Translating to Japanese

Mercury13 · Dec 19, 2024 · c6437be · c6437be
1 parent 1e5d540
commit c6437be
Show file tree

Hide file tree

Showing 7 changed files with 85 additions and 25 deletions.
diff --git a/Lang-src/en/locale.xml b/Lang-src/en/locale.xml
@@ -18,12 +18,16 @@
 	</trigger-langs>
 	<!-- Alphabetical sorting, right now very simple and locale-independent
 		 non-meaning stub of alphabet = underscore _
+		 alp follow="1" (see Russian) — this alphabet is not parallel;
+		   instead it will be greater than everything we have defined
+		   (to split long strings, reduce # of underscores, better sync long alphabets)
 		 Limitations:
 		 • Difference within first 40 MEANING chars of term/block, otherwise they are deemed equal
 		 • Strings that differ with “equal” character only are forbidden:
 		    äbb < abc < äbd is supported, but abc < äbc isn’t.
 		 • 32-bit chars ALLOWED
-		 • Max alphabet 32K
+		 • Sum of all alphabets following each other 32K
+		 • Alphabet with follow=1 MAY contain text (though better not for beauty)
 		 • # if alphabets unlimited -->
 	<alpha-sort>
 		<alp>ABCDEFGHIJKLMNOPQRSTUVWXYZ</alp>

diff --git a/Lang-src/ja/locale.xml b/Lang-src/ja/locale.xml
@@ -3,22 +3,59 @@
 	<trigger-langs>
 		<lang iso="ja" />
 	</trigger-langs>
-	<!-- Alphabetical sorting, right now very simple and locale-independent
-		 non-meaning stub of alphabet = underscore _
-		 Limitations:
-		 • Difference within first 40 MEANING chars of term/block, otherwise they are deemed equal
-		 • Strings that differ with “equal” character only are forbidden:
-		    äbb < abc < äbd is supported, but abc < äbc isn’t.
-		 • 32-bit chars ALLOWED
-		 • Max alphabet 32K
-		 • # if alphabets unlimited -->
 	<alpha-sort>
+		<alp>あいうえぉお</alp><!-- Vowel -->
+		<alp>アイウエォオ</alp>
+		<alp follow="1" /><!-- K/G -->
+		<alp>かがきぎくぐけげこご</alp>
+		<alp>カガキギクグケゲコゴ</alp>
+		<alp follow="1" /><!-- S/Z -->
+		<alp>さざしじすずせぜそぞ</alp>
+		<alp>サザシジスズセゼソゾ</alp>
+		<alp follow="1" /><!-- T/D -->
+		<alp>っただちぢつづてでとど</alp>
+		<alp>ッタダチヂツヅテデトド</alp>
+		<alp follow="1" /><!-- N -->
+		<alp>なにぬねの</alp>
+		<alp>ナニヌネノ</alp>
+		<alp follow="1" /><!-- H/B/P -->
+		<alp>はばぱひびぴふぶぷへべぺほぼぽ</alp>
+		<alp>ハバパヒビピフブプヘベペホボポ</alp>
+		<alp follow="1" /><!-- M -->
+		<alp>まみむめも</alp>
+		<alp>マミムメモ</alp>
+		<alp follow="1" /><!-- Y -->
+		<alp>ゃやゆよ</alp>
+		<alp>ャヤユヨ</alp>
+		<alp follow="1" /><!-- R -->
+		<alp>らラりリるルれレろロ</alp>
+		<alp follow="1" /><!-- W -->
+		<alp>ゎわゐゑを</alp>
+		<alp>ヮワヰヱヲ</alp>
+		<alp follow="1" /><!-- Consonant/V -->
+		<alp>ん__ゔ</alp>
+		<alp>ンヷヸヴヹヺ</alp>
 		<alp>ABCDEFGHIJKLMNOPQRSTUVWXYZ</alp>
 		<alp>abcdefghijklmnopqrstuvwxyz</alp>
-		<alp>____________________Ü_____</alp>
-		<alp>____________________ü_____</alp>
 	</alpha-sort>
-	<alpha-fixup />
+	<alpha-fixup>
+		<!-- For Japanese: they are just conversions to kana -->
+		<blk start="1980" target="しん たい゠ ろ もじ (ふぃんらんどご)" /><!-- Talu -->
+		<blk start="1D00" target="おんせい きごう かくちょう" />
+		<blk start="1D80" target="おんせい きごう かくちょう ほそく" />
+		<blk start="20A0" target="つうか きごう" />
+		<blk start="2100" target="もじ さま きごう" />
+		<blk start="2150" target="すうじ けいしき" />
+		<blk start="2190" target="やじるし" /><!-- the rest arrows will automatically pull -->
+		<blk start="2200" target="すうがく きごう" /><!-- the rest math will automatically pull -->
+		<blk start="2F00" target="かん ぶしゅ" /><!-- Kangxi, the 2nd will automatically pull -->
+		<blk start="3100" target="ちゅう おと じぼ (たま かげ)" /><!-- Bopo -->
+		<blk start="103A0" target="こだい ぺるしゃ もじ" /><!-- Xpeo -->
+		<blk start="10840" target="ていこく あらむ もじ" /><!-- Imperial Aramaic -->
+		<blk start="10C00" target="もじ (ほご-じぇねしす) (るーんご)" /><!-- Orkh -->
+		<blk start="10F70" target="こだい ういぐる もじ" /><!-- Ougr -->
+		<blk start="11580" target="しったん もじ (ぼんじ)" /><!-- Sidd -->
+	</alpha-fixup>
 	<!-- For discoverability some auxiliary blocks are placed next to “parent” ones,
 		 breaking alphabet order and referenceability.
 		 These blocks are marked with ellipsis. -->

