fix_numbers.R

fix_numbers = function(x) {

  # strip out irrelevant characters
  x = gsub('[^a-zA-Z0-9. ]', '', x)

  # map SI units
  si_map = c('k'=10^3, 'M'=10^6, 'G'=10^9)

  for (unit in names(si_map)) {
    zeros = paste0(rep('0', log10(si_map[unit])), collapse='')
    x = sub(paste0('(?<=\\d)', unit, '$'), zeros, x, ignore.case=TRUE, perl=TRUE)
  }

  # detect GST
  gst_re = '\\GST$'
  with_gst = grep(gst_re, x, ignore.case=TRUE)
  x = sub(gst_re, '', x)

  # NA-out any non-digit remnants and convert
  x[!grepl('^\\d+(\\.\\d+)?$', x)] = NA
  x = as.double(x)

  # apply GST
  gst = 0.10
  x[with_gst] = (1 + gst) * x[with_gst]

  return(x)
}

x = c('1,234,545,655.19', '$150K', '$1M', '$324G', ') $134234', 'This is not real', '$123412 + GST')
fix_numbers(x)