ICU-22641 Change const refs in the class to const pointers

The function may return nullptr, we cannot just deref the return value. Change them to const pointer instead.
unicode-org · Jan 23, 2024 · f7f9dbb · f7f9dbb
1 parent e81b872
commit f7f9dbb
Show file tree

Hide file tree

Showing 4 changed files with 21 additions and 11 deletions.
diff --git a/icu4c/source/common/caniter.cpp b/icu4c/source/common/caniter.cpp
@@ -64,6 +64,7 @@ U_NAMESPACE_BEGIN
 
 UOBJECT_DEFINE_RTTI_IMPLEMENTATION(CanonicalIterator)
 
+
 /**
  *@param source string to get results for
  */
@@ -73,10 +74,10 @@ CanonicalIterator::CanonicalIterator(const UnicodeString &sourceStr, UErrorCode
     pieces_lengths(nullptr),
     current(nullptr),
     current_length(0),
-    nfd(*Normalizer2::getNFDInstance(status)),
-    nfcImpl(*Normalizer2Factory::getNFCImpl(status))
+    nfd(Normalizer2::getNFDInstance(status)),
+    nfcImpl(Normalizer2Factory::getNFCImpl(status))
 {
-    if(U_SUCCESS(status) && nfcImpl.ensureCanonIterData(status)) {
+    if(U_SUCCESS(status) && nfcImpl->ensureCanonIterData(status)) {
       setSource(sourceStr, status);
     }
 }
@@ -172,7 +173,7 @@ void CanonicalIterator::setSource(const UnicodeString &newSource, UErrorCode &st
     int32_t i = 0;
     UnicodeString *list = nullptr;
 
-    nfd.normalize(newSource, source, status);
+    nfd->normalize(newSource, source, status);
     if(U_FAILURE(status)) {
       return;
     }
@@ -219,7 +220,7 @@ void CanonicalIterator::setSource(const UnicodeString &newSource, UErrorCode &st
     // on the NFD form - see above).
     for (; i < source.length(); i += U16_LENGTH(cp)) {
         cp = source.char32At(i);
-        if (nfcImpl.isCanonSegmentStarter(cp)) {
+        if (nfcImpl->isCanonSegmentStarter(cp)) {
             source.extract(start, i-start, list[list_length++]); // add up to i
             start = i;
         }
@@ -390,7 +391,7 @@ UnicodeString* CanonicalIterator::getEquivalents(const UnicodeString &segment, i
             //UnicodeString *possible = new UnicodeString(*((UnicodeString *)(ne2->value.pointer)));
             UnicodeString possible(*((UnicodeString *)(ne2->value.pointer)));
             UnicodeString attempt;
-            nfd.normalize(possible, attempt, status);
+            nfd->normalize(possible, attempt, status);
 
             // TODO: check if operator == is semanticaly the same as attempt.equals(segment)
             if (attempt==segment) {
@@ -457,7 +458,7 @@ Hashtable *CanonicalIterator::getEquivalents2(Hashtable *fillinResult, const cha
     for (int32_t i = 0; i < segLen; i += U16_LENGTH(cp)) {
         // see if any character is at the start of some decomposition
         U16_GET(segment, 0, i, segLen, cp);
-        if (!nfcImpl.getCanonStartSet(cp, starts)) {
+        if (!nfcImpl->getCanonStartSet(cp, starts)) {
             continue;
         }
         // if so, see which decompositions match
@@ -518,7 +519,7 @@ Hashtable *CanonicalIterator::extract(Hashtable *fillinResult, UChar32 comp, con
     UnicodeString temp(comp);
     int32_t inputLen=temp.length();
     UnicodeString decompString;
-    nfd.normalize(temp, decompString, status);
+    nfd->normalize(temp, decompString, status);
     if (U_FAILURE(status)) {
         return nullptr;
     }
@@ -582,7 +583,7 @@ Hashtable *CanonicalIterator::extract(Hashtable *fillinResult, UChar32 comp, con
     // brute force approach
     // check to make sure result is canonically equivalent
     UnicodeString trial;
-    nfd.normalize(temp, trial, status);
+    nfd->normalize(temp, trial, status);
     if(U_FAILURE(status) || trial.compare(segment+segmentPos, segLen - segmentPos) != 0) {
         return nullptr;
     }

diff --git a/icu4c/source/common/unicode/caniter.h b/icu4c/source/common/unicode/caniter.h
@@ -183,8 +183,8 @@ class U_COMMON_API CanonicalIterator final : public UObject {
     // transient fields
     UnicodeString buffer;
 
-    const Normalizer2 &nfd;
-    const Normalizer2Impl &nfcImpl;
+    const Normalizer2 *nfd;
+    const Normalizer2Impl *nfcImpl;
 
     // we have a segment, in NFD. Find all the strings that are canonically equivalent to it.
     UnicodeString *getEquivalents(const UnicodeString &segment, int32_t &result_len, UErrorCode &status); //private String[] getEquivalents(String segment)

diff --git a/icu4c/source/test/intltest/apicoll.cpp b/icu4c/source/test/intltest/apicoll.cpp
@@ -2517,6 +2517,13 @@ void CollationAPITest::TestGapTooSmall() {
     }
 }
 
+void CollationAPITest::TestNFCNull() {
+    IcuTestErrorCode errorCode(*this, "TestNFCNull");
+    RuleBasedCollator coll(
+       u"&†<†\u0f81\u0f81\u0f81\u0f81\u0f81|†", errorCode);
+    errorCode.expectErrorAndReset(U_UNSUPPORTED_ERROR);
+}
+
  void CollationAPITest::dump(UnicodeString msg, RuleBasedCollator* c, UErrorCode& status) {
     const char* bigone = "One";
     const char* littleone = "one";
@@ -2560,6 +2567,7 @@ void CollationAPITest::runIndexedTest( int32_t index, UBool exec, const char* &n
     TESTCASE_AUTO(TestIterNumeric);
     TESTCASE_AUTO(TestBadKeywords);
     TESTCASE_AUTO(TestGapTooSmall);
+    TESTCASE_AUTO(TestNFCNull);
     TESTCASE_AUTO_END;
 }
 

diff --git a/icu4c/source/test/intltest/apicoll.h b/icu4c/source/test/intltest/apicoll.h
@@ -174,6 +174,7 @@ class CollationAPITest: public IntlTestCollator {
     void TestIterNumeric();
     void TestBadKeywords();
     void TestGapTooSmall();
+    void TestNFCNull();
 
 private:
     // If this is too small for the test data, just increase it.