In Japanese, ぁ and あ are treated as different characters in anytime. か and が are as well.
But in Safari and Chrome, they are treated as the same characters in its search.
In Japanese, ぁ and あ are treated as different characters in anytime. か and が are as well.
But in Safari and Chrome, they are treated as the same characters in its search.
As you know in English, abc and ABC are treated as the same in a case insensitive context like application searches.
But in Japanese, for example, あった and あつた are always different words in any contexts. Because in Japanese semantics, っ is NOT considered as a small form of つ. These characters are never treated as the same characters.
In the current Unicode Collation Algorithm, っ and つ are in the same order in the primary collation strength.
WebKit uses the primary collation strength in ICU for its search.
I reported this problem in the Unicode ML. (My report)
Mark Davis replied to my report: (His reply)
UTS#10 does not necessarily match the sorting of any particular language.
It means we cannot use ICU's search function directly for application searches. It needs some tailoring in the collation table for some languages.
I wrote a patch for WebKit to add the following tailoring rules for Japanese text search. This patch doesn't have any regression in the other languages.
&ぁ=ァ=ｧ<あ=ア=ｱ<ぃ=ィ=ｨ<い=イ=ｲ<ぅ=ゥ=ｩ<う=ウ=ｳ<ゔ=ヴ<ぇ=ェ=ｪ<え=エ=ｴ<ぉ=ォ=ｫ<お=オ=ｵ <ゕ=ヵ<か=カ=ｶ<が=ガ<き=キ=ｷ<ぎ=ギ<く=ク=ｸ<ぐ=グ<ゖ=ヶ<け=ケ=ｹ<げ=ゲ<こ=コ=ｺ<ご=ゴ <さ=サ=ｻ<ざ=ザ<し=シ=ｼ<じ=ジ<す=ス=ｽ<ず=ズ<せ=セ=ｾ<ぜ=ゼ<そ=ソ=ｿ<ぞ=ゾ <た=タ=ﾀ<だ=ダ<ち=チ=ﾁ<ぢ=ヂ<っ=ッ=ｯ<つ=ツ=ﾂ<づ=ヅ<て=テ=ﾃ<で=デ<と=ト=ﾄ<ど=ド <な=ナ=ﾅ<に=ニ=ﾆ<ぬ=ヌ=ﾇ<ね=ネ=ﾈ<の=ノ=ﾉ <は=ハ=ﾊ<ば=バ<ぱ=パ<ひ=ヒ=ﾋ<び=ビ<ぴ=ピ<ふ=フ=ﾌ<ぶ=ブ<ぷ=プ<へ=ヘ=ﾍ<べ=ベ<ぺ=ペ<ほ=ホ=ﾎ<ぼ=ボ<ぽ=ポ <ま=マ=ﾏ<み=ミ=ﾐ<む=ム=ﾑ<め=メ=ﾒ<も=モ=ﾓ <ゃ=ャ=ｬ<や=ヤ=ﾔ<ゅ=ュ=ｭ<ゆ=ユ=ﾕ<ょ=ョ=ｮ<よ=ヨ=ﾖ <ら=ラ=ﾗ<り=リ=ﾘ<る=ル=ﾙ<れ=レ=ﾚ<ろ=ロ=ﾛ <ゎ=ヮ<わ=ワ=ﾜ<ヷ<ゐ=ヰ<ヸ<ゑ=ヱ<を=ヲ=ｦ<ん=ン=ﾝ
ぁ=ァ=ｧ<あ=ア=ｱ means:
ぁ, ァ and ｧ are in the same order.ぁ, ァ and ｧ are smaller than あ, ア and ｱ.
Index: WebCore/editing/TextIterator.cpp
===================================================================
--- WebCore/editing/TextIterator.cpp (revision 49673)
+++ WebCore/editing/TextIterator.cpp (working copy)
@@ -1443,9 +1443,90 @@
return searcher;
}
+// Tailored collation rules for Japanese text search.
+// The default Unicode Collation Algorithm is unnatural in Japanese.
+// These rules intend to treat the following characters as different characters.
