R/parse_income.R

In runapp-aus/strayr: Ready-to-use Australian common structures and classifications and tools for working with them

#' @title Parse income ranges
#'
#' @name parse_income_range
#'
#' @description parse_income_range extracts dollar figures from income bucket ranges used in Census
#' data and elsewhere
#'
#' @param income_range a character vector containing income ranges, e.g.
#' \code{"b. $18,201 to $37,000"}, \code{"$400-$499 ($20,800-$25,999)"}
#' @param limit either "lower" (the default) and "upper" bounds of the range, or "mid" to return the mid-point
#' @param max_income numeric value of 'and over' value of the highest bracket. \code{Inf} by default.
#' If \code{NULL}, returns a max_income value equal to the lower bound of the
#' highest bracket. Important for \code{limit = "mid"} or \code{limit = "upper"}.
#' @param is_zero a character vector of values to be interpreted as zero. Defaults to \code{c("Nil income")}.
#' @param negative_as_zero a boolean value. If TRUE (the default) the character
#' @param dollar_prefix a character string to denote the dollar delimiter.
#' \code{"$"} by default, but could be e.g. \code{"AUD"} or \code{"USD"}.
#' string "Negative income" is interpreted as \code{0}. If FALSE, \code{NA} is returned.
#' @param .silent a boolean value. If FALSE (the default), the function will warn that  \code{NA}(s) were returned.
#' @param .print_lookup If TRUE prints lookup table displaying inputted income range and the returned parsed value.
#'
#' @return a numeric vector of incomes
#'
#' @importFrom stringr str_detect str_split str_replace_all
#' @importFrom readr parse_number
#' @importFrom dplyr left_join rename select
#'
#' @export
#'

globalVariables(c("transformed", "raw_income"))

parse_income_range <- function(income_range,
                                limit = "lower",
                                max_income = Inf,
                                is_zero = c("Nil income"),
                                negative_as_zero = TRUE,
                                dollar_prefix = "$",
                                .silent = TRUE,
                                .print_lookup = FALSE) {

  parse_income_range_ <- function(income_range,
                                  limit = "lower",
                                  max_income = Inf,
                                  is_zero = c("Nil income"),
                                  negative_as_zero = TRUE,
                                  dollar_prefix = "$",
                                  .silent = TRUE) {

    # correct for use in regex
    if (stringr::str_detect(dollar_prefix, "\\$")) {
      dollar_prefix <- stringr::str_replace_all(dollar_prefix, "\\$", "\\\\$")
    }

    # run
    purrr::map_dbl(income_range,
                   .parse_income_range_single,
                   .limit = limit,
                   .negative_as_zero = negative_as_zero,
                   .is_zero = is_zero,
                   .max_income = max_income,
                   .dollar_prefix = dollar_prefix,
                   silent = .silent)

  }

  .parse_income_range_single <- function(x,
                                         .limit,
                                         .negative_as_zero,
                                         .is_zero,
                                         .max_income,
                                         .dollar_prefix,
                                         silent) {

    # Non-numeric specials
    if (is.na(x)) return(NA_real_)
    if (x %in% .is_zero) return(0)
    if (stringr::str_detect(x, "(N|n)egative")) {
      if (.negative_as_zero) return(0) else return(NA_real_)
    }
    if (x %in% c("Not stated", "Not applicable", "Total", "All")) return(NA_real_)

    # Return NA if there is no number
    if (!stringr::str_detect(x, .dollar_prefix)) return(NA_real_)

    # Remove cruft before .dollar_prefix
    stripped_x <- stringr::str_remove_all(x, paste0("[^0-9", .dollar_prefix, "]*"))

    # 'or more' and 'or less'
    if (stringr::str_detect(x, " more")) { # 'X or more'

      # get first only
      max_lower <- suppressWarnings(readr::parse_number(stripped_x))

      # get max_upper
      if (!is.null(.max_income)) {
        max_upper <- .max_income
      } else {
        max_upper <- max_lower
      }
      all <- c(max_lower, max_upper)

    } else if (stringr::str_detect(x, " less")) { # 'X or less'

      # get first only
      min_upper <- suppressWarnings(readr::parse_number(stripped_x))
      all <- c(0, min_upper)

    } else {
      # otherwise, extract the $numerics
      all <- stringr::str_split(stripped_x, pattern = .dollar_prefix)
      all <- unlist(all)

      all <- suppressWarnings(readr::parse_number(all))
      all <- all[!is.na(all)]
    }

    # too long or short?
    len_all <- length(all)
    # If too long, take the first 2
    if (len_all > 2) {

      if (!silent) message("More than two dollar figures found in ", x, "; using the first two")

      all <-  all[1:2]

      # If none found, return NA
    } else if (len_all == 0) {

      if (!silent) message("No dollar signs found in ", x, ". Returning NA")

      return(NA_real_)
    }

    if (.limit == "lower") return(all[1])
    if (.limit == "upper") return(all[2])
    if (.limit == "mid")   return((sum(round(all, -1))) / 2)

  }

  orig <- as.data.frame(income_range)
  colnames(orig) = c('raw_income')

  lookup <- orig %>%
    unique() %>%
    mutate(transformed = parse_income_range_(income_range = raw_income,
                                             limit,
                                             max_income,
                                             is_zero,
                                             negative_as_zero,
                                             dollar_prefix,
                                             .silent))

  parsed_output <- orig %>%
    left_join(lookup, by = 'raw_income') %>%
    select(transformed) %>%
    unlist() %>%
    as.vector()

  if (.print_lookup) {
    lookup %>% rename(`Income Range` = raw_income,
                      `Parsed Income Value` = transformed)
  } else {
    return(parsed_output)
  }

}