kjelsrud.dev/node_modules/micromark-util-decode-numeric-character-reference/index.js
2023-07-19 21:31:30 +02:00

36 lines
1.1 KiB
JavaScript

/**
* Turn the number (in string form as either hexa- or plain decimal) coming from
* a numeric character reference into a character.
*
* Sort of like `String.fromCharCode(Number.parseInt(value, base))`, but makes
* non-characters and control characters safe.
*
* @param {string} value
* Value to decode.
* @param {number} base
* Numeric base.
* @returns {string}
* Character.
*/
export function decodeNumericCharacterReference(value, base) {
const code = Number.parseInt(value, base)
if (
// C0 except for HT, LF, FF, CR, space.
code < 9 ||
code === 11 ||
(code > 13 && code < 32) ||
// Control character (DEL) of C0, and C1 controls.
(code > 126 && code < 160) ||
// Lone high surrogates and low surrogates.
(code > 55295 && code < 57344) ||
// Noncharacters.
(code > 64975 && code < 65008) /* eslint-disable no-bitwise */ ||
(code & 65535) === 65535 ||
(code & 65535) === 65534 /* eslint-enable no-bitwise */ ||
// Out of range
code > 1114111
) {
return '\uFFFD'
}
return String.fromCharCode(code)
}