CLDR Data subset

CLDRGenerator.java

import java.io.IOException;
import java.util.HashSet;
import java.util.Map;
import java.util.Map.Entry;
import java.util.Set;
import java.util.function.Function;
import java.util.stream.Collectors;

import com.google.gson.JsonElement;
import com.google.gson.JsonObject;

/**
 * (c) Swissquote 02.05.17
 *
 * @author sgoetz
 */
public class CLDRGenerator {

	private final CldrFileGetter cldrFileGetter;

	public CLDRGenerator() throws IOException {
		cldrFileGetter = new CldrFileGetter();
	}

	public JsonObject getCLDRBaseData() throws IOException {

		JsonObject results = new JsonObject();

		// Modules: currency, date, message, number, plural, relativeTime, unit
		// Disabled; we'll get only the likelySubtags that might be needed with the current locale
		//results = mergeRecursive(results, getCLDRFile("supplemental/likelySubtags.json"));

		// Modules: currency, date, number, relativeTime, unit
		// Disabled; we'll get only the numberingSystems used in the current locale
		//results = mergeRecursive(results, getCLDRFile("supplemental/numberingSystems.json"));

		// Modules: currency, plural, relativeTime, unit
		// Disabled; we'll get only the plurals needed for the current locale
		//results = mergeRecursive(results, getCLDRFile("supplemental/plurals.json"));
		//results = mergeRecursive(results, getCLDRFile("supplemental/ordinals.json"));

		// Modules: currency
		results = mergeRecursive(results, getCLDRFile("supplemental/currencyData.json"));
		results.getAsJsonObject("supplemental").getAsJsonObject("currencyData").remove("region");

		// Modules: date
		results = mergeRecursive(results, getCLDRFile("supplemental/timeData.json"));
		results = mergeRecursive(results, getCLDRFile("supplemental/weekData.json"));

		return results;
	}

	public JsonObject getCLDRLocaleData(String locale) throws IOException {

		String cldrLocale = convertToCDRLocale(locale);

		//Modules: currency, date, number, relativeTime, unit
		JsonObject results = getCLDRFile("main/" + cldrLocale + "/numbers.json");

		// Get only the numbering systems actually used in this locale
		// Modules: currency, date, number, relativeTime, unit
		results = mergeRecursive(results, getNumberingSystems(results, cldrLocale));

		// Get only the locales that are needed
		// Modules: currency, plural, relativeTime, unit
		results = mergeRecursive(results, getPlurals(cldrLocale));

		// Get only the likelySubtags that might be needed
		// Modules: currency, date, message, number, plural, relativeTime, unit
		results = mergeRecursive(results, getLikelySubtags(cldrLocale));

		//Modules: currency
		//results = mergeRecursive(results, getCLDRFile("main/" + cldrLocale + "/currencies.json"));

		//Modules: date
		results = mergeRecursive(results, getCLDRFile("main/" + cldrLocale + "/ca-gregorian.json"));
		results = mergeRecursive(results, getCLDRFile("main/" + cldrLocale + "/timeZoneNames.json"));

		//Modules: relativeTime
		results = mergeRecursive(results, getCLDRFile("main/" + cldrLocale + "/dateFields.json"));

		//Modules: unit
		// We don't load this module on the front
		//results = mergeRecursive(results, getCLDRFile("main/" + cldrLocale + "/units.json"));

		return results;
	}

	void addValid(JsonObject source, JsonObject destination, final Set<String> validkeys) {
		addValid(source, destination, (Entry<String, JsonElement> entry) -> validkeys.contains(entry.getKey()));
	}

	void addValid(JsonObject source, JsonObject destination, Function<Entry<String, JsonElement>, Boolean> callback) {
		for (Map.Entry<String, JsonElement> entry : source.entrySet()) {
			if (callback.apply(entry)) {
				destination.add(entry.getKey(), entry.getValue());
			}
		}
	}

	JsonObject getLikelySubtags(String cldrLocale) throws IOException {

		Set<String> validLocales = getValidLocales(cldrLocale).stream()
				.map(item -> item.toLowerCase())
				.collect(Collectors.toSet());

		JsonObject supplemental = new JsonObject();
		JsonObject likelySubtagsUsed = new JsonObject();
		supplemental.add("likelySubtags", likelySubtagsUsed);

