fix: incorrect display for zh-* culture names in lang switcher (#172)

Also: Remove uncultured language names from language switcher when a
cultured variant is found

Closes: #171
Closes: #157

CHANGELOG.md

@@ -8,8 +8,11 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 ## [unreleased]
 
 ### Fixed
+- Incorrect language display names for some locales (#171)
 
 ### Changed
+- Uncultured language variants (e.g. `en`) are not displayed in the language switcher when cultured variants (e.g. 
+  `en-US`) are registered. (#171)
 
 ### Removed
 

Editor/UI/LanguageSwitcher.cs

@@ -35,18 +35,10 @@ public static void DrawImmediate()
             var curLang = LanguagePrefs.Language;
 
             var curIndex = LanguagePrefs.RegisteredLanguages.IndexOf(curLang);
-            var DisplayNames = LanguagePrefs.RegisteredLanguages.Select(
-                    lang =>
-                    {
-                        try
-                        {
-                            return CultureInfo.GetCultureInfo(lang).NativeName;
-                        }
-                        catch (Exception)
-                        {
-                            return lang;
-                        }
-                    })
+            var DisplayNames = LanguagePrefs.RegisteredLanguages
+                .Where(lang => lang.Contains("-") ||
+                               LanguagePrefs.RegisteredLanguages.All(l2 => !l2.StartsWith(lang + "-")))
+                .Select(LanguagePrefs.GetLocaleNativeName)
                 .ToArray();
 
             var newIndex = EditorGUILayout.Popup("Editor Language", curIndex, DisplayNames);

Editor/UI/Localization/LanguagePrefs.cs

@@ -1,7 +1,10 @@
 using System;
 using System.Collections.Generic;
 using System.Collections.Immutable;
+using System.Globalization;
+using System.Linq;
 using System.Runtime.CompilerServices;
+using Newtonsoft.Json;
 using UnityEditor;
 using UnityEngine;
 
@@ -12,9 +15,13 @@ namespace nadena.dev.ndmf.localization
     /// </summary>
     public static class LanguagePrefs
     {
+        private const string LocaleNameDatasetPath =
+            "Packages/nadena.dev.ndmf/Editor/UI/Localization/language_names.json";
         private const string EditorPrefKey = "nadena.dev.ndmf.language-selection";
         private static string _curLanguage = "en-us";
 
+        private static ImmutableDictionary<String, String> LocaleNames;
+
         [InitializeOnLoadMethod]
         private static void Init()
         {
@@ -111,8 +118,40 @@ private static void TriggerLanguageChangeCallbacks()
         /// </summary>
         public static ImmutableSortedSet<string> RegisteredLanguages { get; private set; }
 
+        internal static string GetLocaleNativeName(string locale)
+        {
+            if (LocaleNames.TryGetValue(locale.ToLowerInvariant().Replace("-", "_"), out var name))
+            {
+                return name;
+            }
+            else
+            {
+                try
+                {
+                    return CultureInfo.CreateSpecificCulture(locale).NativeName;
+                }
+                catch (Exception e)
+                {
+                    return locale;
+                }
+            }
+        } 
+
         static LanguagePrefs()
         {
+            try
+            {
+                var localeNameJson = System.IO.File.ReadAllText(LocaleNameDatasetPath);
+                LocaleNames = JsonConvert.DeserializeObject<Dictionary<string, string>>(localeNameJson)
+                    .Select(kvp => { return new KeyValuePair<String, String>(kvp.Key.ToLowerInvariant(), kvp.Value); })
+                    .ToImmutableDictionary();
+            }
+            catch (Exception e)
+            {
+                Debug.LogException(e);
+                LocaleNames = ImmutableDictionary<string, string>.Empty;
+            }
+
             RegisteredLanguages = ImmutableSortedSet<string>.Empty;
         }
 

Editor/UI/Localization/import_icu.py

@@ -0,0 +1,13 @@
+import icu
+import json
+
+locales = icu.Locale.getAvailableLocales()
+
+displayNames = {}
+for locale in locales:
+    icu_locale = icu.Locale(locale)
+    displayNames[locale] = icu_locale.getDisplayName(icu_locale)
+
+displayNames["__comment__"] = "Derived from ICU dataset. See import_icu.py";
+
+print(json.dumps(displayNames, indent=4, ensure_ascii=False))