DEV Community

Discussion on: Code Smell 65 - Variables Named After Types

View post
Collapse
 
qm3ster profile image
Mihail Malo • • Edited

Well, I mean... regex is the type. I get that this code won't confuse anyone, but it's as type-named as string or bytes or num and as uninformative as data or object.

public bool verifyPasswordFormat(string password) {
  // 2-7 lowercase chars, then 3-4 digits
  return new Regex(@"[a-z]{2,7}[1-9]{3,4}").IsMatch(password);
}
Enter fullscreen mode Exit fullscreen mode

the above is probably ideal in terms of readability, but if you have to for example hoist the Regex in a particular language to avoid recompilation, would this be a close second?

static Regex has3To7LowercaseCharsFollowedBy3or4Numbers = new Regex(@"[a-z]{2,7}[1-9]{3,4}");
public bool verifyPasswordFormat(string password) {
  return has3To7LowercaseCharsFollowedBy3or4Numbers.IsMatch(password);
}
Enter fullscreen mode Exit fullscreen mode

Some will say yes, because "it's good to declare things that do one job, and will be replaced rather than modified".

But personally, I'd end up with

// 2-7 lowercase chars, then 3-4 digits
static Regex passwordRegex = new Regex(@"[a-z]{2,7}[1-9]{3,4}");
public bool verifyPasswordFormat(string password) {
  return passwordRegex.IsMatch(password);
}
Enter fullscreen mode Exit fullscreen mode

Yes, it has a type in its name. But in a wider scope like that it might be a boon and not a sin, as that still reflects its "role" in the code, and is not an implementation detail of how the role is accomplished. You could replace that with "Pattern" but I think that's purely harmful.

And of course I renamed the function.