R/interface.loadData.R

In feranpre/feR: Basic functions that I use all the time

#' Data loading
#'
#' This function helps with the data loading by wraping different functions accessible
#' in the feR package and that together make a complete and seamless data loading procedure
#'
#' @param file                      String; file name or empty if you want the `folder` listed for you to choose file
#' @param folder                    String; folder inside the project filder where data files can be found
#' @param sheetIndex                numeric (default=1); if data file is an excel/calc file what sheet should be loaded
#' @param var.to.upper              boolean (default=FALSE); return all caps
#' @param var.forbiden.chars        string vector (see details); characters to ve removed from the string
#' @param var.forbiden.last.char    string vector (see details); characters to remove if they are the last character in the string
#' @param var.empty.starts.char     strings vector (see details); characters that if found at the very beggining of the string could mean that the variable is empty
#' @param guess.factor              boolean (default=TRUE); try to guess variables that are indeed factors
#' @param max.factor.cat            num (default=10); max number of unique values after which a variable is not considered a factor anymore
#' @param verbose                   boolean (default=TRUE); if you want the function to be verbose
#' @param DEBUG                     boolean (default=FALSE); even more verbose
#'
#' @return                          data.frame; data frame with variable names corrected and nameless vars removed
#'
#' @seealso
#'
#'   This function wrap [strClean][feR::strClean()] for string cleaning (used to clean var names), [feR::loadFile()], wrapped by the function
#'
#' @seealso [roxygen2::roxygenize()], `browseVignettes("roxygen2")`
#'
#' @examples
#'
#' @importFrom crayon bold green red
#' @export
loadData <- function(file = NULL, sheetIndex=1, folder = "DATOS/",
                     var.to.upper = TRUE,
                     var.forbiden.chars = c(" ",".",":",";","-","__","(",")","[","]","{","}","$","%","&","/","\\","!"),
                     var.forbiden.last.char = c("_"," "),
                     var.empty.starts.char = c("_"),
                     guess.factor = TRUE,
                     max.factor.cat = 10,
                     verbose=TRUE,
                     DEBUG=FALSE) {

  crayon <- "crayon" %in% installed.packages()
  if (crayon) {
    crayon.x <- crayon::red(crayon::bold("✗  "))
    crayon.v <- crayon::green(crayon::bold("✔ "))
  } else {
    crayon.x <- "✗  "
    crayon.v <- "✔  "
  }



  data <- tryCatch(feR::loadFile(file, folder = folder, sheetIndex = sheetIndex),
                   error = function(e) {
                     message(crayon.x,"[loadData] File not loaded")

                   }
                   )
  # data <- feR::loadFile(file, folder = folder, sheetIndex = sheetIndex)
  if(!is.data.frame(data)) stop("No data could be loaded")
  if(verbose) message("[loadData] loading data")
  data <- feR::cleanVarNames(data)
  data <- feR::removeNamelessVars(data)

  if(guess.factor){
    if(verbose) message("[loadData] guessing factors")
      for (v in names(data)) {
        value <- dplyr::pull(data,v)
        feR:::.guess.factor(value, max.factor.cat = max.factor.cat)
        data[,v] <- value
      }
  }

  return(data)

}