R/ir_remove_missing.R

In ir: Functions to Handle and Preprocess Infrared Spectra

#' Removes empty data values in an object of class `ir`
#'
#' `ir_remove_missing` takes and object of class `ir` and removes all
#' rows in the `data.frame`s of the list column `spectra` that have
#' `NA` intensity values (column `y`). Additionally, one can specify
#' to remove rows in the `ir` object to discard if they contain empty
#' spectra.
#'
#' @param x An object of class [`ir`][ir_new_ir()].
#'
#' @param remove_rows A logical value indicating if rows in `x` with empty
#' spectra should be discarded (`remove_rows = TRUE`) or not
#' (`remove_rows = FALSE`).
#'
#' @return `x` with cleaned spectra.
#'
#' @examples
#' # create sample data with some missing rows and one entire missing spectra
#' x <-
#'    ir::ir_sample_data
#' x$spectra[[1]] <- x$spectra[[1]][0, ]
#' x$spectra[[2]][1:100, "y"] <- NA_real_
#'
#' # remove missing values (but remove no rows in x)
#' x1 <-
#'    x %>%
#'    ir::ir_remove_missing(remove_rows = FALSE)
#'
#' # remove missing values (and remove rows in x if a compete spectrum is
#' # missing)
#' x2 <-
#'    x %>%
#'    ir::ir_remove_missing(remove_rows = TRUE)
#'
#' nrow(x)
#' nrow(x1)
#' nrow(x2)
#'
#' @export
ir_remove_missing <- function(x,
                              remove_rows = FALSE) {

  ir_check_ir(x)
  if(!is.logical(remove_rows)) {
    rlang::abort(paste0("`remove_rows` must be a logical vector, not", class(remove_rows)[[1]], "."))
  }
  if(length(remove_rows) != 1) {
    rlang::abort("`remove_rows` must be of length 1, not ", length(remove_rows), ".")
  }

  x$spectra <- purrr::map(x$spectra, function(x) x[!is.na(x$y),])

  if(remove_rows) {
    x <- x[purrr::map_int(x$spectra, nrow) != 0,]
  }
  x

}