		JsonObject likelySubtags = getCLDRFile("supplemental/likelySubtags.json")
				.getAsJsonObject("supplemental")
				.getAsJsonObject("likelySubtags");

		addValid(likelySubtags, likelySubtagsUsed,
				entry -> validLocales.stream().anyMatch(item -> entry.getKey().toLowerCase().indexOf(item) > -1));

		JsonObject root = new JsonObject();
		root.add("supplemental", supplemental);

		return root;
	}

	Set<String> getValidLocales(String cldrLocale) {
		Set<String> validLocales = new HashSet<>();
		validLocales.add(cldrLocale);

		String moreLikelyToBeATag = cldrLocale.replaceAll("_", "-");

		// If the language tag was generated with "Locale.toString" chances are we have this wrong
		validLocales.add(moreLikelyToBeATag);

		// recursive bcp-47 fallback using a less specific tag, for example if
		// we only have values for "en" and "fr" but "fr-FR" was requested.
		// it is generally preferable to use "fr"
		String primaryLanguage = moreLikelyToBeATag;
		while (primaryLanguage.contains("-")) {
			primaryLanguage = primaryLanguage.substring(0, primaryLanguage.lastIndexOf('-'));
			validLocales.add(primaryLanguage);
		}

		return validLocales;
	}

	/**
	 * Pluralization rules are stored in a common place,
	 * but we only need the ones for the current locale.
	 */
	JsonObject getPlurals(String cldrLocale) throws IOException {

		Set<String> validLocales = getValidLocales(cldrLocale);

		JsonObject supplemental = new JsonObject();
		JsonObject cardinalUsed = new JsonObject();
		supplemental.add("plurals-type-cardinal", cardinalUsed);
		JsonObject ordinalUsed = new JsonObject();
		supplemental.add("plurals-type-ordinal", ordinalUsed);

		JsonObject cardinal = getCLDRFile("supplemental/plurals.json")
				.getAsJsonObject("supplemental")
				.getAsJsonObject("plurals-type-cardinal");

		addValid(cardinal, cardinalUsed, validLocales);

		JsonObject ordinal = getCLDRFile("supplemental/ordinals.json")
				.getAsJsonObject("supplemental")
				.getAsJsonObject("plurals-type-ordinal");

		addValid(ordinal, ordinalUsed, validLocales);

		JsonObject root = new JsonObject();
		root.add("supplemental", supplemental);

		return root;
	}

	/**
	 * Instead of adding all numbering systems, we only
	 * load the ones that are declared in the current locale
	 */
	JsonObject getNumberingSystems(JsonObject results, String cldrLocale) throws IOException {
		JsonObject numbers = results.getAsJsonObject("main").getAsJsonObject(cldrLocale).getAsJsonObject("numbers");
		String defaultNumberingSystem = numbers.getAsJsonPrimitive("defaultNumberingSystem").getAsString();

		Set<String> set = new HashSet<>();
		set.add(defaultNumberingSystem);

		JsonObject otherNumberingSystems = numbers.getAsJsonObject("otherNumberingSystems");

		for (Entry<String, JsonElement> entry : otherNumberingSystems.entrySet()) {
			set.add(entry.getValue().getAsString());
		}

		JsonObject allNumberingSystems = getCLDRFile("supplemental/numberingSystems.json")
				.getAsJsonObject("supplemental")
				.getAsJsonObject("numberingSystems");

		JsonObject usedNumberingSystems = new JsonObject();

		addValid(allNumberingSystems, usedNumberingSystems, set);

		JsonObject supplemental = new JsonObject();
		JsonObject root = new JsonObject();

		supplemental.add("numberingSystems", usedNumberingSystems);
		root.add("supplemental", supplemental);

		return root;
	}

	String convertToCDRLocale(String locale) {
		switch (locale.toLowerCase()) {
			case "cn":
			case "zh-cn":
			case "zh_cn":
				return "zh";
			case "tw":
			case "zh-tw":
			case "zh_tw":
				return "zh-Hant";
			default:
				return locale;
		}
	}

	JsonObject getCLDRFile(String file) throws IOException {
		return cldrFileGetter.getFile(file).getAsJsonObject();
	}
}