UnicodeData.txt to JSON converter

unicodedata2json.js

// Script runs on the browser console of the webpage: http://unicode.org/Public/UNIDATA/UnicodeData.txt

const entries = document.querySelector("body > pre").innerText.split('\n');
entries.pop();

let decompTags = ['<font>','<noBreak>','<initial>','<medial>','<final>','<isolated>','<circle>','<super>','<sub>','<vertical>','<wide>','<narrow>','<small>','<square>','<fraction>','<compat>']
function parseDecompMapping(mapping) {
    if (!mapping) return null;
  
    mapping = mapping.split(' ');
    let mapobj = { tag: null, characters: [] };

    if (decompTags.includes(mapping[0])) {
        mapobj.tag = mapping[0];
        mapping.shift();
    }
    mapping.forEach(charcode => {
        mapobj.characters.push(Number('0x' + charcode));
    });
    return mapobj;
}

let chars = [];
entries.forEach(entry => {
  let fields = entry.split(';');
  let char = {};

  char.code = Number('0x' + fields[0]);
  char.name = fields[1];
  char.category = fields[2];
  char.canonicalCombiningClass = Number('0x' + fields[3]);
  char.bidirectionalCategory = fields[4];
  char.decompositionMapping = parseDecompMapping(fields[5]) || null;
  char.decimalDigitValue = Number('0x' + fields[6]) || null;
  char.digitValue = Number('0x' + fields[7]) || null;
  char.numericValue = Number('0x' + fields[8]) || null;
  char.mirrored = fields[9] === 'Y';
  char.oldName = fields[10] || null;
  char.comment = fields[11] || null;
  char.uppercaseMapping = Number('0x' + fields[12]) || null;
  char.lowercaseMapping = Number('0x' + fields[13]) || null;
  char.titlecaseMapping = Number('0x' + fields[14]) || null;
  
  chars.push(char);
});

// Output is now available as a JS object in the "chars" variable.
// Pretty-print the JSON for copy
const unicodejson = JSON.stringify(chars, null, 2);
console.info(unicodejson);