clean bad Unicode

clean-bad-unicode.js

/*
date: 2022/8/7 23:34
*/
export default function (str) {
  const garbageSpaces = [
    '\u00A0', // No-Break Space
    '\u3000', // ideographic space
    '\u2000', // En quad space
    '\u2001', // Em quad space
    '\u2002', // En space
    '\u2003', // Em space
    '\u2004', // 3/em space
    '\u2005', // 4/em space
    '\u2006', // 6/em space
    '\u2007', // figure space
    '\u2008', // punctuation space
    '\u2009', // thin space
    '\u200A', // hair space
    '\u200B', // zero width space
    '\u202F', // narrow no break space
    '\u205F', // math space
    '\u303F', // ideographic half space
    '\uFEFF', // zero width No Break Space
    '\u{E0020}' // tag space
  ]

  const ASCIIVariants = [
    '\uFF01!',
    '\uFF02"',
    '\uFF03#',
    '\uFF04$',
    '\uFF05%',
    '\uFF06&',
    '\uFF07\'',
    '\uFF08(',
    '\uFF09)',
    '\uFF0A*',
    '\uFF0B+',
    '\uFF0C,',
    '\uFF0D-',
    '\uFF0E.',
    '\uFF0F/',
    '\uFF100',
    '\uFF111',
    '\uFF122',
    '\uFF133',
    '\uFF144',
    '\uFF155',
    '\uFF166',
    '\uFF177',
    '\uFF188',
    '\uFF199',
    '\uFF1A:',
    '\uFF1B;',
    '\uFF1C<',
    '\uFF1D=',
    '\uFF1E>',
    '\uFF1F?',
    '\uFF20@',
    '\uFF21A',
    '\uFF22B',
    '\uFF23C',
    '\uFF24D',
    '\uFF25E',
    '\uFF26F',
    '\uFF27G',
    '\uFF28H',
    '\uFF29I',
    '\uFF2AJ',
    '\uFF2BK',
    '\uFF2CL',
    '\uFF2DM',
    '\uFF2EN',
    '\uFF2FO',
    '\uFF30P',
    '\uFF31Q',
    '\uFF32R',
    '\uFF33S',
    '\uFF34T',
    '\uFF35U',
    '\uFF36V',
    '\uFF37W',
    '\uFF38X',
    '\uFF39Y',
    '\uFF3AZ',
    '\uFF3B[',
    '\uFF3C\\',
    '\uFF3D]',
    '\uFF3E^',
    '\uFF3F_',
    '\uFF40`',
    '\uFF41a',
    '\uFF42b',
    '\uFF43c',
    '\uFF44d',
    '\uFF45e',
    '\uFF46f',
    '\uFF47g',
    '\uFF48h',
    '\uFF49i',
    '\uFF4Aj',
    '\uFF4Bk',
    '\uFF4Cl',
    '\uFF4Dm',
    '\uFF4En',
    '\uFF4Fo',
    '\uFF50p',
    '\uFF51q',
    '\uFF52r',
    '\uFF53s',
    '\uFF54t',
    '\uFF55u',
    '\uFF56v',
    '\uFF57w',
    '\uFF58x',
    '\uFF59y',
    '\uFF5Az',
    '\uFF5B{',
    '\uFF5C|',
    '\uFF5D}',
    '\uFF5E~',
  ]
  garbageSpaces.forEach(s => { str = str.replaceAll(s, ' ') })
  ASCIIVariants.forEach(c => { str = str.replaceAll(c[0], c[1]) })
  return str
}