swc_ecma_minifier support charset option

[ahaoboy]

Describe the feature

For nerd fonts characters, convert them to Unicode by default. It is expected that there will be an option similar to esbuild charset

const a = "󰋚";
console.log("a", a);

console.log("a", "\uDB80\uDEDA");

esbuild

esbuild ./a.js --bundle  --charset=utf8 

var a = "󰋚";
console.log("a", a);

Babel plugin or link to the feature description

No response

Additional context

web-infra-dev/rspack#7734

https://play.swc.rs/?version=1.7.21&code=H4sIAAAAAAAAA0vOzyvOz0nVy8lP11BKVNJRUPq8oXsWiH62tfvF%2BqlKmtYA2egJaCMAAAA%3D&config=H4sIAAAAAAAAA42VzW7bMAyA732KwOcetsOAYg%2FQ1xAUiXKUyaIg0m6Mou8%2B2nHSZKaNXRybH38k%2FuXz5XBozuSa34dPeZWPYitBvX%2BLhMbM9iKSBlxnydVYuHm90TNNKNhEMIu%2BrqRhW1vg2Yp%2BLepNQiS4qS%2ByLuYYxseAnc1tggfJ5A5LggHSs%2FGM%2FgAU45IlyrYD2tIIu7TUOFgG%2BcWiKUV4U6Rkg63x54%2BnDNxzIAoOu1KB6Pkykpq%2Bg8xaIFsrfkyAa%2F8oPyImsHmHGEsmZoYWqubYYUq2EJjBVsXLdFK5C6EWYoI9g7%2BnZ8WzjxwxS8w19WC9cehBQbGC4ziAZiaxxCwTJlDuM2MPx75t5379xxoGm3qpqEIuc0nktIrXE0ZiE%2FqspfAKN3JwhUtyV%2B0TTAXua17bnTHmjZr8b2dLP21Z73d9zEGmj0eFy6hqt8zQTmMSY1AyO2UGKketmhV872DKrNOOs%2BCN9FH0YCAE6RXFNX1EdictKI8FMChA6muD1lVXYO5TuMGngdjB73JL1hts0egsn7Ypjd0R006ADviEfkdBSsG4jatsiUvZ5n32IK0BXlXpaQbrJSADwGjSvPdXvSHjIR5Nm%2FD4vSZuG%2FPl9px35%2Fefwqzz9Rc%2BNnjbpwYAAA%3D%3D

[CPunisher]

This could be expected with default target Es5.

swc/crates/swc_ecma_codegen/src/tests.rs

Line 635 in 62ed065

es5("ퟻ", "\"\\uD7FB\"");

[ahaoboy]

This could be expected with default target Es5.

The expected behavior should be to display the characters directly instead of Unicode encoding, because in some engines, multi-byte encoding cannot be handled and only characters can be used. And the strange thing is that only some characters seem to be converted to unicode codes. These three are characters from the same font.

export const History = "󰋚"
export const Heart = ""
export const Random = ""

[ahaoboy]

utf8 seems to take up less file size than unicode

const fs = require('fs')
// const a = "🌏";  // 🌏 or 
const a = "";  // 🌏 or 
const encoder = new TextEncoder();

function getUnicodeEncoding(char) {
  const unicodeArray = [];
  for (let i = 0; i < char.length; i++) {
    const codePoint = char.charCodeAt(i).toString(16).toUpperCase();
    unicodeArray.push(`\\u${codePoint.padStart(4, "0")}`);
  }
  return unicodeArray.join('');
}

const n = 1024
fs.writeFileSync("./utf8.js", `"${a.repeat(n)}"`)
fs.writeFileSync("./unicode.js", `"${getUnicodeEncoding(a).repeat(n)}"`)

Both single-byte and double-byte characters in utf8 encoding take up less space than unicode

-rw-r--r-- 1 ace None 6.1K Dec 26 20:54 unicode.js
-rw-r--r-- 1 ace None 3.1K Dec 26 20:54 utf8.js

-rw-r--r-- 1 ace None  13K Dec 26 20:54 unicode.js
-rw-r--r-- 1 ace None 4.1K Dec 26 20:54 utf8.js