across objects,
#' the combined \code{'testargs'} object is constructed by simply taking the union
#' of all argument and diagnostic names. Then, \code{rbind.fill()} is used to
#' combine the diagnostic data, producing intentional \code{NA} values where appropriate.
#'
#'
#' @param x object of class \code{'testargs'}
#' @param ... objects of class \code{'testargs'} to be combined with \code{x}
#' @return An object of class \code{'testargs'}, the result of combining \code{x} and \code{...}
#' @export
setMethod("c", signature="testargs", function(x, ...) {

  object_list <- list(...)

  if (!all(sapply(object_list, function(x) is(x, "testargs"))))
    stop("all objects to be combined should be of class testargs")

  object_list <- c(list(x), object_list)

  new_df <- rbind.fill(lapply(object_list, function(x) x@diagnostics_df))

  new_arg_names  <- x@arg_names
  new_diag_names <- x@diagnostic_names
  for (i in 2:length(object_list)) {
    new_arg_names  <- union(new_arg_names,  object_list[[i]]@arg_names)
    new_diag_names <- union(new_diag_names, object_list[[i]]@diagnostic_names)
  }

  return(new("testargs",
             diagnostics_df = new_df,
             arg_names = new_arg_names,
             diagnostic_names = new_diag_names
  ))
})