R/pii_table.R

In DevikaNair90/prybar: Toolkit for Identifying Hidden PII in Text Data

#' Creates a summary table of PII for a dataset.
#'
#' This function takes a dataframe and searches each column for presence of 
#' any text that appears as PII: emails, DOB, SSNs, and phone numbers. 
#'
#' @param df A dataframe with (vector-able) variables.
#' @param writeout TRUE/FALSE indicator of whether or not full detailed results to be written out. An individual table will be written out for each column, with PII results for every row of the table. 
#' @param path destination for tables to be written out
#' @import stringr
#' @import maditr
#' @suggest generator
#' @export
#' @examples
#' set.seed(2)
#' testcase <- data.frame(name = generator::r_full_names(2),  
#' snn = generator::r_national_identification_numbers(2), 
#' dob = generator::r_date_of_births(2), 
#' email = generator::r_email_addresses(2), 
# ip = generator::r_ipv4_addresses(2), 
#' phone = generator::r_phone_numbers(2),  
#' credit_card = generator::r_credit_card_numbers(2), 
#' lat = generator::r_latitudes(2), 
#' lon = generator::r_longitudes(2), 
#' stringsAsFactors = FALSE)
#' 
#' pii_table(testcase, writeout = FALSE)
#' 
# devtools::load_all()
# devtools::use_package(package = "stringr", type = "import")
# devtools::use_package(package = "dplyr", type = "import")

pii_table <- function(df, path, writeout) {
  df_name <- deparse(substitute(df))
  df <- maditr::as.data.table(df)
  class_table <- maditr::data.table("Column" = colnames(df)) %>%
    maditr::dt_mutate("Reference" = paste0(df_name, "$", Column)) %>% 
    maditr::dt_mutate(class = character(length = length(colnames(df))))
  
  for (i in 1:ncol(df)) {
    class_table$class[i] <- class(df[[i]])
  }
  
  # numvars <- class_table %>% filter(class %in% c("numeric", "integer", "double")) 
  # charvars <- class_table %>% filter(class == "character")
  # 
  # df_numbers <- df[, as.character(numvars$Column)]
  
  # if (is.data.frame(df_numbers) == FALSE) {
  #   df_numbers <- data.frame(NUM = df_numbers) 
  # }
  # 
  # columnsummary <- data.frame("Column" = colnames(df_numbers))
  
  dt_email <- list(data.table::data.table())
  dt_ssn <- list(data.table::data.table())
  dt_dob <- list(data.table::data.table())
  dt_phone <- list(data.table::data.table())
  #df_zipcode <- list(data.frame())
  
  for (i in 1:ncol(df)) {
    var <- as.character(df[[i]])
    dt_dob[[i]] <- search_DOB(var, output = "dt")
    dt_ssn[[i]] <- search_ssn(var, "dt")
    dt_email[[i]] <- search_email(var, "dt")
    dt_phone[[i]] <- search_phone(var, "dt")
   # df_zipcode[[i]] <- search_zipcode(var, "df")
    class_table$dobrisk[i] <- dt_dob[[i]] %>% .$DOBYN %>% sum #sum(search_DOB(var), na.rm = TRUE)
    class_table$emailrisk[i] <- dt_email[[i]] %>% .$EmailYN %>% sum  #sum(search_email(var), na.rm = TRUE)
    class_table$phonerisk[i] <- dt_phone[[i]] %>% .$PhoneYN %>% sum  #sum(search_phone(var), na.rm = TRUE)
    class_table$ssnrisk[i] <- dt_ssn[[i]] %>% .$SSNYN %>% sum  # sum(search_ssn(var), na.rm = TRUE)
    
  }
  
  if (writeout == TRUE) {
    saveRDS(object = class_table, file = paste0(path, df_name, "_PII_0_Summary_", Sys.Date(), ".RDS")) 
    
    saveRDS(object = dt_dob, file = paste0(path, df_name, "_PII_1_DOB_", Sys.Date(), ".RDS"))
    saveRDS(object = dt_ssn, file = paste0(path, df_name, "_PII_2_SSN_", Sys.Date(), ".RDS"))
    saveRDS(object = dt_email, file = paste0(path, df_name, "_PII_3_EMAIL_", Sys.Date(), ".RDS"))
    saveRDS(object = dt_phone, file = paste0(path, df_name, "_PII_4_PHONE_", Sys.Date(), ".RDS"))
  }
  
  print("summary completed")
  print(Sys.time())
  
  print(class_table)
  
}