Update Azure locale mapping

CHANGELOG.md

@@ -54,6 +54,7 @@ and this project adheres to [Semantic Versioning](http://semver.org/spec/v2.0.0.
 -  Resolves [#2213](https://github.com/microsoft/BotFramework-WebChat/issues/2213). Added customization for typing activity, by [@compulim](https://github.com/compulim), in PR [#2912](https://github.com/microsoft/BotFramework-WebChat/pull/2912)
 -  Resolves [#2754](https://github.com/microsoft/BotFramework-WebChat/issues/2754). Added [telemetry system](https://github.com/microsoft/BotFramework-WebChat/tree/master/docs/TELEMETRY.md), by [@compulim](https://github.com/compulim), in PR [#2922](https://github.com/microsoft/BotFramework-WebChat/pull/2922)
 -  Resolves [#2857](https://github.com/microsoft/BotFramework-WebChat/issues/2857). Added the ability to customize the avatar on a per activity basis, by [@compulim](https://github.com/compulim), in PR [#2943](https://github.com/microsoft/BotFramework-WebChat/pull/2943)
+-  Resolves [#2944](https://github.com/microsoft/BotFramework-WebChat/issues/2944). Updated Azure locale mapping in embed page, by [@compulim](https://github.com/compulim) in PR [#2965](https://github.com/microsoft/BotFramework-WebChat/pull/2965)
 
 ### Fixed
 

packages/embed/src/index.js

@@ -1,6 +1,7 @@
 import 'core-js/modules/es.array.includes';
 import 'core-js/modules/es.array.iterator';
 import 'core-js/modules/es.promise';
+import 'core-js/modules/es.string.starts-with';
 import 'core-js/modules/es.symbol';
 import 'url-search-params-polyfill';
 import 'whatwg-fetch';

packages/embed/src/locale.js

@@ -1,4 +1,4 @@
-// Supported Azure languages as of 2019-04-25
+// Supported Azure languages as of 2020-02-28
 // The first part is language (localization), the second part is regional format (internationalization)
 
 // en.en-us
@@ -18,46 +18,51 @@
 // sv.sv-se
 // tr.tr-tr
 // zh-hans.zh-cn
+// zh-hans.zh-sg
+// zh-hant.zh-hk
+// zh-hant.zh-mo
 // zh-hant.zh-tw
 
 // Please note that Arabic and Hebrew are not currently supported by Azure
 
-const AZURE_LOCALE_PATTERN = /^(([a-z]{2})(-[a-z]{2,})?)\.([a-z]{2})/;
-const JAVASCRIPT_LOCALE_PATTERN = /^([a-z]{2})-([A-Z]{2,})?$/;
+const AZURE_LOCALE_PATTERN = /^(([a-z]{2})(-[a-z]{2,})?)\.(([a-z]{2})(-[a-z]{2,})?)/;
+const JAVASCRIPT_LOCALE_PATTERN = /^([a-z]{2})(-([A-Z][\w]+))*$/;
 
 const AZURE_LOCALE_MAPPING = {
-  ar: 'ar-EG',
-  bg: 'bg-BG',
-  cs: 'cs-CZ',
-  de: 'de-DE',
-  en: 'en-US',
-  es: 'es-ES',
-  fr: 'fr-FR',
-  he: 'he-IL',
-  hu: 'hu-HU',
-  it: 'it-IT',
-  ja: 'ja-JP',
-  ko: 'ko-KR',
-  nl: 'nl-NL',
-  pl: 'pl-PL',
-  'pt-br': 'pt-BR',
-  'pt-pt': 'pt-PT',
-  ru: 'ru-RU',
-  sv: 'sv-SE',
-  tr: 'tr-TR',
-  'zh-hans': 'zh-HANS',
-  'zh-hant': 'zh-HANT'
+  cs: { '*': 'cs-CZ' },
+  de: { '*': 'de-DE' },
+  en: { '*': 'en-US' },
+  es: { '*': 'es-ES' },
+  fr: { '*': 'fr-FR' },
+  hu: { '*': 'hu-HU' },
+  it: { '*': 'it-IT' },
+  ja: { '*': 'ja-JP' },
+  ko: { '*': 'ko-KR' },
+  nl: { '*': 'nl-NL' },
+  pl: { '*': 'pl-PL' },
+  'pt-br': { '*': 'pt-BR' },
+  'pt-pt': { '*': 'pt-PT' },
+  ru: { '*': 'ru-RU' },
+  sv: { '*': 'sv-SE' },
+  tr: { '*': 'tr-TR' },
+  'zh-hans': { 'zh-sg': 'zh-Hans-SG', '*': 'zh-Hans' },
+  'zh-hant': { 'zh-hk': 'zh-Hant-HK', 'zh-mo': 'zh-Hant-MO', '*': 'zh-Hant' }
 };
 
 function normalize(language) {
-  const azureLocaleMatch = AZURE_LOCALE_PATTERN.exec(language);
-  const javaScriptLocaleMatch = JAVASCRIPT_LOCALE_PATTERN.exec(language);
   let result;
 
