GetPhonebookIndex: Fix handling for minor Japanese kana characters

This adds minor Japanese kana character handling to meet the
unicode collation algorithm.

-Normalize digraphs (yori, koto) into first reading letter.
-Treat dakuten, dot, onbiki, iteration marks as a symbol.
-Add handling for minor small katakana letter.
Signed-off-by: Yutaro Ogasawara <yutaro.ogasawara@gmail.com>

android/PhonebookIndex.cpp

@@ -169,19 +169,23 @@ int32_t GetPhonebookIndex(UCharIterator * iter, const char * locale, UChar * out
   c = android::GetNormalizedCodePoint(c, next, NULL);
 
   // Traditional grouping of Hiragana characters
-  if (0x3042 <= c && c <= 0x309F) {
+  if (0x3041 <= c && c <= 0x309F) {
     if (c < 0x304B) c = 0x3042;         // a
     else if (c < 0x3055) c = 0x304B;    // ka
     else if (c < 0x305F) c = 0x3055;    // sa
     else if (c < 0x306A) c = 0x305F;    // ta
     else if (c < 0x306F) c = 0x306A;    // na
     else if (c < 0x307E) c = 0x306F;    // ha
-    else if (c < 0x3084) c = 0x307E;    // ma
+    else if (c < 0x3083) c = 0x307E;    // ma
     else if (c < 0x3089) c = 0x3084;    // ya
-    else if (c < 0x308F) c = 0x3089;    // ra
-    else c = 0x308F;                    // wa
+    else if (c < 0x308E) c = 0x3089;    // ra
+    else if (c < 0x3094) c = 0x308F;    // wa
+    else return 0;                      // Others are not readable
     out[0] = c;
     return 1;
+  } else if (0x30A0 <= c && c <= 0x30FF) {
+    // Dot, onbiki, iteration marks are not readable
+    return 0;
   }
 
   if (is_CJK(c)) {

android/PhoneticStringUtils.cpp

@@ -153,12 +153,45 @@ static int GetNormalizedHiragana(int codepoint) {
         case 0x3045:
         case 0x3047:
         case 0x3049:
+        case 0x3063:
+        case 0x3083:
+        case 0x3085:
+        case 0x3087:
         case 0x308E: // xwa
             return codepoint + 1;
         case 0x3095: // xka
             return 0x304B;
-        case 0x3096: // xku
+        case 0x3096: // xke
+            return 0x3051;
+        case 0x31F0: // xku
             return 0x304F;
+        case 0x31F1: // xsi
+            return 0x3057;
+        case 0x31F2: // xsu
+            return 0x3059;
+        case 0x31F3: // xto
+            return 0x3068;
+        case 0x31F4: // xnu
+            return 0x306C;
+        case 0x31F5: // xha
+            return 0x306F;
+        case 0x31F6: // xhi
+            return 0x3072;
+        case 0x31F7: // xhu
+            return 0x3075;
+        case 0x31F8: // xhe
+            return 0x3078;
+        case 0x31F9: // xho
+            return 0x307B;
+        case 0x31FA: // xmu
+            return 0x3080;
+        case 0x31FB: // xra
+        case 0x31FC: // xri
+        case 0x31FD: // xru
+        case 0x31FE: // xre
+        case 0x31FF: // xro
+            // ra: 0x3089
+            return 0x3089 + (codepoint - 0x31FB);
         default:
             return codepoint;
     }
@@ -172,6 +205,12 @@ static int GetNormalizedKana(char32_t codepoint,
         // Make fullwidth katakana same as hiragana.
         // 96 == 0x30A1 - 0x3041c
         codepoint = codepoint - 96;
+    } else if (codepoint == 0x309F) {
+        // Digraph YORI; Yo
+        codepoint = 0x3088;
+    } else if (codepoint == 0x30FF) {
+        // Digraph KOTO; Ko
+        codepoint = 0x3053;
     } else {
         codepoint = GetHiraganaFromHalfwidthKatakana(
                 codepoint, next_codepoint, next_is_consumed);