Unicode characters are mangled in JavaScript kata output

[paul-calvelage]

Unicode characters in the JavaScript kata output are being mangled.

Each byte of the UTF-8 encoding seems to be printing as a separate Unicode character. So the Chinese greeting 你好 displays as ä½ å¥½.

This is very bad for anyone needing more than 7-bit ASCII.

Now for some examples of what I think may be happening. This code in a JavaScript kata:

console.log("£");

displays as the two characters £ ("\u00c2\u00a3"). The Unicode code point for £ ("\u00a3") is normally encoded in UTF-8 as 0xc2a3. But Codewars apparently re-encodes each byte: 0xc2, 0xa3 to get £.

This:

console.log("\uffff")

is displayed as three characters ï¿¿ ("\u00ef\u00bf\u00bf"). The Unicode code point 0xffff is normally encoded in UTF-8 as 0xefbfbf. But as above, Codewars then seems to re-encode 0xef, 0xbf, 0xbf to ï¿¿.

I could give as many examples as there are multiple-byte UTF-8 encodings, but this suffices to show the pattern for a single character. Longer strings just repeat the problem, so that console.log("£££££"); displays as £££££ for example.

As I said, this seems to be pretty serious for anyone needing Unicode.

---

Note: I discovered this while completing Simple Change Machine, which uses the pound symbol.

[paul-calvelage]

The Chinese Numeral Encoder kata has lots of Unicode characters, so it is affected by this bug. In this screenshot it is noticeable in the test output.

[paul-calvelage]

My previous screenshot was in Chrome on Windows 10. In Firefox I get a slightly different (but still incorrect) output:

[kazk]

#902