ICU-22511 Fixed an infinite loop problem caused by backward collation…

icu4c/source/i18n/utf16collationiterator.cpp

@@ -295,9 +295,21 @@ FCDUTF16CollationIterator::previousCodePoint(UErrorCode &errorCode) {
                 return U_SENTINEL;
             }
             c = *--pos;
-            if(CollationFCD::hasLccc(c)) {
-                if(CollationFCD::maybeTibetanCompositeVowel(c) ||
-                        (pos != start && CollationFCD::hasTccc(*(pos - 1)))) {
+            if (CollationFCD::hasLccc(c)) {
+                UBool normalizePrevSegment = CollationFCD::maybeTibetanCompositeVowel(c);
+                if (!normalizePrevSegment && pos != start) {
+                    if (U16_IS_TRAIL(*(pos - 1))) {
+                        // CollationFCD::hasTccc(<low-surrogate>) always returns false.
+                        // To test possible trailing ccc, we need to check high surrogate
+                        // (or previous character for broken surrogate pair).
+                        if (pos - 1 != start) {
+                            normalizePrevSegment = CollationFCD::hasTccc(*(pos - 2));
+                        }
+                    } else {
+                        normalizePrevSegment = CollationFCD::hasTccc(*(pos - 1));
+                    }
+                }
+                if (normalizePrevSegment) {
                     ++pos;
                     if(!previousSegment(errorCode)) {
                         return U_SENTINEL;

icu4c/source/test/intltest/collationtest.cpp

@@ -90,6 +90,7 @@ class CollationTest : public IntlTest {
     void TestCollatorPredicateTypes();
     void TestUCollatorPredicateTypes();
     void TestCollatorMap();
+    void TestColItrInfiniteLoop22511();
 
 private:
     void checkFCD(const char *name, CollationIterator &ci, CodePointIterator &cpi);
@@ -168,6 +169,7 @@ void CollationTest::runIndexedTest(int32_t index, UBool exec, const char *&name,
     TESTCASE_AUTO(TestCollatorPredicateTypes);
     TESTCASE_AUTO(TestUCollatorPredicateTypes);
     TESTCASE_AUTO(TestCollatorMap);
+    TESTCASE_AUTO(TestColItrInfiniteLoop22511);
     TESTCASE_AUTO_END;
 }
 
@@ -1912,6 +1914,7 @@ void CollationTest::TestHang22414() {
         errorCode.reset();
     }
 }
+
 void CollationTest::TestBuilderContextsOverflow() {
     IcuTestErrorCode errorCode(*this, "TestBuilderContextsOverflow");
     // ICU-20715: Bad memory access in what looks like a bogus CharsTrie after
@@ -2123,4 +2126,30 @@ void CollationTest::TestCollatorMap() {
     assertEquals(R"(u16m["a"])", 2, um[u"a"]);
 }
 
+void CollationTest::TestColItrInfiniteLoop22511() {
+    IcuTestErrorCode errorCode(*this, "TestColItrInfiniteLoop22511");
+    char16_t str1[] = {
+        0x0100, 0x032a, 0x01e0, 0xd804, 0xdd00, 0x031c
+    };
+    char16_t str2[] = {
+        0x0041, 0x0304, 0x032a, 0x01e0, 0xd804, 0xdd00, 0x031c
+    };
+    int32_t num_locales = 0;
+    const icu::Locale* locales = icu::Locale::getAvailableLocales(num_locales);
+    for (int32_t i = 0; i < num_locales; i++) {
+        errorCode.reset();
+        icu::Locale l = locales[i];
+        LocalPointer<Collator> coll(Collator::createInstance(l, errorCode));
+        errorCode.assertSuccess();
+        coll->setStrength(icu::Collator::IDENTICAL);
+        UCollationResult result = coll->compare(
+            str1, sizeof(str1)/sizeof(char16_t),
+            str2, sizeof(str2)/sizeof(char16_t),
+            errorCode);
+        errorCode.assertSuccess();
+        assertEquals(UnicodeString("Locale ") + l.getName(), UCOL_EQUAL, result);
+    }
+}
+
+
 #endif  // !UCONFIG_NO_COLLATION

icu4j/main/collate/src/main/java/com/ibm/icu/impl/coll/FCDUTF16CollationIterator.java

@@ -137,8 +137,20 @@ public int previousCodePoint() {
                 }
                 c = seq.charAt(--pos);
                 if (CollationFCD.hasLccc(c)) {
-                    if (CollationFCD.maybeTibetanCompositeVowel(c)
-                            || (pos != start && CollationFCD.hasTccc(seq.charAt(pos - 1)))) {
+                    boolean normalizePrevSegment = CollationFCD.maybeTibetanCompositeVowel(c);
+                    if (!normalizePrevSegment && pos != start) {
+                        if (Character.isLowSurrogate(seq.charAt(pos - 1))) {
+                            // CollationFCD.hasTccc(<low-surrogate>) always returns false.
+                            // To test possible trailing ccc, we need to check high surrogate
+                            // (or previous character for broken surrogate pair).
+                            if (pos - 1 != start) {
+                                normalizePrevSegment = CollationFCD.hasTccc(seq.charAt(pos - 2));
+                            }
+                        } else {
+                            normalizePrevSegment = CollationFCD.hasTccc(seq.charAt(pos - 1));
+                        }
+                    }
+                    if (normalizePrevSegment) {
                         ++pos;
                         previousSegment();
                         c = seq.charAt(--pos);

icu4j/main/collate/src/test/java/com/ibm/icu/dev/test/collator/CollationTest.java

@@ -1960,4 +1960,20 @@ public void TestBuilderContextsOverflow() {
             errln("unexpected type of exception: " + e);
         }
     }
+
+    @Test
+    public void TestColItrInfiniteLoop22511() {
+        // ICU-22511 Locale vi and wo triggers infinite loop for getting
+        // collation key for these strings.
+        final String str1 = "\u0100\u032a\u01e0\ud804\udd00\u031c";
+        final String str2 = "\u0041\u0304\u032a\u01e0\ud804\udd00\u031c";
+
+        ULocale[] locales = ULocale.getAvailableLocales();
+        for (ULocale loc : locales) {
+            Collator coll = Collator.getInstance(loc);
+            coll.setStrength(Collator.IDENTICAL);
+            int cmp = coll.compare(str1, str2);
+            assertEquals("Locale " + loc.toString(), 0, cmp);
+        }
+    }
 }