ICU-22549 Improve fuzzer to test more locale

We found bogus locale cause crash in DateFormat so here we enhance the fuzzer to also test locale name which are not return by the available locale list.
unicode-org · Dec 18, 2023 · 11d1148 · 11d1148
1 parent 4ba5d91
commit 11d1148
Show file tree

Hide file tree

Showing 4 changed files with 54 additions and 22 deletions.
diff --git a/icu4c/source/test/fuzzer/break_iterator_fuzzer.cpp b/icu4c/source/test/fuzzer/break_iterator_fuzzer.cpp
@@ -66,6 +66,26 @@ extern "C" int LLVMFuzzerTestOneInput(const uint8_t* data, size_t size) {
   for (int32_t p = bi->first(); p != icu::BreakIterator::DONE; p = bi->next()) {}
 
   utext_close(fuzzstr);
+
+  icu::Locale locale2(reinterpret_cast<const char*>(data));
+  switch (rnd8 % 5) {
+    case 0:
+      bi.reset(icu::BreakIterator::createWordInstance(locale2, status));
+      break;
+    case 1:
+      bi.reset(icu::BreakIterator::createLineInstance(locale2, status));
+      break;
+    case 2:
+      bi.reset(icu::BreakIterator::createCharacterInstance(locale2, status));
+      break;
+    case 3:
+      bi.reset(icu::BreakIterator::createSentenceInstance(locale2, status));
+      break;
+    case 4:
+      bi.reset(icu::BreakIterator::createTitleInstance(locale2, status));
+      break;
+  }
+
   return 0;
 }
 
diff --git a/icu4c/source/test/fuzzer/collator_compare_fuzzer.cpp b/icu4c/source/test/fuzzer/collator_compare_fuzzer.cpp
@@ -37,22 +37,25 @@ extern "C" int LLVMFuzzerTestOneInput(const uint8_t* data, size_t size) {
   if (size > 4096) {
       size = 4096;
   }
-  std::unique_ptr<char16_t> compbuff1(new char16_t[size/4]);
+  std::unique_ptr<char16_t[]> compbuff1(new char16_t[size/4]);
   std::memcpy(compbuff1.get(), data, (size/4)*2);
   data = data + size/2;
-  std::unique_ptr<char16_t> compbuff2(new char16_t[size/4]);
+  std::unique_ptr<char16_t[]> compbuff2(new char16_t[size/4]);
   std::memcpy(compbuff2.get(), data, (size/4)*2);
 
 
   icu::LocalPointer<icu::Collator> fuzzCollator(
       icu::Collator::createInstance(locale, status), status);
-  if (U_FAILURE(status))
-    return 0;
+  if (U_SUCCESS(status)) {
 
-  fuzzCollator->setStrength(strength);
+    fuzzCollator->setStrength(strength);
 
-  fuzzCollator->compare(compbuff1.get(), size/4,
-                        compbuff2.get(), size/4);
+    fuzzCollator->compare(compbuff1.get(), size/4,
+                          compbuff2.get(), size/4);
+  }
+  status = U_ZERO_ERROR;
 
+  fuzzCollator.adoptInstead(
+      icu::Collator::createInstance(icu::Locale(reinterpret_cast<const char*>(data)), status));
   return 0;
 }
diff --git a/icu4c/source/test/fuzzer/date_format_fuzzer.cpp b/icu4c/source/test/fuzzer/date_format_fuzzer.cpp
@@ -83,5 +83,11 @@ extern "C" int LLVMFuzzerTestOneInput(const uint8_t* data, size_t size) {
     if (U_SUCCESS(status)) {
         df->format(date, appendTo);
     }
+
+    icu::Locale locale2(fuzzData.data());
+    df.reset(
+        icu::DateFormat::createDateTimeInstance(dateStyle, timeStyle, locale2));
+    df.reset(
+        icu::DateFormat::createDateTimeInstance(dateStyle2, timeStyle2, locale2));
     return EXIT_SUCCESS;
 }
diff --git a/icu4c/source/test/fuzzer/date_time_pattern_generator_fuzzer.cpp b/icu4c/source/test/fuzzer/date_time_pattern_generator_fuzzer.cpp
@@ -29,28 +29,31 @@ extern "C" int LLVMFuzzerTestOneInput(const uint8_t* data, size_t size) {
 
   icu::LocalPointer<icu::DateTimePatternGenerator > gen(
       icu::DateTimePatternGenerator::createInstance(locale, status), status);
-  if (U_FAILURE(status))
-    return 0;
+  if (U_SUCCESS(status)) {
 
-  status = U_ZERO_ERROR;
-  gen->getSkeleton(fuzzstr, status);
+      status = U_ZERO_ERROR;
+      gen->getSkeleton(fuzzstr, status);
 
-  status = U_ZERO_ERROR;
-  gen->getBaseSkeleton(fuzzstr, status);
+      status = U_ZERO_ERROR;
+      gen->getBaseSkeleton(fuzzstr, status);
 
-  status = U_ZERO_ERROR;
-  gen->getBaseSkeleton(fuzzstr, status);
+      status = U_ZERO_ERROR;
+      gen->getBaseSkeleton(fuzzstr, status);
 
-  status = U_ZERO_ERROR;
-  gen->getPatternForSkeleton(fuzzstr);
+      status = U_ZERO_ERROR;
+      gen->getPatternForSkeleton(fuzzstr);
 
-  status = U_ZERO_ERROR;
-  gen->getBestPattern(fuzzstr, status);
+      status = U_ZERO_ERROR;
+      gen->getBestPattern(fuzzstr, status);
 
-  status = U_ZERO_ERROR;
-  icu::DateTimePatternGenerator::staticGetSkeleton(fuzzstr, status);
+      status = U_ZERO_ERROR;
+      icu::DateTimePatternGenerator::staticGetSkeleton(fuzzstr, status);
+
+      status = U_ZERO_ERROR;
+      icu::DateTimePatternGenerator::staticGetBaseSkeleton (fuzzstr, status);
+  }
 
   status = U_ZERO_ERROR;
-  icu::DateTimePatternGenerator::staticGetBaseSkeleton (fuzzstr, status);
+  gen.adoptInstead(icu::DateTimePatternGenerator::createInstance(icu::Locale(reinterpret_cast<const char*>(data)), status));
   return 0;
 }