unaccent_rules_builder.js

/**
 * Generates unaccent.rules for Postgres.
 * Rule set will convert bold/italic UTF8 characters to normal characters.
 * @type {number[]}
 */

/**
 * !IMPORTANT!
 * the ℎ letter (3rd one) is not correctly recognized, needs to be manually updated
 */

// characters from https://yaytext.com/bold-italic/
// and https://www.obliquity.com/computer/html/unicode1D400.html
const diffsUpper = ['𝐀', '𝗔', '𝐴', '𝘈', '𝑨', '𝘼', '𝒜', '𝙰'].map((ch) => ch.codePointAt(0) - 'A'.codePointAt(0));
const diffsLower = ['𝐚', '𝗮', '𝑎', '𝘢', '𝒂', '𝙖', '𝒶', '𝚊'].map((ch) => ch.codePointAt(0) - 'a'.codePointAt(0));

const generateRule = (ch1, ch2) => `${ch1}\t${ch2}`;

const generateAlphabet = (from, to) => {
  const alphabet = [];
  const start = from.charCodeAt(0);
  const last  = to.charCodeAt(0);
  for (var i = start; i <= last; ++i) {
    alphabet.push(String.fromCharCode(i));
  }

  return alphabet;
};

const generateRuleSet = (alphabet, diffs) => {
  const result = [];
  alphabet.forEach(ch => {
    diffs.forEach(diff => {
      result.push(generateRule(String.fromCodePoint(ch.codePointAt(0) + diff), ch))
    });
  });
  return result;
};

const rules = generateRuleSet(generateAlphabet('A', 'Z'), diffsUpper)
  .concat(generateRuleSet(generateAlphabet('a', 'z'), diffsLower));
console.log(rules.join("\n"));