prepare-subtag-registry.js

import {
	promises
} from "fs";

const {
	readFile,
	writeFile
} = promises;

// https://www.iana.org/assignments/language-subtag-registry/language-subtag-registry
const registryText = await readFile("./registry.txt", "utf8");

const regex = /%%\n(Type): (.+?)\n(Subtag|Tag): (.+?)(?:\n(Description): ((?:.+?)(?:\n {2}(?:.+?))*)(?:\n\5: (?:(?:.+?)(?:\n {2}(?:.+?))*))*)*(?:\n(Added|Suppress-Script|Deprecated|Preferred-Value|Comments): (?:(?:.+?)(?:\n {2}(?:.+?))*)(?:\n\7: (?:(?:.+?)(?:\n {2}(?:.+?))*))*)*(?:\n(Prefix): ((?:.+?)(?:\n {2}(?:.+?))*)(?:\n\8: (?:(?:.+?)(?:\n {2}(?:.+?))*))*)*(?:\n\7: (?:(?:.+?)(?:\n {2}(?:.+?))*)(?:\n\7: (?:(?:.+?)(?:\n {2}(?:.+?))*))*)*(?:\n(Macrolanguage): ((?:.+?)(?:\n {2}(?:.+?))*)(?:\n\10: (?:(?:.+?)(?:\n {2}(?:.+?))*))*)*(?:\n\7: (?:(?:.+?)(?:\n {2}(?:.+?))*)(?:\n\7: (?:(?:.+?)(?:\n {2}(?:.+?))*))*)*(?:\n(Scope): ((?:.+?)(?:\n {2}(?:.+?))*)(?:\n\12: (?:(?:.+?)(?:\n {2}(?:.+?))*))*)*(?:\n(?:Added|Suppress-Script|Deprecated|Preferred-Value|Comments): (?:(?:.+?)(?:\n {2}(?:.+?))*)(?:\n\7: (?:(?:.+?)(?:\n {2}(?:.+?))*))*)*(?=\n%%|$)/g;

const replaceString = "\"$4\":{\"type\":\"$2\",\"tagType\":\"$3\",\"description\":\"$6\",\"prefix\":\"$9\",\"macrolanguage\":\"$11\",\"scope\":\"$13\"},";

const processed = registryText
	.replace(/^(.+?)\n/, "")
	.replace(/"/g, "\\\"")
	.replace(regex, replaceString)
	.replace(/\n| {2,}/g, "")
	.slice(0, -1);

const json = `{${processed}}`;

const registry = JSON.parse(json);

for (const tag in registry) {
	for (const key in registry[tag]) {
		if (registry[tag][key] === "") {
			delete registry[tag][key];
		}
	}
}

await writeFile("./registry.json", JSON.stringify(registry));