Parse unicode escapes \uxxxx \Uxxxxxxxx about luajit HOT 6 CLOSED

What about Lua5.3-compatible \u{XXX} where XXX is a sequence of 1 or more hexadecimal digits representing the character code point?

from luajit.

Maybe in addition. And, yes, I think this is just another pointless and stupid Lua 5.3 weirdness. Why should one be bothered to enter raw UTF-8 as hex values with \u{abcd}?!? This can already be entered as \xab\xcd -- that proprietary \u{} syntax doesn't help at all.

The point is to be able to enter or cut'n'paste a hex code point literally, not having to convert it to UTF-8 by hand. What an ill-conceived 'feature'.

Following the lead of Python, Java, Javascript etc. and implementing \u and \U just makes so much more sense.

from luajit.

I've already implemented this lexer extension in TvmJIT

see https://github.com/fperrad/tvmjit/blob/master/src/tj_lex.c#L171

from luajit.

This can already be entered as \xab\xcd -- that proprietary \u{} syntax doesn't help at all.

"\u{ABCD}" == "\xEA\xAF\x8D", NOT "\xAB\xCD".
Also "\u{0}" == "\0" and "\u{10301}" == "\xF0\x90\x8C\x81". It's variable-length so you don't need to have to handle \u and \U separately, you can just handle both.

The point is to be able to enter or cut'n'paste a hex code point literally, not having to convert it to UTF-8 by hand.

Exactly. Lua 5.3 also removes the need to add padding 😉

from luajit.

Rust also uses the "\u{ABCD}"-style escapes. These have the advantages mentioned above.

I am going to try this.

from luajit.

Thanks! Applied to my git master (with modifications).

from luajit.