Allow underscores _ as digit separates in float and integer literals.

[momumi]

Overview

Allow the use of underscore _ in float and integer literals as a visual separator to enhance readability of long number literals. For example:

const n = 1_000_000;
const c = 299_792_458;
const pi = 3.1415_9265_3589_7932;

const x = 0xDEAD_BEEF;
const b = 0b1110_0000;

This is a common feature of several other languages and can make code much easier to read.

Modifications to grammar.

Underscores could be freely used inside number literals except:

• _ can't be the first or last character of a number literal
• _ can't appear next to .-+ep in float literals.

So the grammar for integer and float literals would become:

hex <- [0-9a-fA-F]
hex_ <- (_ / hex)
dec <- [0-9]
dec_ <- (_ / dec)

dec_int <- dec (dec_* dec)?
hex_int <- hex (hex_* hex)?

INTEGER
    <- "0b" [_01]* [01]  skip
     / "0o" [_0-7]* [0-7] skip
     / "0x" hex_* hex  skip
     / dec_int skip

FLOAT
    <- "0x" hex_* hex "." hex_int ([pP] [-+]? hex_it )? skip
     /      dec_int "." dec_int ([eE] [-+]? dec_int)? skip
     / "0x" hex_* hex "."? [pP] [-+]? hex_int skip
     /      dec_int "."? [eE] [-+]? dec_int skip

Examples of valid literals

1_000
1_0_0_0
1__0____0

0b___1
0b__10
0b_100

1_000.0
1e1_000
1_000e1_000

0x_abc
0xabc_def.123

Examples of invalid literals

_100
100_

0x_
0x0_
_0x0
0_x0

_._e_
1._e10
_.1e10
1.1e_

1_.0e10
1._0e10
1.0_e10
1.1e_1
1.1e-_1

1_e6
1e6_
_1e6

[emekoi]

see #504.

[momumi]

Ah, sorry for the duplicate I'm sure I searched for that