generate numeric unicode character table

index.js

(async () => {
  const [unicodeDataTxt, scriptsTxt] = await Promise.all([
    fetch('https://www.unicode.org/Public/UCD/latest/ucd/UnicodeData.txt').then(res => res.text()),
    fetch('https://www.unicode.org/Public/UCD/latest/ucd/Scripts.txt').then(res => res.text()),
  ]);

  const scriptMap = [];

  // Parse Scripts.txt to map codepoint ranges → script
  for (const line of scriptsTxt.split('\n')) {
    const cleaned = line.replace(/#.*/, '').trim();
    if (!cleaned) continue;

    const [rangePart, scriptName] = cleaned.split(';').map(s => s.trim());
    let start, end;

    if (rangePart.includes('..')) {
      [start, end] = rangePart.split('..').map(cp => parseInt(cp, 16));
    } else {
      start = end = parseInt(rangePart, 16);
    }

    scriptMap.push({ start, end, script: scriptName });
  }

  // Function to find script for a codepoint
  const getScript = (codePoint) => {
    for (const { start, end, script } of scriptMap) {
      if (codePoint >= start && codePoint <= end) {
        return script;
      }
    }
    return 'Unknown';
  };

  const unicodeData = {};

  for (const line of unicodeDataTxt.split('\n')) {
    if (!line.trim()) continue;

    const fields = line.split(';');
    const codeHex = fields[0];
    const category = fields[2];
    const decimal = fields[6];
    const digit = fields[7];
    const numeric = fields[8];

    const codePoint = parseInt(codeHex, 16);
    const char = String.fromCodePoint(codePoint);

    const script = getScript(codePoint);

    const entry = {
      script: `'${script}'`,
      decimalDigit: category === 'Nd',
    };

    if (decimal !== '') entry.decimalValue = Number(decimal);
    if (numeric !== '') {
      entry.numeric = numeric.includes('/')
        ? (() => {
            const [n, d] = numeric.split('/').map(Number);
            return n / d;
          })()
        : Number(numeric);
    }

    // Only include useful entries
    if (
      entry.decimalDigit ||
      entry.decimalValue !== undefined ||
      entry.numeric !== undefined
    ) {
      unicodeData[char] = entry;
    }
  }

  // Output copy-paste-ready JS
  console.log('let unicodeData = {');
  for (const [char, data] of Object.entries(unicodeData)) {
    const key = `'${char}'`;
    const props = Object.entries(data)
      .map(([k, v]) => `${k}: ${v}`)
      .join(', ');
    console.log(`  ${key}: { ${props} },`);
  }
  console.log('};');
})();