tinycast/Scripts/gen-countries.js

#!/usr/bin/env node
// Generate Tinycast/Features/Calculator/Model/CountryZoneData.generated.swift.
//
// Usage: node Scripts/gen-countries.js [zone.tab cldr-territories.json]
// Downloads the sources when paths aren't given. Run occasionally, commit the output.
//
// Two sources, joined on the ISO 3166 code:
//   - IANA's zone.tab decides *which* clock a country answers with. It lists every zone per country,
//     most populous first where geography allows, so its first row is the country's main clock.
//     Foundation carries the zones themselves but no notion of which country they belong to.
//   - CLDR's `en` territory names decide what humans *call* the country, including the short forms
//     ("UK", "US"). Read from cldr-json rather than the host's `Intl`, whose output shifts with ICU.
//
// A spelling two countries share is left out rather than guessed.
"use strict";

const fs = require("fs");
const path = require("path");

const ZONE_TAB = "https://data.iana.org/time-zones/tzdb/zone.tab";
const CLDR_TERRITORIES =
  "https://raw.githubusercontent.com/unicode-org/cldr-json/main/cldr-json/cldr-localenames-full/main/en/territories.json";

// zone.tab orders by geography before population, which puts these countries' remote edges first.
// Each answers with its capital's clock instead; the script fails if one stops being listed.
const CAPITAL_ZONES = {
  AU: "Australia/Sydney",
  BR: "America/Sao_Paulo",
  CA: "America/Toronto",
  FM: "Pacific/Pohnpei",
  RU: "Europe/Moscow",
  UA: "Europe/Kyiv",
  UZ: "Asia/Tashkent",
};
// Territories with no capital and clocks a hemisphere apart name no single time at all.
const NO_SINGLE_CLOCK = new Set(["AQ", "UM"]);

// zone.tab lists ~250 codes; well below that is a bad response.
const MIN_EXPECTED = 230;

async function loadText(url, argPath) {
  if (argPath) return fs.readFileSync(argPath, "utf8");
  const response = await fetch(url);
  if (!response.ok) throw new Error(`${url} -> HTTP ${response.status}`);
  return response.text();
}

/// Every lookup form of a name: folded and lowercased as the query is, plus the spellings a keyboard
/// produces for CLDR's typographic `&` and `’`.
function spellings(name) {
  const base = name
    .normalize("NFD")
    .replace(/[̀-ͯ]/g, "")
    .toLowerCase()
    .replace(/\s+/g, " ")
    .trim();
  const forms = new Set([base, base.replace(/’/g, "'")]);
  for (const form of [...forms]) if (form.includes(" & ")) forms.add(form.replace(/ & /g, " and "));
  return forms;
}

function claim(map, key, code) {
  if (!map.has(key)) map.set(key, new Set());
  map.get(key).add(code);
}

function swiftString(value) {
  if (value.includes('"') || value.includes("\\")) throw new Error(`unsafe literal ${JSON.stringify(value)}`);
  return `"${value}"`;
}

/// Country code -> zones, in zone.tab's own order.
function zonesByCountry(text) {
  const zones = new Map();
  for (const line of text.split("\n")) {
    if (!line || line.startsWith("#")) continue;
    const [code, , zone] = line.split("\t");
    if (!/^[A-Z]{2}$/.test(code) || !zone) throw new Error(`unexpected zone.tab row: ${line}`);
    if (!zones.has(code)) zones.set(code, []);
    zones.get(code).push(zone);
  }
  return zones;
}

function mainClock(code, zones) {
  const capital = CAPITAL_ZONES[code];
  if (!capital) return zones[0];
  if (!zones.includes(capital)) throw new Error(`${capital} is no longer listed for ${code}`);
  return capital;
}

async function main() {
  const zones = zonesByCountry(await loadText(ZONE_TAB, process.argv[2]));
  const cldr = JSON.parse(await loadText(CLDR_TERRITORIES, process.argv[3]));
  const names = cldr?.main?.en?.localeDisplayNames?.territories;
  if (!names) throw new Error("unexpected CLDR shape: main.en.localeDisplayNames.territories missing");
  if (zones.size < MIN_EXPECTED) throw new Error(`suspiciously few countries: ${zones.size}`);

  const claims = new Map();
  const unnamed = [];
  for (const [code, list] of zones) {
    if (NO_SINGLE_CLOCK.has(code)) continue;
    // `GB`, `GB-alt-short` and `GB-alt-variant` are all names for the one country.
    const keys = Object.keys(names).filter((key) => key === code || key.startsWith(`${code}-alt-`));
    if (keys.length === 0) unnamed.push(code);
    const zone = mainClock(code, list);
    for (const key of keys) for (const form of spellings(names[key])) claim(claims, form, zone);
  }

  const table = [...claims]
    .filter(([, owners]) => owners.size === 1)
    .map(([name, owners]) => [name, [...owners][0]])
    .sort((a, b) => a[0].localeCompare(b[0]));
  const contested = claims.size - table.length;

  const out = path.resolve(
    __dirname,
    "..",
    "Tinycast/Features/Calculator/Model/CountryZoneData.generated.swift",
  );
  fs.writeFileSync(
    out,
    "// Generated by Scripts/gen-countries.js — do not edit by hand.\n" +
      "// Clocks from IANA zone.tab; country names from Unicode CLDR (en).\n" +
      "enum CountryZoneData {\n" +
      "    /// Folded, lowercased country name → the IANA zone of its main clock.\n" +
      "    static let zones: [String: String] = [\n" +
      table.map(([name, zone]) => `        ${swiftString(name)}: ${swiftString(zone)},`).join("\n") +
      "\n    ]\n" +
      "}\n",
  );
  console.log(
    `wrote ${out} — ${table.length} names for ${zones.size - NO_SINGLE_CLOCK.size} countries ` +
      `(${contested} contested, unnamed: ${unnamed.join(", ") || "none"})`,
  );
}

main().catch((e) => {
  console.error(e);
  process.exit(1);
});