ICU-13569 rbbi table compression, work in progress.

X-SVN-Rev: 40873

diff --git a/icu4c/source/common/rbbirb.cpp b/icu4c/source/common/rbbirb.cpp
index 817a955a9698b57b53e67e71b749a6c6e5ce1757..3becda31278a06fef462fe0bbd6f4b52d9af69b3 100644 (file)
--- a/icu4c/source/common/rbbirb.cpp
+++ b/icu4c/source/common/rbbirb.cpp
@@ -357,11 +357,17 @@ void RBBIRuleBuilder::optimizeTables() {
     int32_t leftClass;
     int32_t rightClass;
 
-    leftClass = 1;
-    rightClass = 2;
+    leftClass = 3;
+    rightClass = 4;
+    printf("Optimizing tables ...\n");
     while (fForwardTables->findDuplCharClassFrom(leftClass, rightClass)) {
+        printf("Merging duplicate columns (%d, %d)\n", leftClass, rightClass);
+        
         fSetBuilder->mergeCategories(leftClass, rightClass);
         fForwardTables->removeColumn(rightClass);
+        fReverseTables->removeColumn(rightClass);
+        fSafeFwdTables->removeColumn(rightClass);
+        fSafeRevTables->removeColumn(rightClass);
     }
 
 

diff --git a/icu4c/source/common/rbbitblb.cpp b/icu4c/source/common/rbbitblb.cpp
index a8bc0486199dad26c43db0126c66eb376dc2a2d4..b71921a9dee16d44adf482323291d7501a42aaef 100644 (file)
--- a/icu4c/source/common/rbbitblb.cpp
+++ b/icu4c/source/common/rbbitblb.cpp
@@ -1090,7 +1090,7 @@ bool RBBITableBuilder::findDuplCharClassFrom(int32_t &baseCategory, int32_t &dup
     uint16_t table_base;
     uint16_t table_dupl;
     for (; baseCategory < numCols-1; ++baseCategory) {
-        for (; duplCategory < numCols; ++duplCategory) {
+        for (duplCategory=baseCategory+1; duplCategory < numCols; ++duplCategory) {
              for (int32_t state=0; state<numStates; state++) {
                  RBBIStateDescriptor *sd = (RBBIStateDescriptor *)fDStates->elementAt(state);
                  table_base = (uint16_t)sd->fDtran->elementAti(baseCategory);

diff --git a/icu4c/source/test/intltest/rbbitst.cpp b/icu4c/source/test/intltest/rbbitst.cpp
index 2565ef4f61cfb6a8ea28c51b3811ee2cbf47a363..c6e0f457fbe10f8d36ce171af0885a336a934c61 100644 (file)
--- a/icu4c/source/test/intltest/rbbitst.cpp
+++ b/icu4c/source/test/intltest/rbbitst.cpp
@@ -4461,9 +4461,24 @@ void RBBITest::TestBug12677() {
 
 void RBBITest::TestTableRedundancies() {
     UErrorCode status = U_ZERO_ERROR;
+    
+    UnicodeString rules {u"$s0=[;,*]; \n"
+                "$s1=[a-z]; \n"
+                "$s2=[i-n]; \n"
+                "$s3=[x-z]; \n"
+                "!!forward; \n"
+                "($s0 | '?')*; \n"
+                "($s1 | $s2 | $s3)*; \n" };
+    RuleBasedBreakIterator *lbi = 
+        (RuleBasedBreakIterator *)BreakIterator::createLineInstance(Locale::getEnglish(), status);
+    rules = lbi->getRules();
+    UParseError pe {};
     RuleBasedBreakIterator *bi =
-            (RuleBasedBreakIterator *)BreakIterator::createLineInstance(Locale::getEnglish(), status);
-    // bi->dumpTables();
+    //         (RuleBasedBreakIterator *)BreakIterator::createLineInstance(Locale::getEnglish(), status);
+           new RuleBasedBreakIterator(rules, pe, status);
+    assertSuccess(WHERE, status);
+    if (U_FAILURE(status)) return;
+    bi->dumpTables();
 
     RBBIDataWrapper *dw = bi->fData;
     const RBBIStateTable *fwtbl = dw->fForwardTable;

diff --git a/icu4c/source/test/testdata/rbbitst.txt b/icu4c/source/test/testdata/rbbitst.txt
index 1450a98d7be848ece5a2626aedd51295f08cbb9e..0d4c1633b62e47a311c5b7bc193102d79102fb32 100644 (file)
--- a/icu4c/source/test/testdata/rbbitst.txt
+++ b/icu4c/source/test/testdata/rbbitst.txt
@@ -39,18 +39,16 @@
 
 #   Temp debugging tests
 <locale en>
-<word>
-<data><0>コンピューター<400>は<400>、<0>本質<400>的<400>に<400>は<400>数字<400>しか<400>扱う<400>こと<400>が<400>でき<400>ま<400>せん<400>。<0>\
-コンピューター<400>は<400>、<0>文字<400>や<400>記号<400>など<400>の<400>それぞれに<400>番号<400>を<400>割り振る<400>こと<400>によって<400>扱える<400>\
-よう<400>にし<400>ます<400>。<0>ユニ<400>コード<400>が<400>出来る<400>まで<400>は<400>、<0>これらの<400>番号<400>を<400>割り振る<400>仕組み<400>が<400>\
-何<400>百<400>種類<400>も<400>存在<400>しま<400>した<400>。<0>どの<400>一つ<400>を<400>とっても<400>、<0>十分<400>な<400>文字<400>を<400>含<400>\
-んで<400>は<400>いま<400>せん<400>で<400>した<400>。<0>例えば<400>、<0>欧州<400>連合<400>一つ<400>を<400>見<400>て<400>も<400>、<0>その<400>\
-すべて<400>の<400>言語<400>を<400>カバー<400>する<400>ため<400>に<400>は<400>、<0>いくつか<400>の<400>異なる<400>符号<400>化<400>の<400>仕組み<400>\
-が<400>必要<400>で<400>した<400>。<0>英語<400>の<400>よう<400>な<400>一つ<400>の<400>言語<400>に<400>限<400>って<400>も<400>、<0>一つ<400>だけ<400>\
-の<400>符号<400>化<400>の<400>仕組み<400>では<400>、<0>一般<400>的<400>に<400>使<400>われる<400>すべて<400>の<400>文字<400>、<0>句読点<400>、<0>\
-。<0></data>
-
-#<data><0>コンピューター<400>は<400>、<0>本質<400>的<400>に<400>は<400>数字<400>しか<400>扱う<400>こと<400>が<400>でき<400>ま<400>せん<400>。<0>\
+<rules>
+$s0=[;,*];
+$s1=[a-z];
+$s2=[i-n];
+$s3=[x-z];
+!!forward;
+($s0 | '?')*
+($s1 | $s2 | $s3)*;
+</rules>
+<data>•hello• •</data>
 
 ## FILTERED BREAK TESTS