toASCII returns incorrect result

[nucks]

While converting various emojis, I've found that very often it is not converting them correctly.

What I'd expect:
1. punycode.toASCII( "❄️🙊" ) should equal "xn--tdix719m" where instead it equals "xn--tdiy444e5vxg" (invalid punycode)
2. punycode.toASCII( "👰✉️" ) should equal "xn--4bi6168m" where instead it equals "xn--4biw254ehqwg" (invalid punycode)

I could give many more examples, but I haven't found a pattern in the mistake.

[mathiasbynens]

What makes you say that’s the expected output, and that the current output is “invalid Punycode”? Could you elaborate please?

Update: Entering e.g. ❄️🙊.la in Chrome’s address bar displays xn--tdix719m.la.

[nucks]

Yes, sorry. I've been converting many emojis and because the punycode isn't correct many of them haven't been appearing correctly, which is what pushed me to investigate further. I tested on many conversion sites to test that out. The main one I trust most is Punycoder. If you try converting those strings above, you will find the results I was describing.

[hashtag2949]

[hashtag2949]

Cluster-Nobes has given a networking drive & wifi-drive. Having first error on Nobe-3

[mjethani]

This is not really a bug. It's a limitation.

> [...'❄️🙊']
[ '❄', '️', '🙊' ]
> 

The second character is U+FE0F. If you drop it from the input, you get the correct result. Take a look at idna-uts46 for a more detailed explanation.

Here's a quick alternative that works in the browser as well as Node.js:

let toASCII = d => d.split('.').map(l => new URL(`ws://${l}`).hostname).join('.');