You don't "sanitize" input, you "sane-itize" input. The whole point of checking the input is to make sure it's valid, not to try to scrape away cruft you don't think is valid.

Example: phone numbers. Is the input you accepted a phone number? Well, if you were just "sanitizing input", you might pass it through some generic "sanitizing function" that just checks if it had "malicious characters" or something and strip those out. But what you should actually be checking is is this a phone number? By making sure the input is what it's supposed to be, you not only gain a better security posture, but you improve your program's reliability by making it operate in the way you expect it to.

Some input fields like "give me a random block of text and I'll store it" are very hard to validate, so for those fields you can encode them as Base64 for storage, and at output time decide how to format them safely.

Also consider wrappers for any functions which pass input from a user. Perl's Taint Mode (https://en.wikipedia.org/wiki/Taint_checking) is a global way to enforce this, but for languages that don't have a Taint Mode, you'll have to implement it yourself.