-  if (javaScriptLocaleMatch) {
-    result = language;
-  } else if (azureLocaleMatch) {
-    result = AZURE_LOCALE_MAPPING[azureLocaleMatch[1]];
+  if (language !== 'en') {
+    const azureLocaleMatch = AZURE_LOCALE_PATTERN.exec(language);
+    const javaScriptLocaleMatch = JAVASCRIPT_LOCALE_PATTERN.exec(language);
+
+    if (javaScriptLocaleMatch) {
+      result = language;
+    } else if (azureLocaleMatch) {
+      const mapping = AZURE_LOCALE_MAPPING[azureLocaleMatch[1]];
+
+      result = mapping[azureLocaleMatch[4]] || mapping['*'];
+    }
   }
 
   return result || 'en-US';
@@ -75,26 +80,33 @@ function toAzureLocale(language) {
     case 'pt-PT':
       return 'pt-pt.pt-pt';
 
-    case 'zh-CN':
-    case 'zh-SG':
-      return `zh-hans.${language.toLowerCase()}`;
+    case 'yue':
+    case 'zh-Hant-HK':
+      return 'zh-hant.zh-hk';
 
-    case 'zh-HANS':
+    case 'zh-Hans':
       return 'zh-hans.zh-cn';
 
-    case 'zh-HANT':
+    case 'zh-Hans-SG':
+      return `zh-hans.zh-sg`;
+
+    case 'zh-Hant':
       return 'zh-hant.zh-tw';
 
-    case 'zh-HK':
-    case 'zh-MO':
-    case 'zh-TW':
-      return `zh-hant.${language.toLowerCase()}`;
+    case 'zh-Hant-MO':
+      return 'zh-hant.zh-mo';
   }
 
-  const match = JAVASCRIPT_LOCALE_PATTERN.exec(language);
+  if (
+    Object.keys(AZURE_LOCALE_MAPPING).some(azureLocaleFirstPart =>
+      language.toLowerCase().startsWith(azureLocaleFirstPart)
+    )
+  ) {
+    const match = JAVASCRIPT_LOCALE_PATTERN.exec(language);
 
-  if (match) {
-    return `${match[1]}.${match[1]}-${match[2].toLowerCase()}`;
+    if (match) {
+      return `${match[1]}.${match[1]}-${match[3].toLowerCase()}`;
+    }
   }
 }
 

packages/embed/src/locale.spec.js

@@ -1,17 +1,9 @@
 import { normalize, toAzureLocale } from './locale';
 
-test('Normalizing "ar.ar-eg"', () => {
-  expect(normalize('ar.ar-eg')).toBe('ar-EG');
-});
-
 test('Normalizing "en.en-us"', () => {
   expect(normalize('en.en-us')).toBe('en-US');
 });
 
-test('Normalizing "bg.bg-bg"', () => {
-  expect(normalize('bg.bg-bg')).toBe('bg-BG');
-});
-
 test('Normalizing "cs.cs-cz"', () => {
   expect(normalize('cs.cs-cz')).toBe('cs-CZ');
 });
@@ -28,10 +20,6 @@ test('Normalizing "fr.fr-fr"', () => {
   expect(normalize('fr.fr-fr')).toBe('fr-FR');
 });
 
-test('Normalizing "he.he-IL"', () => {
-  expect(normalize('he.he-IL')).toBe('he-IL');
-});
-
 test('Normalizing "hu.hu-hu"', () => {
   expect(normalize('hu.hu-hu')).toBe('hu-HU');
 });
@@ -77,11 +65,23 @@ test('Normalizing "tr.tr-tr"', () => {
 });
 
 test('Normalizing "zh-hans.zh-cn"', () => {
-  expect(normalize('zh-hans.zh-cn')).toBe('zh-HANS');
+  expect(normalize('zh-hans.zh-cn')).toBe('zh-Hans');
+});
+
+test('Normalizing "zh-hans.zh-sg"', () => {
+  expect(normalize('zh-hans.zh-sg')).toBe('zh-Hans-SG');
+});
+
+test('Normalizing "zh-hant.zh-hk"', () => {
+  expect(normalize('zh-hant.zh-hk')).toBe('zh-Hant-HK');
+});
+
+test('Normalizing "zh-hant.zh-mo"', () => {
+  expect(normalize('zh-hant.zh-mo')).toBe('zh-Hant-MO');
 });
 
 test('Normalizing "zh-hant.zh-tw"', () => {
-  expect(normalize('zh-hant.zh-tw')).toBe('zh-HANT');
+  expect(normalize('zh-hant.zh-tw')).toBe('zh-Hant');
 });
 
 test('Normalizing "en.zh-hk" should become "en-US"', () => {
@@ -92,58 +92,80 @@ test('Normalizing "en"', () => {
   expect(normalize('en')).toBe('en-US');
 });
 
