Invalid invisible unicode characters in phone numbers

[coolaj86]

I came across a phone number that ended in a zero-width, non-printing (U+202C, "Pop Directional Formatting").

This could be fixed either by cleaning up the database or in client side code by replacing all whitespace with a single whitespace, and then removing any characters except those in the ascii ranges of

I'll write in JavaScript for ease of translation to any other frontend or backend language:

function sanitize(tel) {
phone = phone
.replace(/\s+/g, ' ') // replace all whitespace with a single space
.replace(/[^ -;]/, '') // remove all characters except in the ascii range between space and semicolon (includes # + , . ( ) etc)
.trim() // trim leading and trailing whitespace
}

// Test cases

// I've pasted this with the U+202C (granted, it may not survive the BB submit)
sanitize(`+1 (385) 555-1337‬`) // `+ 1 (385) 555-1337`
// this could easily be extended to include alphas (e.g. p, w) if desired
sanitize(`+1 (ban) ana-1337`) // `+1 () -1337`
sanitize(`+ 1 (385) 555-1337#10,2,2;2*`) // `+ 1 (385) 555-1337#10,2,2;2*`
sanitize(" ") // ""

Also, since there are a very limited number of phone number formats around the globe, this could be further cleaned up in terms on consistent formatting by recognizing the +1 or country of the individual, removing non-touch-tone separator characters ( ) . - and leave control characters like # , etc and if the length of the number before # matches the expected number of characters and pattern, then rewrite the number in canonical form in the database (e.g. +13855551337), and apply a client-side formatter to represent a localized form (e.g. (385) 555-1337).

Similarly, a pattern could be used to apply formatting to invalid data for easier recognition by ward leaders who are trying to clean up the data.