R/utils.strClean.R

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

#' strClean
#'
#' @param x                   string to be cleaned
#' @param DEBUG               T/F verbose output (default = FALSE)
#' @param upper               transform to upper case (default = FALSE)
#' @param forbiden.char       vector with characters to change for '_' if found anywhere (default = c(" ",".",":","-","__","(",")","[","]","{","}","$","%","&","/","\\","!"))
#' @param forbiden.last.char  vector with characters to remove if found at the end of the string (default = c("_"," "))

#'
#' @return String cleaned
#'
#' @examples
#'
#' strClean("VAR_NAM;E") #... will return "VAR_NAM_E"
#'
#' @importFrom crayon bold green red
#' @export
strClean <- function(x,DEBUG=FALSE, upper = FALSE,
                     forbiden.char = c(" ",".",":",";","-","__","(",")","[","]","{","}","$","%","&","/","\\","!"),
                     forbiden.last.char = c("_"," ")
                     ){
  # if(is.null(x)) stop("string to given")
  crayon <- "crayon" %in% installed.packages()

  if(!is.character(x)) {
    stop("x is not a string")
  }
  texto <- iconv(iconv(x, from = 'UTF-8', to = 'ASCII//TRANSLIT'), to = 'UTF-8', from = 'ASCII//TRANSLIT')

  for(c in forbiden.char){
    texto <- gsub(c,"_",texto,fixed = TRUE)
  }


  if(upper) texto <- toupper(texto)
  clean = FALSE
  while(!clean){

    last_char <- substr(texto, nchar(texto), nchar(texto))
    if(DEBUG) {
      message("Texto: ",texto)
      message("Last character: ",last_char)
    }
    if(last_char %in% forbiden.last.char) {
      texto <- substr(texto,1, nchar(texto)-1)
      if(DEBUG) message("... ",ifelse(crayon,crayon::bold("CLEANING"),"CLEANING")," ...")
    } else {
      if(DEBUG) message("... ",ifelse(crayon,crayon::bold("DONE"),"DONE")," ...")
      clean = TRUE
    }
  }

  return(texto)
}