ndc_cleanup_R.R

cleanup_ndc <- function(package_path, product_path){
  require(dplyr)
  require(stringr)
  
  dpackage <- read_delim(package_path, delim = "\t")
  dproduct <- read_delim(product_path, delim = "\t")
  
  names(dproduct) <- tolower(names(dproduct))
  names(dpackage) <- tolower(names(dpackage))
  
  d <- inner_join(dproduct, dpackage)
  
  isFmt442 <- grepl("[[:alnum:]]{4}-[[:alnum:]]{4}-[[:alnum:]]{2}", d$ndcpackagecode)
  isFmt532 <- grepl("[[:alnum:]]{5}-[[:alnum:]]{3}-[[:alnum:]]{2}", d$ndcpackagecode)
  isFmt541 <- grepl("[[:alnum:]]{5}-[[:alnum:]]{4}-[[:alnum:]]{1}", d$ndcpackagecode)
  fmt <- rep(NA, nrow(d))
  fmt[isFmt442] <- "4-4-2 format"
  fmt[isFmt532] <- "5-3-2 format"
  fmt[isFmt541] <- "5-4-1 format"
  table(fmt, useNA="ifany")
  d$fmtPackageCode <- format(fmt)
  
  ndcSplit <- strsplit(d$ndcpackagecode, "-")
  ndcSplit <- data.frame(matrix(unlist(ndcSplit), ncol=3, byrow=TRUE))
  
  d$ndcPackageFmt11[isFmt442] <- paste0("0", ndcSplit$X1[isFmt442], ndcSplit$X2[isFmt442], ndcSplit$X3[isFmt442])
  d$ndcPackageFmt11[isFmt532] <- paste0(ndcSplit$X1[isFmt532], "0", ndcSplit$X2[isFmt532], ndcSplit$X3[isFmt532])
  d$ndcPackageFmt11[isFmt541] <- paste0(ndcSplit$X1[isFmt541], ndcSplit$X2[isFmt541], "0", ndcSplit$X3[isFmt541])
  
  return(d)
}