diff --git a/Lang-src/ru/locale.xml b/Lang-src/ru/locale.xml
@@ -16,9 +16,12 @@
 	</trigger-langs>
 	<alpha-sort>
 		<!-- Dash is for Yi: «И — слоги» < «Иврит» -->
-		<alp>—АБВГДЕЁЖЗИЙКЛМНОПРСТУФХЦЧШЩЪЫЬЭЮЯABCDEFGHIJKLMNOPQRSTUVWXYZ</alp>
-		<alp>_абвгдеёжзийклмнопрстуфхцчшщъыьэюяabcdefghijklmnopqrstuvwxyz</alp>
-		<alp>_á____é_________ó___________________________________________</alp>
+		<alp>—АБВГДЕЁЖЗИЙКЛМНОПРСТУФХЦЧШЩЪЫЬЭЮЯ</alp>
+		<alp>_абвгдеёжзийклмнопрстуфхцчшщъыьэюя</alp>
+		<alp>_á____é_________ó_________________</alp>
+		<alp follow="1" />
+		<alp>ABCDEFGHIJKLMNOPQRSTUVWXYZ</alp>
+		<alp>abcdefghijklmnopqrstuvwxyz</alp>
 	</alpha-sort>
 	<alpha-fixup />
 	<ellipsis text="…">

diff --git a/Lang-src/uk/locale.xml b/Lang-src/uk/locale.xml
@@ -4,10 +4,13 @@
 		<lang iso="uk" />
 	</trigger-langs>
 	<alpha-sort>
-		<alp>АБВГДЕЄЖЗИІЇЙКЛМНОПРСТУФХЦЧШЩЬЮЯABCDEFGHIJKLMNOPQRSTUVWXYZ</alp>
-		<alp>абвгдеєжзиіїйклмнопрстуфхцчшщьюяabcdefghijklmnopqrstuvwxyz</alp>
-		<alp>á__ґ_é____í______ó________________________________________</alp>
-		<alp>___Ґ______________________________________________________</alp>
+		<alp>АБВГДЕЄЖЗИІЇЙКЛМНОПРСТУФХЦЧШЩЬЮЯ</alp>
+		<alp>абвгдеєжзиіїйклмнопрстуфхцчшщьюя</alp>
+		<alp>á__ґ_é____í______ó______________</alp>
+		<alp>___Ґ____________________________</alp>
+		<alp follow="1" />
+		<alp>ABCDEFGHIJKLMNOPQRSTUVWXYZ</alp>
+		<alp>abcdefghijklmnopqrstuvwxyz</alp>
 	</alpha-sort>
 	<alpha-fixup />
 	<ellipsis text="…">

diff --git a/Unicodia/LocList.cpp b/Unicodia/LocList.cpp
@@ -180,11 +180,18 @@ namespace {
 
         r.sortOrder.clear();
         auto hAlphaSort = hLocale.child("alpha-sort");
+        int base = 0, greatest = -1;
         for (auto& v : hAlphaSort.children("alp")) {
+            if (v.attribute("follow").as_bool(false)) {
+                base = greatest + 1;
+            }
             auto alph = mojibake::toS<std::u32string>(v.text().as_string());
             for (size_t i = 0; i < alph.length(); ++i) {
-                if (auto c = alph[i]; c != '_')
-                    r.sortOrder[c] = i;
+                if (auto c = alph[i]; c != '_') {
+                    int newCode = base + i;
+                    r.sortOrder[c] = newCode;
+                    greatest = std::max(greatest, newCode);
+                }
             }
         }
 

diff --git a/Unicodia/Uc/UcData.cpp b/Unicodia/Uc/UcData.cpp
@@ -1636,10 +1636,16 @@ void uc::finishTranslation(
             std::u8string_view keyName = blk.loc.name;
             if (blk.alphaKey.ecScript != EcScript::NONE) {
                 keyName = script.loc.name;
-                if (script.mainBlock)
+                if (script.mainBlock) {
+                    if (auto q = alphaFixup.find(script.mainBlock->startingCp); q != alphaFixup.end()) {
+                        buildSortKey(q->second, SpecialSort::NO, sortOrder, blk.loc.sortKey);
+                        goto alreadyBuilt;
+                    }
                     keyName = script.mainBlock->loc.name;
+                }
             }
             buildSortKey(keyName, SpecialSort::NO, sortOrder, blk.loc.sortKey);
+        alreadyBuilt: ;
         }
     }
 

diff --git a/Unicodia/Uc/UcData.h b/Unicodia/Uc/UcData.h
@@ -766,10 +766,10 @@ namespace uc {
 
     DEFINE_ENUM_OPS(Bfg)
 
-    enum class MyName { INST };
+    enum class MyName : unsigned char { INST };
 
     ///
-    /// \brief The BlockAlpha struct
+    /// Use someone’s else name for sorting
     ///
     struct AlphaKey {
         EcScript ecScript = EcScript::NONE;