-test('Normalizing "zh-HK"', () => {
-  expect(normalize('zh-HK')).toBe('zh-HK');
+test('Normalizing "zh-Hant-HK"', () => {
+  expect(normalize('zh-Hant-HK')).toBe('zh-Hant-HK');
 });
 
 test('Normalizing "*"', () => {
   expect(normalize('*')).toBe('en-US');
 });
 
-test('Convert "en-US" to Azure locale', () => {
-  expect(toAzureLocale('en-US')).toBe('en.en-us');
-});
-
 test('Convert "fr" to Azure locale', () => {
   expect(toAzureLocale('fr')).toBe('fr.fr-fr');
 });
 
-test('Convert "pt-BR" to Azure locale', () => {
-  expect(toAzureLocale('pt-BR')).toBe('pt-br.pt-br');
-});
-
-test('Convert "pt-PT" to Azure locale', () => {
-  expect(toAzureLocale('pt-PT')).toBe('pt-pt.pt-pt');
-});
-
-test('Convert "zh-CN" to Azure locale', () => {
-  expect(toAzureLocale('zh-CN')).toBe('zh-hans.zh-cn');
-});
-
-test('Convert "zh-HANT" to Azure locale', () => {
-  expect(toAzureLocale('zh-HANT')).toBe('zh-hant.zh-tw');
-});
-
-test('Convert "zh-HANS" to Azure locale', () => {
-  expect(toAzureLocale('zh-HANS')).toBe('zh-hans.zh-cn');
-});
-
-test('Convert "zh-HK" to Azure locale', () => {
-  expect(toAzureLocale('zh-HK')).toBe('zh-hant.zh-hk');
-});
-
-test('Convert "zh-MO" to Azure locale', () => {
-  expect(toAzureLocale('zh-MO')).toBe('zh-hant.zh-mo');
-});
-
-test('Convert "zh-SG" to Azure locale', () => {
-  expect(toAzureLocale('zh-SG')).toBe('zh-hans.zh-sg');
-});
-
-test('Convert "zh-TW" to Azure locale', () => {
-  expect(toAzureLocale('zh-TW')).toBe('zh-hant.zh-tw');
-});
-
 test('Convert "*" to Azure locale', () => {
   expect(toAzureLocale('*')).toEqual(undefined);
 });
+
+test('Convert all Azure supported languages to Azure locale', () => {
+  const expected = {
+    'ar-EG': undefined,
+    'ar-JO': undefined,
+    'ar-SA': undefined,
+    'bg-BG': undefined,
+    'ca-ES': undefined,
+    'cs-CZ': 'cs.cs-cz',
+    'da-DK': undefined,
+    'de-DE': 'de.de-de',
+    'el-GR': undefined,
+    'en-US': 'en.en-us',
+    'es-ES': 'es.es-es',
+    'et-EE': undefined,
+    'eu-ES': undefined,
+    'fi-FI': undefined,
+    'fr-FR': 'fr.fr-fr',
+    'gl-ES': undefined,
+    'he-IL': undefined,
+    'hi-IN': undefined,
+    'hr-HR': undefined,
+    'hu-HU': 'hu.hu-hu',
+    'id-ID': undefined,
+    'it-IT': 'it.it-it',
+    'ja-JP': 'ja.ja-jp',
+    'kk-KZ': undefined,
+    'ko-KR': 'ko.ko-kr',
+    'lt-LT': undefined,
+    'lv-LV': undefined,
+    'ms-MY': undefined,
+    'nb-NO': undefined,
+    'nl-NL': 'nl.nl-nl',
+    'pl-PL': 'pl.pl-pl',
+    'pt-BR': 'pt-br.pt-br',
+    'pt-PT': 'pt-pt.pt-pt',
+    'ro-RO': undefined,
+    'ru-RU': 'ru.ru-ru',
+    'sk-SK': undefined,
+    'sl-SI': undefined,
+    'sr-Cyrl': undefined,
+    'sr-Latn': undefined,
+    'sv-SE': 'sv.sv-se',
+    'th-TH': undefined,
+    'tr-TR': 'tr.tr-tr',
+    'uk-UA': undefined,
+    'vi-VN': undefined,
+    yue: 'zh-hant.zh-hk',
+    'zh-Hans': 'zh-hans.zh-cn',
+    'zh-Hans-SG': 'zh-hans.zh-sg',
+    'zh-Hant': 'zh-hant.zh-tw',
+    'zh-Hant-HK': 'zh-hant.zh-hk',
+    'zh-Hant-MO': 'zh-hant.zh-mo'
+  };
+
+  const actual = Object.keys(expected).reduce(
+    (actual, language) => ({ ...actual, [language]: toAzureLocale(language) }),
+    {}
+  );
+
+  expect(actual).toEqual(expected);
+});