TLDR; Seeking feedback on what, if any, normalisations and validations should be provided when accepting input that represents a person’s name.

Context

When you accept input for the name of a person, what validation or normalisation do you normally do?. CLDR 48 will be out very soon and with it new releases of many of the ex_cldr libraries.

One of these is ex_cldr_person_names which supports formatting names in a locale-sensitive way. I have an opportunity to provide some validations or normalisations as optional capabilities.

Person Name Validation - current practise

When implementations allow entry of person names, they can be too strict; there are instances where people can’t enter their real names, such as O’Brian, Stéphanie, Wałęsa, Þjóðólfr.

When you accept input for a person’s name do you:

• Do nothing - as long as its UTF-8 it’s ok?
• Restrict to a single script? Maybe thats too lenient - even the Latin script has over 1_400 characters in it
• Do you allow emoji characters, like the name B​b?
• What about the always fun Zalgo text?

Possible normalisations

• Transform to Unicode NFC format
• Replacement of arbitrary sequences of whitespace characters by a single space .
  • \p{whitespace}{2,∞} → U+0020
• Replacement of U+2010 HYPHEN and U+2011 NON-BREAKING HYPHEN
  • [‐‑] → -

Additional possible constraints

• Too many identical grapheme clusters in a sequence
  • (Tóóóóóm)
• Too many non-letters in a row
  • (Jean—Luc Jr.., MD)
• Too many combining marks in a row
  • Faruq̣̣̈̈

Not in scope

Removing profane words or otherwise accepting or rejecting words according to some dictionary.