+//
+// - Small kana letters and normal kana letters
+// - Voiceless letters, voiced letters and semi-voiced letters
+//
+static const UChar JAPANESE_KANA_COLLATION_RULES[] = {
+ '&', 0x3041, '=', 0x30a1, '=', 0xff67, '<', 0x3042,
+ '=', 0x30a2, '=', 0xff71, '<', 0x3043, '=', 0x30a3,
+ '=', 0xff68, '<', 0x3044, '=', 0x30a4, '=', 0xff72,
+ '<', 0x3045, '=', 0x30a5, '=', 0xff69, '<', 0x3046,
+ '=', 0x30a6, '=', 0xff73, '<', 0x3094, '=', 0x30f4,
+ '<', 0x3047, '=', 0x30a7, '=', 0xff6a, '<', 0x3048,
+ '=', 0x30a8, '=', 0xff74, '<', 0x3049, '=', 0x30a9,
+ '=', 0xff6b, '<', 0x304a, '=', 0x30aa, '=', 0xff75,
+ '<', 0x3095, '=', 0x30f5, '<', 0x304b, '=', 0x30ab,
+ '=', 0xff76, '<', 0x304c, '=', 0x30ac, '<', 0x304d,
+ '=', 0x30ad, '=', 0xff77, '<', 0x304e, '=', 0x30ae,
+ '<', 0x304f, '=', 0x30af, '=', 0xff78, '<', 0x3050,
+ '=', 0x30b0, '<', 0x3096, '=', 0x30f6, '<', 0x3051,
+ '=', 0x30b1, '=', 0xff79, '<', 0x3052, '=', 0x30b2,
+ '<', 0x3053, '=', 0x30b3, '=', 0xff7a, '<', 0x3054,
+ '=', 0x30b4, '<', 0x3055, '=', 0x30b5, '=', 0xff7b,
+ '<', 0x3056, '=', 0x30b6, '<', 0x3057, '=', 0x30b7,
+ '=', 0xff7c, '<', 0x3058, '=', 0x30b8, '<', 0x3059,
+ '=', 0x30b9, '=', 0xff7d, '<', 0x305a, '=', 0x30ba,
+ '<', 0x305b, '=', 0x30bb, '=', 0xff7e, '<', 0x305c,
+ '=', 0x30bc, '<', 0x305d, '=', 0x30bd, '=', 0xff7f,
+ '<', 0x305e, '=', 0x30be, '<', 0x305f, '=', 0x30bf,
+ '=', 0xff80, '<', 0x3060, '=', 0x30c0, '<', 0x3061,
+ '=', 0x30c1, '=', 0xff81, '<', 0x3062, '=', 0x30c2,
+ '<', 0x3063, '=', 0x30c3, '=', 0xff6f, '<', 0x3064,
+ '=', 0x30c4, '=', 0xff82, '<', 0x3065, '=', 0x30c5,
+ '<', 0x3066, '=', 0x30c6, '=', 0xff83, '<', 0x3067,
+ '=', 0x30c7, '<', 0x3068, '=', 0x30c8, '=', 0xff84,
+ '<', 0x3069, '=', 0x30c9, '<', 0x306a, '=', 0x30ca,
+ '=', 0xff85, '<', 0x306b, '=', 0x30cb, '=', 0xff86,
+ '<', 0x306c, '=', 0x30cc, '=', 0xff87, '<', 0x306d,
+ '=', 0x30cd, '=', 0xff88, '<', 0x306e, '=', 0x30ce,
+ '=', 0xff89, '<', 0x306f, '=', 0x30cf, '=', 0xff8a,
+ '<', 0x3070, '=', 0x30d0, '<', 0x3071, '=', 0x30d1,
+ '<', 0x3072, '=', 0x30d2, '=', 0xff8b, '<', 0x3073,
+ '=', 0x30d3, '<', 0x3074, '=', 0x30d4, '<', 0x3075,
+ '=', 0x30d5, '=', 0xff8c, '<', 0x3076, '=', 0x30d6,
+ '<', 0x3077, '=', 0x30d7, '<', 0x3078, '=', 0x30d8,
+ '=', 0xff8d, '<', 0x3079, '=', 0x30d9, '<', 0x307a,
+ '=', 0x30da, '<', 0x307b, '=', 0x30db, '=', 0xff8e,
+ '<', 0x307c, '=', 0x30dc, '<', 0x307d, '=', 0x30dd,
+ '<', 0x307e, '=', 0x30de, '=', 0xff8f, '<', 0x307f,
+ '=', 0x30df, '=', 0xff90, '<', 0x3080, '=', 0x30e0,
+ '=', 0xff91, '<', 0x3081, '=', 0x30e1, '=', 0xff92,
+ '<', 0x3082, '=', 0x30e2, '=', 0xff93, '<', 0x3083,
+ '=', 0x30e3, '=', 0xff6c, '<', 0x3084, '=', 0x30e4,
+ '=', 0xff94, '<', 0x3085, '=', 0x30e5, '=', 0xff6d,
+ '<', 0x3086, '=', 0x30e6, '=', 0xff95, '<', 0x3087,
+ '=', 0x30e7, '=', 0xff6e, '<', 0x3088, '=', 0x30e8,
+ '=', 0xff96, '<', 0x3089, '=', 0x30e9, '=', 0xff97,
+ '<', 0x308a, '=', 0x30ea, '=', 0xff98, '<', 0x308b,
+ '=', 0x30eb, '=', 0xff99, '<', 0x308c, '=', 0x30ec,
+ '=', 0xff9a, '<', 0x308d, '=', 0x30ed, '=', 0xff9b,
+ '<', 0x308e, '=', 0x30ee, '<', 0x308f, '=', 0x30ef,
+ '=', 0xff9c, '<', 0x30f7, '<', 0x3090, '=', 0x30f0,
+ '<', 0x30f8, '<', 0x3091, '=', 0x30f1, '<', 0x3092,
+ '=', 0x30f2, '=', 0xff66, '<', 0x3093, '=', 0x30f3,
+ '=', 0xff9d, 0
+};
+
static UStringSearch* searcher()
{
static UStringSearch* searcher = createSearcher();
+ static UCollator* collator = 0;
+ if (!collator) {
+ // Set tailored collation rules to fix Japanese text search.
+ // See the comments before JAPANESE_KANA_COLLATION_RULES for details.
+ UErrorCode status = U_ZERO_ERROR;
+ collator = ucol_openRules(JAPANESE_KANA_COLLATION_RULES, -1, UCOL_DEFAULT,
+ UCOL_DEFAULT_STRENGTH, 0, &status);
+ ASSERT(status == U_ZERO_ERROR);
+ status = U_ZERO_ERROR;
+ usearch_setCollator(searcher, collator, &status);
+ ASSERT(status == U_ZERO_ERROR);
+ usearch_reset(searcher);
+ }
return searcher;
}
