Fallback hyphenation locale

When the `lang` attribute does not match the hyphenation
dictionary exactly, this patch computes fallback languages,
by incrementally removing the last component separated by
U+002D HYPHEN-MINUS character.

This allows, for instance, `lang="fr-fr"` or `lang=`fr-ca`"
to find the `fr` dictionary.

Bug: 815059
Change-Id: Ia4b76621dfb18b5aef604e8c7351ed9322f056dc
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2505353
Commit-Queue: Koji Ishii <kojii@chromium.org>
Reviewed-by: Ian Kilpatrick <ikilpatrick@chromium.org>
Reviewed-by: Kent Tamura <tkent@chromium.org>
Reviewed-by: Yoshifumi Inoue <yosin@chromium.org>
Cr-Commit-Position: refs/heads/master@{#822059}

third_party/blink/renderer/platform/text/hyphenation/hyphenation_minikin.cc

@@ -199,6 +199,15 @@ static LocaleMap CreateLocaleFallbackMap() {
       {"und-Orya", "or"},  // Oriya
       {"und-Taml", "ta"},  // Tamil
       {"und-Telu", "te"},  // Telugu
+
+      // List of locales with hyphens not to fall back.
+      {"de-1901", nullptr},
+      {"de-1996", nullptr},
+      {"de-ch-1901", nullptr},
+      {"en-gb", nullptr},
+      {"en-us", nullptr},
+      {"mn-cyrl", nullptr},
+      {"und-ethi", nullptr},
   };
   LocaleMap map;
   for (const auto& it : locale_fallback_data)
@@ -206,12 +215,28 @@ static LocaleMap CreateLocaleFallbackMap() {
   return map;
 }
 
+// static
+AtomicString HyphenationMinikin::MapLocale(const AtomicString& locale) {
+  DEFINE_STATIC_LOCAL(LocaleMap, locale_fallback, (CreateLocaleFallbackMap()));
+  for (AtomicString mapped_locale = locale;;) {
+    const auto& it = locale_fallback.find(mapped_locale);
+    if (it != locale_fallback.end()) {
+      if (it->value)
+        return it->value;
+      return mapped_locale;
+    }
+    const wtf_size_t last_hyphen = mapped_locale.ReverseFind('-');
+    if (last_hyphen == kNotFound || !last_hyphen)
+      return mapped_locale;
+    mapped_locale = AtomicString(mapped_locale.GetString().Left(last_hyphen));
+  }
+}
+
 scoped_refptr<Hyphenation> Hyphenation::PlatformGetHyphenation(
     const AtomicString& locale) {
-  DEFINE_STATIC_LOCAL(LocaleMap, locale_fallback, (CreateLocaleFallbackMap()));
-  const auto& it = locale_fallback.find(locale);
-  if (it != locale_fallback.end())
-    return LayoutLocale::Get(it->value)->GetHyphenation();
+  const AtomicString mapped_locale = HyphenationMinikin::MapLocale(locale);
+  if (mapped_locale.Impl() != locale.Impl())
+    return LayoutLocale::Get(mapped_locale)->GetHyphenation();
 
   scoped_refptr<HyphenationMinikin> hyphenation(
       base::AdoptRef(new HyphenationMinikin));

third_party/blink/renderer/platform/text/hyphenation/hyphenation_minikin.h

@@ -28,6 +28,8 @@ class PLATFORM_EXPORT HyphenationMinikin final : public Hyphenation {
                                 wtf_size_t before_index) const override;
   Vector<wtf_size_t, 8> HyphenLocations(const StringView&) const override;
 
+  static AtomicString MapLocale(const AtomicString& locale);
+
   static scoped_refptr<HyphenationMinikin> FromFileForTesting(base::File);
 
  private:

third_party/blink/renderer/platform/text/hyphenation_test.cc

@@ -72,6 +72,33 @@ TEST_F(HyphenationTest, Get) {
   EXPECT_EQ(nullptr, LayoutLocale::Get("en-UK")->GetHyphenation());
 }
 
+#if defined(USE_MINIKIN_HYPHENATION)
+TEST_F(HyphenationTest, MapLocale) {
+  EXPECT_EQ(HyphenationMinikin::MapLocale("de-de"), "de-1996");
+  EXPECT_EQ(HyphenationMinikin::MapLocale("de-de-xyz"), "de-1996");
+  EXPECT_EQ(HyphenationMinikin::MapLocale("de-li"), "de-1996");
+  EXPECT_EQ(HyphenationMinikin::MapLocale("de-li-1901"), "de-ch-1901");
+  EXPECT_EQ(HyphenationMinikin::MapLocale("en"), "en-gb");
+  EXPECT_EQ(HyphenationMinikin::MapLocale("en-gu"), "en-us");
+  EXPECT_EQ(HyphenationMinikin::MapLocale("en-gu-xyz"), "en-us");
+  EXPECT_EQ(HyphenationMinikin::MapLocale("en-xyz"), "en-gb");
+  EXPECT_EQ(HyphenationMinikin::MapLocale("en-xyz-xyz"), "en-gb");
+  EXPECT_EQ(HyphenationMinikin::MapLocale("fr-ca"), "fr");
+  EXPECT_EQ(HyphenationMinikin::MapLocale("fr-fr"), "fr");
+  EXPECT_EQ(HyphenationMinikin::MapLocale("fr-fr-xyz"), "fr");
+  EXPECT_EQ(HyphenationMinikin::MapLocale("mn-xyz"), "mn-cyrl");
+  EXPECT_EQ(HyphenationMinikin::MapLocale("und-Deva-xyz"), "hi");
+
+  const char* no_map_locales[] = {"en-us", "fr"};
+  for (const char* locale_str : no_map_locales) {
+    AtomicString locale(locale_str);
+    AtomicString mapped_locale = HyphenationMinikin::MapLocale(locale);
+    // If no mapping, the same instance should be returned.
+    EXPECT_EQ(locale.Impl(), mapped_locale.Impl());
+  }
+}
+#endif
+
 #if defined(USE_MINIKIN_HYPHENATION) || defined(OS_MAC)
 TEST_F(HyphenationTest, HyphenLocations) {
   scoped_refptr<Hyphenation> hyphenation = GetHyphenation("en-us");