Generate unicode forms

index.js

(async () => {
  const url = 'https://www.unicode.org/Public/UCD/latest/ucd/UnicodeData.txt';
  const res = await fetch(url);
  const text = await res.text();

  const lines = text.split('\n');
  const digitEntries = [];

  for (const line of lines) {
    if (!line.trim()) continue;
    const fields = line.split(';');
    const code = parseInt(fields[0], 16);
    const category = fields[2];
    const decimal = fields[6];

    if (category === 'Nd' && decimal !== '') {
      digitEntries.push({ code, decimal: Number(decimal), name: fields[1] });
    }
  }

  // Sort and group into 10-digit ranges with decimals 0–9
  const ranges = [];
  let group = [];

  for (let i = 0; i < digitEntries.length; i++) {
    const curr = digitEntries[i];
    group.push(curr);

    if (group.length === 10) {
      const validDecimals = group.map(e => e.decimal).sort((a, b) => a - b).join(',');
      if (validDecimals === '0,1,2,3,4,5,6,7,8,9') {
        ranges.push([...group]);
      }
      group = [];
    } else if (i < digitEntries.length - 1) {
      const next = digitEntries[i + 1];
      if (next.code !== curr.code + 1) group = [];
    }
  }

  // Generate the getForm() function
  console.log('function getForm(codePoint) {');
  for (let i = 0; i < ranges.length; i++) {
    const group = ranges[i];
    const start = group[0].code;
    const end = group[9].code;
    const label = group[0].name.toLowerCase().replace(/[^a-z]/g, '_').replace(/_+/g, '_').replace(/^_|_$/g, '');

    console.log(`  if (codePoint >= 0x${start.toString(16).toUpperCase()} && codePoint <= 0x${end.toString(16).toUpperCase()}) return '${label}';`);
  }
  console.log("  return 'other';");
  console.log('}');
})();