Validating Input

Part of the Logic & Flow section of Coddy's R journey. Lesson 55 of 64.

Input from users or files is text and can be anything. Check it before using it: suppressWarnings(as.numeric(s)) gives NA for text that is not a number, and is.na() tests for that:

read_age <- function(s) {
  n <- suppressWarnings(as.numeric(s))
  if (is.na(n)) return("not a number")
  if (n < 0 || n > 130) return("out of range")
  paste("age", n)
}
print(read_age("36"))
print(read_age("abc"))
print(read_age("200"))

Output:

[1] "age 36"
[1] "not a number"
[1] "out of range"

A validator can collect every problem instead of stopping at the first, which gives more useful feedback. Here it returns a vector of messages, empty when the input is fine:

check_password <- function(p) {
  problems <- c()
  if (nchar(p) < 8) problems <- c(problems, "too short")
  if (!grepl("[0-9]", p)) problems <- c(problems, "needs a digit")
  problems
}
print(check_password("abc"))
print(length(check_password("secret123")))

Output:

[1] "too short"     "needs a digit"
[1] 0

Pick the style by who calls the function: return a message when the caller expects bad input and shows it to a user; raise an error with stop() when bad input means a bug that should not go unnoticed.

# the caller expects bad input and shows the message
read_age("abc")                     # "not a number"
# bad input is a mistake in the program
if (!is.numeric(x)) stop("x must be numeric")

Validating with tryCatch() around the parsing step keeps the main code free of checks: each record is either converted or reported:

parse_record <- function(line) {
  parts <- strsplit(line, ",")[[1]]
  if (length(parts) != 2) stop("expected 2 fields")
  qty <- suppressWarnings(as.integer(parts[2]))
  if (is.na(qty)) stop("quantity is not a whole number")
  list(item = parts[1], qty = qty)
}
for (line in c("tea,2", "jam", "oil,x")) {
  r <- tryCatch(parse_record(line), error = function(e) conditionMessage(e))
  cat(line, ":", if (is.list(r)) paste("ok", r$qty) else r, "\n")
}

Output:

tea,2 : ok 2 
jam : expected 2 fields 
oil,x : quantity is not a whole number

Challenge

Easy

Complete check_signup(name, age, email). Collect every problem, in this order: name is empty (after removing spaces at the ends), age must be a number, age must be 13 or older (only when the age is a number), and email needs @. Return ok when there is no problem, otherwise the problems joined with ; .

The supplied code reads the name, the age and the email (one per line) and prints the returned value.

Try it yourself

check_signup <- function(name, age, email) {
  # Write your code here
  "ok"
}

# Supplied input/output code: keep it as it is
input <- suppressWarnings(readLines(file("stdin")))
cat(check_signup(input[1], input[2], input[3]), sep = "\n")