Merge pull request #1523 from yext/feature/chinese-i18n

chinese i18n tweaks

update edge supported locales for voice search (#1519)
rename zh-CN to zh-Hans and zh-TW to zh-Hant (#1517)

conf/i18n/constants.js

@@ -134,15 +134,15 @@ const LANGUAGES_TO_LOCALES = {
   sv: [
     'sv_SE'
   ],
-  'zh-CN': [
-    'zh-CN_CN',
-    'zh_CN_HK',
-    'zh_CN_SG'
+  'zh-Hans': [
+    'zh-Hans_CN',
+    'zh-Hans_HK',
+    'zh-Hans_SG'
   ],
-  'zh-TW': [
-    'zh-TW_HK',
-    'zh-TW_MO',
-    'zh-TW_TW'
+  'zh-Hant': [
+    'zh-Hant_HK',
+    'zh-Hant_MO',
+    'zh-Hant_TW'
   ]
 };
 exports.LANGUAGES_TO_LOCALES = LANGUAGES_TO_LOCALES;
@@ -186,5 +186,28 @@ exports.SPEECH_RECOGNITION_LOCALES_SUPPORTED_BY_EDGE = [
   'de-at',
   'de-ch',
   'de-de',
-  'ja-jp'
+  'ja-jp',
+  'ar-ae',
+  'ar-bh',
+  'ar-dz',
+  'ar-iq',
+  'ar-kw',
+  'ar-ly',
+  'ar-om',
+  'ar-ps',
+  'ar-qa',
+  'ar-sa',
+  'nl-be',
+  'nl-nl',
+  'pt-br',
+  'pt-pt',
+  'sv-se',
+  'zh-hans-cn',
+  'zh-hans-hk',
+  'zh-hant-hk',
+  'zh-hant-tw',
+
+  // Below are locales that are not the SDK's built-in locales, but are supported by edge
+  'zh-cn',
+  'zh-tw'
 ];

src/core/speechrecognition/locales.js

@@ -1,4 +1,5 @@
 import { SPEECH_RECOGNITION_LOCALES_SUPPORTED_BY_EDGE } from '../constants';
+import { parseLocale } from '../utils/i18nutils';
 
 /**
  * Transforms the given locale to a locale Microsoft Edge can understand.
@@ -9,15 +10,39 @@ import { SPEECH_RECOGNITION_LOCALES_SUPPORTED_BY_EDGE } from '../constants';
  * @param {string} locale
  * @returns {string}
  */
-export function transformSpeechRecognitionLocaleForEdge (locale) {
-  const underscoreIndex = locale.indexOf('_');
-  if (underscoreIndex === -1) {
-    return locale;
+export function transformSpeechRecognitionLocaleForEdge (rawLocale) {
+  const { language, modifier, region } = parseLocale(rawLocale);
+  if (!modifier && !region) {
+    return language;
   }
-  locale = locale.replace('_', '-');
-  const isCompatibleWithEdge = SPEECH_RECOGNITION_LOCALES_SUPPORTED_BY_EDGE.includes(locale.toLowerCase());
+  const locale = formatLocaleForEdge(language, modifier, region);
+  const isCompatibleWithEdge = SPEECH_RECOGNITION_LOCALES_SUPPORTED_BY_EDGE.includes(locale);
   if (isCompatibleWithEdge) {
     return locale;
   }
-  return locale.substring(0, underscoreIndex);
+  if (modifier) {
+    return formatLocaleForEdge(language, modifier);
+  }
+  return language;
+}
+
+/**
+ * Formats a locale code given its constituent parts for Edge (which does not accept underscores).
+ * Edge does not care about capitalization, but converting to full lowercase allows for easier lookup
+ * within the SPEECH_RECOGNITION_LOCALES_SUPPORTED_BY_EDGE array.
+ *
+ * @param {string} language zh in zh-Hans_CH
+ * @param {string?} modifier Hans in zh-Hans_CH
+ * @param {string?} region CH in zh-Hans_CH
+ * @returns
+ */
+function formatLocaleForEdge (language, modifier, region) {
+  let result = language;
+  if (modifier) {
+    result += '-' + modifier;
+  }
+  if (region) {
+    result += '-' + region;
+  }
+  return result.toLowerCase();
 }

