si2 role for myst

si2.mjs

const UNITS = {
  // SI base units
  ampere: 'A',
  candela: 'cd',
  kelvin: 'K',
  kilogram: 'kg',
  metre: 'm',
  meter: 'm',
  mole: 'mol',
  second: 's',
  // SI derived units
  becquerel: 'Bq',
  degreeCelsius: '°C',
  coulomb: 'C',
  farad: 'F',
  gray: 'Gy',
  hertz: 'Hz',
  henry: 'H',
  joule: 'J',
  lumen: 'lm',
  katal: 'kat',
  lux: 'lx',
  newton: 'N',
  ohm: 'Ω',
  pascal: 'Pa',
  radian: 'rad',
  siemens: 'S',
  sievert: 'Sv',
  steradian: 'sr',
  tesla: 'T',
  volt: 'V',
  watt: 'W',
  weber: 'Wb',
  // Non-SI units accepted for use with SI
  astronomicalunit: 'au',
  bel: 'B',
  dalton: 'Da',
  day: 'd',
  decibel: 'dB',
  degree: '°',
  electronvolt: 'eV',
  hectare: 'ha',
  hour: 'h',
  litre: 'L',
  liter: 'L',
  arcminute: '′', // minute (plane angle) U+2032
  minute: 'min', // minute (time)
  arcsecond: '″', // second (plane angle) U+2033
  neper: 'Np',
  tonne: 't',
  // SI prefixes
  yocto: 'y', // -24
  zepto: 'z', // -21
  atto: 'a', // -18
  femto: 'f', // -15
  pico: 'p', // -12
  nano: 'n', // -9
  micro: 'µ', // -6
  milli: 'm', // -3
  centi: 'c', // -2
  deci: 'd', // -1
  deca: 'da', // 1
  hecto: 'h', // 2
  kilo: 'k', // 3
  mega: 'M', // 6
  giga: 'G', // 9
  tera: 'T', // 12
  peta: 'P', // 15
  exa: 'E', // 18
  zetta: 'Z', // 21
  yotta: 'Y', // 24
  // Special
  angstrom: 'Å',
};

/** Format a number string as LaTeX: commas → {,}, eN → \times 10^{N} */
function formatNumber(num) {
  if (!num) return '';
  const eMatch = num.match(/^(-?[\d,.]+)e(\d+)$/);
  if (eMatch) {
    const base = eMatch[1].replace(/,/g, '{,}');
    return `${base} \\times 10^{${eMatch[2]}}`;
  }
  return num.replace(/,/g, '{,}');
}

const PREFIXES = [
  'yocto', 'zepto', 'atto', 'femto', 'pico', 'nano', 'micro', 'milli',
  'centi', 'deci', 'deca', 'hecto', 'kilo', 'mega', 'giga', 'tera',
  'peta', 'exa', 'zetta', 'yotta',
];

/** Resolve a unit name to its symbol, handling combined prefix+unit names like 'millimeter' */
function resolveSymbol(name) {
  if (name in UNITS) return UNITS[name];
  for (const prefix of PREFIXES) {
    if (name.startsWith(prefix)) {
      const remainder = name.slice(prefix.length);
      if (remainder in UNITS) {
        return (UNITS[prefix] ?? prefix) + (UNITS[remainder] ?? remainder);
      }
    }
  }
  return name;
}

/** Convert a unit name to a \mathrm{} LaTeX token */
function unitToLatex(name) {
  return `\\mathrm{${resolveSymbol(name)}}`;
}

const si2Role = {
  name: 'si2',
  doc: 'Format a number with SI units, e.g. {si2}`1.5e3 <\\kilo\\hertz>`',
  body: {
    type: String,
    required: true,
  },
  run(data) {
    const value = data.body;
    // Supports: optional sign, comma-grouped integer, optional decimal, optional exponent
    const match = value.match(
      /^(-?(\d+(,\d+)*(\.\d+)?)?(e\d+)?)\s?<([\\a-zA-Z\s]+)>$/,
    );
    if (!match) {
      return [{ type: 'inlineMath', value }];
    }
    const number = match[1];
    const commands = match[match.length - 1];
    const unitNames = [...commands.matchAll(/\\([a-zA-Z]+)/g)].map(([, c]) => c);

    // Split on \per: units before → numerator, units after → denominator (^{-1})
    const perIdx = unitNames.indexOf('per');
    const numerator = perIdx === -1 ? unitNames : unitNames.slice(0, perIdx);
    const denominator = perIdx === -1 ? [] : unitNames.slice(perIdx + 1);

    const numLatex = numerator.map(unitToLatex).join('\\,');
    const denLatex = denominator.map((u) => `${unitToLatex(u)}^{-1}`).join('\\,');
    const unitLatex = [numLatex, denLatex].filter(Boolean).join('\\,');

    const latex = `${formatNumber(number)}\\,${unitLatex}`;

    return [{ type: 'inlineMath', value: latex }];
  },
};

const plugin = { name: 'SI Units', roles: [si2Role] };

export default plugin;