@mostra/cli 0.0.0-stage → 0.2.0

package/src/telemetry/zone-sets.ts

@@ -0,0 +1,182 @@
+/**
+ * Builds the ask-first time zones from tzdata (D140 item 10): every `zone.tab` name of the
+ * countries where Mostra asks before sending telemetry, every `backward` alias of those names, the
+ * legacy and UTC names, and the set of names tzdata knows at all. `scripts/ask-first-zones.ts`
+ * writes the result into `src/telemetry/zones.generated.ts`; a test checks the committed file
+ * against this function over the vendored tzdata files in `data/tzdata/`.
+ */
+
+/** The ISO 3166 codes of the countries and territories where Mostra asks first. */
+export const ASK_FIRST_COUNTRIES: readonly string[] = [
+  // The EU 27.
+  "AT",
+  "BE",
+  "BG",
+  "HR",
+  "CY",
+  "CZ",
+  "DK",
+  "EE",
+  "FI",
+  "FR",
+  "DE",
+  "GR",
+  "HU",
+  "IE",
+  "IT",
+  "LV",
+  "LT",
+  "LU",
+  "MT",
+  "NL",
+  "PL",
+  "PT",
+  "RO",
+  "SK",
+  "SI",
+  "ES",
+  "SE",
+  // The outermost regions with their own codes, and Åland.
+  "GP",
+  "MQ",
+  "GF",
+  "RE",
+  "YT",
+  "MF",
+  "AX",
+  // The EEA.
+  "IS",
+  "LI",
+  "NO",
+  // The UK, the Crown dependencies and Gibraltar.
+  "GB",
+  "GG",
+  "JE",
+  "IM",
+  "GI",
+  // Switzerland, the Faroe Islands, Svalbard and Jan Mayen.
+  "CH",
+  "FO",
+  "SJ",
+  // Inside the EU, some of them tzdata links to EU zones.
+  "AD",
+  "MC",
+  "SM",
+  "VA",
+];
+
+/** Names that are UTC, unknown or legacy European zones, asked first wherever they are. */
+const ASK_FIRST_NAMES: readonly string[] = [
+  "CET",
+  "EET",
+  "WET",
+  "MET",
+  "UTC",
+  "GMT",
+  "Universal",
+  "Zulu",
+  "Greenwich",
+  "Factory",
+];
+
+/** The ask-first zones and every zone name tzdata knows. */
+export interface ZoneSets {
+  readonly version: string;
+  readonly askFirst: readonly string[];
+  readonly known: readonly string[];
+}
+
+/**
+ * The `Link TARGET NAME` lines of a tzdata file, with the zone a `#= ZONE` comment names: the one
+ * the alias meant before tzdata merged it (`Iceland` links to `Africa/Abidjan`, `#=
+ * Atlantic/Reykjavik`).
+ */
+function links(text: string): { target: string; name: string; meant?: string }[] {
+  const found: { target: string; name: string; meant?: string }[] = [];
+  for (const line of text.split("\n")) {
+    const fields = line.replace(/#.*/, "").trim().split(/\s+/);
+    if (fields[0] !== "Link" || !fields[1] || !fields[2]) continue;
+    const meant = /#=\s*(\S+)/.exec(line)?.[1];
+    found.push({ target: fields[1], name: fields[2], ...(meant !== undefined ? { meant } : {}) });
+  }
+  return found;
+}
+
+/** The `Zone NAME …` names of a tzdata file. */
+function zones(text: string): string[] {
+  const found: string[] = [];
+  for (const line of text.split("\n")) {
+    const fields = line.replace(/#.*/, "").trim().split(/\s+/);
+    if (fields[0] === "Zone" && fields[1]) found.push(fields[1]);
+  }
+  return found;
+}
+
+/** Builds the sets from the tzdata files' texts. */
+export function buildZoneSets(files: {
+  readonly version: string;
+  readonly zoneTab: string;
+  readonly backward: string;
+  readonly etcetera: string;
+  readonly factory: string;
+}): ZoneSets {
+  const countries = new Set(ASK_FIRST_COUNTRIES);
+  const known = new Set<string>();
+  const askFirst = new Set<string>(ASK_FIRST_NAMES);
+  for (const line of files.zoneTab.split("\n")) {
+    if (line.startsWith("#") || line.trim() === "") continue;
+    const [code, , name] = line.split("\t");
+    if (!code || !name) continue;
+    known.add(name);
+    if (countries.has(code)) askFirst.add(name);
+  }
+  const etcetera = [...zones(files.etcetera), ...links(files.etcetera).map((link) => link.name)];
+  for (const name of etcetera) {
+    known.add(name);
+    // UTC and its offsets: Etc/* is asked first, whatever its offset (D140 item 10).
+    if (name.startsWith("Etc/") || name === "GMT") askFirst.add(name);
+  }
+  for (const name of zones(files.factory)) known.add(name);
+  // Aliases of an ask-first name are ask-first; matched as given, never by their target.
+  const backward = links(files.backward);
+  let grew = true;
+  while (grew) {
+    grew = false;
+    for (const { target, name, meant } of backward) {
+      known.add(name);
+      known.add(target);
+      const asks = askFirst.has(target) || (meant !== undefined && askFirst.has(meant));
+      if (asks && !askFirst.has(name)) {
+        askFirst.add(name);
+        grew = true;
+      }
+    }
+  }
+  for (const name of ASK_FIRST_NAMES) known.add(name);
+  return {
+    version: files.version.trim(),
+    askFirst: [...askFirst].sort(),
+    known: [...known].sort(),
+  };
+}
+
+/** The generated module's text. */
+export function zonesModule(sets: ZoneSets): string {
+  const list = (names: readonly string[]): string =>
+    names.map((name) => `  ${JSON.stringify(name)},`).join("\n");
+  return `// Generated by scripts/ask-first-zones.ts from tzdata ${sets.version} (data/tzdata/). Don't edit.
+
+/** The tzdata release the zones come from (VERIFIED_FACTS, WP-10). */
+export const TZDATA_VERSION = ${JSON.stringify(sets.version)};
+
+/** The time zones where Mostra asks before sending telemetry (D140 item 10), matched as given. */
+export const ASK_FIRST_ZONES: readonly string[] = [
+${list(sets.askFirst)}
+];
+
+/** Every zone name tzdata ${sets.version} knows; anything else is an unknown zone. */
+export const KNOWN_ZONES: readonly string[] = [
+${list(sets.known)}
+];
+`;
+}