src/core/utils/i18nutils.js

@@ -0,0 +1,51 @@
+/**
+ * Parses a locale code into its constituent parts.
+ * Performs case formatting on the result.
+ *
+ * @param {string} localeCode
+ * @returns { language: string, modifier?: string, region?: string }
+ */
+export function parseLocale (localeCode) {
+  const localeCodeSections = localeCode.replace(/-/g, '_').split('_');
+  const language = localeCodeSections[0].toLowerCase();
+  const parseModifierAndRegion = () => {
+    const numSections = localeCodeSections.length;
+    if (numSections === 1) {
+      return {};
+    } else if (numSections === 2 && language === 'zh') {
+      const ambiguous = localeCodeSections[1].toLowerCase();
+      if (['hans', 'hant'].includes(ambiguous)) {
+        return { modifier: ambiguous };
+      } else {
+        return { region: ambiguous };
+      }
+    } else if (numSections === 2) {
+      return { region: localeCodeSections[1] };
+    } else if (numSections === 3) {
+      return {
+        modifier: localeCodeSections[1],
+        region: localeCodeSections[2]
+      };
+    } else if (numSections > 3) {
+      throw new Error(
+        `Encountered strangely formatted locale "${localeCode}", ` +
+        `with ${numSections} sections.`);
+    }
+  };
+  const capitalizeFirstLetterOnly = raw => {
+    return raw.charAt(0).toUpperCase() + raw.slice(1).toLowerCase();
+  };
+  const parsedLocale = {
+    language,
+    ...parseModifierAndRegion()
+  };
+
+  if (parsedLocale.modifier) {
+    parsedLocale.modifier = capitalizeFirstLetterOnly(parsedLocale.modifier);
+  }
+  if (parsedLocale.region) {
+    parsedLocale.region = parsedLocale.region.toUpperCase();
+  }
+
+  return parsedLocale;
+}

tests/core/speechrecognition/locales.js

@@ -5,27 +5,31 @@ jest.mock('../../../src/core/constants', () => ({
     require('../../../conf/i18n/constants').SPEECH_RECOGNITION_LOCALES_SUPPORTED_BY_EDGE
 }));
 
-it('does nothing when no underscore', () => {
+it('works for plain languages', () => {
   expect(transformSpeechRecognitionLocaleForEdge('en')).toEqual('en');
+  expect(transformSpeechRecognitionLocaleForEdge('ZH_Hans')).toEqual('zh-hans');
 });
 
 it('will recognize supported locales that have dashes', () => {
-  expect(transformSpeechRecognitionLocaleForEdge('en-US')).toEqual('en-US');
-  expect(transformSpeechRecognitionLocaleForEdge('en-GB')).toEqual('en-GB');
+  expect(transformSpeechRecognitionLocaleForEdge('en-US')).toEqual('en-us');
+  expect(transformSpeechRecognitionLocaleForEdge('en-GB')).toEqual('en-gb');
+  expect(transformSpeechRecognitionLocaleForEdge('zh-Hant-tw')).toEqual('zh-hant-tw');
 });
 
 it('defaults Edge incompatible locales to the language code', () => {
   expect(transformSpeechRecognitionLocaleForEdge('ja_FAKE')).toEqual('ja');
   expect(transformSpeechRecognitionLocaleForEdge('en_AI')).toEqual('en');
+  expect(transformSpeechRecognitionLocaleForEdge('zH_hAns_fake')).toEqual('zh-hans');
+  expect(transformSpeechRecognitionLocaleForEdge('ZH-HANS-FAKE')).toEqual('zh-hans');
 });
 
 it('replaces underscores with dashes for supported locales', () => {
-  expect(transformSpeechRecognitionLocaleForEdge('en_US')).toEqual('en-US');
-  expect(transformSpeechRecognitionLocaleForEdge('en_GB')).toEqual('en-GB');
+  expect(transformSpeechRecognitionLocaleForEdge('en_US')).toEqual('en-us');
+  expect(transformSpeechRecognitionLocaleForEdge('en_GB')).toEqual('en-gb');
 });
 
-it('is case insensitive', () => {
-  expect(transformSpeechRecognitionLocaleForEdge('en_us')).toEqual('en-us');
-  expect(transformSpeechRecognitionLocaleForEdge('EN_AI')).toEqual('EN');
-  expect(transformSpeechRecognitionLocaleForEdge('jA_AI')).toEqual('jA');
+it('canonicalizes case', () => {
+  expect(transformSpeechRecognitionLocaleForEdge('en_uS')).toEqual('en-us');
+  expect(transformSpeechRecognitionLocaleForEdge('EN_AI')).toEqual('en');
+  expect(transformSpeechRecognitionLocaleForEdge('jA_AI')).toEqual('ja');
 });