update canonicalizeLocale to handle 中文(chinese) (#239)

J=SLAP-1526
TEST=auto,manual

unit tests
ran build with zh-hans-ch in locale config, saw that the generated source code used the correct locale in the page
tried building a page with a locale with >3 sections, got a jambo warning but build continued without changing the locale

src/utils/i18nutils.js

@@ -1,25 +1,87 @@
+const UserError = require('../errors/usererror');
+
 /**
  * Normalizes a locale code
  *
  * @param {string} localeCode
  * @returns {string}
  */
-canonicalizeLocale = function(localeCode) {
+exports.canonicalizeLocale = function(localeCode) {
   if (!localeCode) {
     return;
   }
-  const localeCodeSections = localeCode.replace('-', '_')
-    .split('_');
-
-  const languageIndex = 0;
-  const regionIndex = 1;
+  const { language, modifier, region } = parseLocale(localeCode);
+  return formatLocale(language, modifier, region);
+}
 
-  localeCodeSections[languageIndex] = localeCodeSections[languageIndex].toLowerCase();
+/**
+ * Parses a locale code into its constituent parts.
+ * Performs case formatting on the result.
+ * 
+ * @param {string} localeCode 
+ * @returns { language: string, modifier?: string, region?: string } 
+ */
+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 UserError(
+        `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 (localeCodeSections.length > regionIndex) {
-    localeCodeSections[regionIndex] = localeCodeSections[regionIndex].toUpperCase();
+  if (parsedLocale.modifier) {
+    parsedLocale.modifier = capitalizeFirstLetterOnly(parsedLocale.modifier);
+  }
+  if (parsedLocale.region) {
+    parsedLocale.region = parsedLocale.region.toUpperCase();
   }
 
-  return localeCodeSections.join('_');
+  return parsedLocale;
 }
-exports.canonicalizeLocale = canonicalizeLocale;
+exports.parseLocale = parseLocale;
+
+/**
+ * Formats a locale code given its constituent parts.
+ *
+ * @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 formatLocale(language, modifier, region) {
+  let result = language.toLowerCase();
+  if (modifier) {
+    result += '-' + modifier;
+  }
+  if (region) {
+    result += '_' + region;
+  }
+  return result;
+}

tests/acceptance/suites/themeupgrader.js

@@ -5,7 +5,7 @@ const ThemeManager = require('../../../src/utils/thememanager');
 const path = require('path');
 
 ThemeManager.getRepoForTheme = () => {
-    return path.resolve(__dirname, '../test-themes/basic-flow');
+  return path.resolve(__dirname, '../test-themes/basic-flow');
 };
 
 //Silence error logs
@@ -14,15 +14,15 @@ console.error = jest.fn();
 afterAll(() => console.error = error);
 
 it('tests upgrade fail', () => runInPlayground(async t => {
-    await git.cwd(process.cwd());
-    await t.jambo('init');
-    await t.jambo(
-        'import --themeUrl ../test-themes/basic-flow');
-    await git.add('.');
-    await git.commit('theme');
-    await expect(t.jambo('upgrade --branch fail')).rejects.toThrow(
-        /Remote branch fail not found in upstream origin/);
-    const diff = await git.diff();
-    expect(diff).toBe('');
-  }));
+  await git.cwd(process.cwd());
+  await t.jambo('init');
+  await t.jambo(
+    'import --themeUrl ../test-themes/basic-flow');
+  await git.add('.');
+  await git.commit('theme');
+  await expect(t.jambo('upgrade --branch fail')).rejects.toThrow(
+    /Remote branch fail not found in upstream origin/);
+  const diff = await git.diff();
+  expect(diff).toBe('');
+}));
 

tests/utils/i18nutils.js

@@ -1,4 +1,4 @@
-const { canonicalizeLocale } = require('../../src/utils/i18nutils');
+const { canonicalizeLocale, parseLocale } = require('../../src/utils/i18nutils');
 
 describe('canonicalizeLocale correctly normalizes locales', () => {
   it('converts language to lower case and region to upper case', () => {
@@ -12,4 +12,62 @@ describe('canonicalizeLocale correctly normalizes locales', () => {
     const canonicalizedLocale = canonicalizeLocale(locale);
     expect(canonicalizedLocale).toEqual('fr_CH');
   });
-});
+
+  describe('works for chinese', () => {
+    function runTest(testName, inputLocale, expectedLocale) {
+      it(testName, () => {
+        expect(canonicalizeLocale(inputLocale)).toEqual(expectedLocale);
+      });
+    }
+    const testCases = [
+      ['using dashes', 'zh-Hans-CH'],
+      ['using underscores', 'zh_Hans_CH'],
+      ['underscore then dash', 'zh_Hans-CH'],
+      ['dash then underscore', 'zh-Hans_CH'],
+      ['updates casing', 'ZH-hans_Ch'],
+      ['does not have region code', 'zh-hans', 'zh-Hans'],
+      ['has region but no modifier', 'zh-cH', 'zh_CH']
+    ];
+    const expected = 'zh-Hans_CH';
+    for (const [testName, inputLocale, specificExpected] of testCases) {
+      runTest(testName, inputLocale, specificExpected || expected);
+    }
+  });
+});
+
+describe('parseLocale', () => {
+  it('performs case formatting', () => {
+    expect(parseLocale('Zh-hans-Ch')).toEqual({
+      language: 'zh',
+      modifier: 'Hans',
+      region: 'CH'
+    })
+  });
+
+  it('chinese with modifier only', () => {
+    expect(parseLocale('ZH_HANS')).toEqual({
+      language: 'zh',
+      modifier: 'Hans'
+    })
+  });
+
+  it('chinese with region only', () => {
+    expect(parseLocale('ZH-cH')).toEqual({
+      language: 'zh',
+      region: 'CH'
+    })
+  });
+
+  it('2 section non-chinese locale', () => {
+    expect(parseLocale('FR-freE')).toEqual({
+      language: 'fr',
+      region: 'FREE'
+    });
+  });
+
+  it('simple language', () => {
+    expect(parseLocale('FR')).toEqual({
+      language: 'fr'
+    });
+  });
+});