ICU-21699 Concatenate Katakana chars

See #1962

diff --git a/icu4c/source/common/dictbe.cpp b/icu4c/source/common/dictbe.cpp
index 64b4fbf63912e8ff7b1e45230cdfd6c04dda6e2d..c0af19ef171300601f5786641ebc45d658ea972e 100644 (file)
--- a/icu4c/source/common/dictbe.cpp
+++ b/icu4c/source/common/dictbe.cpp
@@ -1366,8 +1366,11 @@ CjkBreakEngine::divideUpDictionaryRange( UText *inText,
                 // Calculate the length by using the code unit.
                 length = inString.moveIndex32(0, prevIdx) - codeUnitIdx;
                 prevIdx = i;
-                // Skip the breakpoint if it belongs to the particle or Hiragana.
-                if (!fSkipSet.containsKey(inString.tempSubString(codeUnitIdx, length))) {
+                // Keep the breakpoint if the pattern is not in the fSkipSet and continuous Katakana
+                // characters don't occur.
+                if (!fSkipSet.containsKey(inString.tempSubString(codeUnitIdx, length))
+                    && (!isKatakana(inString.char32At(codeUnitIdx -1))
+                           || !isKatakana(inString.char32At(codeUnitIdx)))) {
                     t_boundary.addElement(i, status);
                     numBreaks++;
                 }

diff --git a/icu4c/source/test/testdata/rbbitst.txt b/icu4c/source/test/testdata/rbbitst.txt
index 54c612da22c583ff76af9ed3fc1b220f874c8d53..1be45e9f3c5d2e86cd4cec337ca3253b26a14d40 100644 (file)
--- a/icu4c/source/test/testdata/rbbitst.txt
+++ b/icu4c/source/test/testdata/rbbitst.txt
@@ -1892,10 +1892,12 @@ Bangkok)•</data>
 <data>•\uff19\u6708\u306b•\u6771\u4eac\u304b\u3089•\u53cb\u9054\u304c•\u904a\u3073\u306b•\u6765\u305f•</data>
 #る文字「そうだ、京都」-> る•文字•「そうだ、•京都」•
 <data>•\u308b•\u6587\u5b57•\u300c\u305d\u3046\u3060\u3001•\u4eac\u90fd\u300d•</data>
-#乗車率９０％程度だろうか -> 乗車•率•９０％•程度だ•ろうか•
+#乗車率９０％程度だろうか。 -> 乗車•率•９０％•程度だ•ろうか。•
 <data>•\u4e57\u8eca•\u7387•\uff19\uff10\uff05•\u7a0b\u5ea6\u3060•\u308d\u3046\u304b\u3002•</data>
 #［携帯電話］正しい選択 -> ［携帯•電話］•正しい•選択•
 <data>•\uff3b\u643a\u5e2f•\u96fb\u8a71\uff3d•\u6b63\u3057\u3044•\u9078\u629e•</data>
+#純金製百人一首にサッカーボール -> 純金•製•百人一首に•サッカーボール
+<data>•\u7D14\u91D1•\u88FD•\u767E\u4EBA\u4E00\u9996\u306B•\u30B5\u30C3\u30AB\u30FC\u30DC\u30FC\u30EB•</data>
 
 
 ####################################################################################