Function TEST

cpp/src/arrow/util/utf8_util_test.cc:319–347

}

TEST(SkipUTF8BOM, Basics) {
  auto CheckOk = [](const std::string& s, size_t expected_offset) -> void {
    const uint8_t* data = reinterpret_cast<const uint8_t*>(s.data());
    const uint8_t* res;
    ASSERT_OK_AND_ASSIGN(res, SkipUTF8BOM(data, static_cast<int64_t>(s.size())));
    ASSERT_NE(res, nullptr);
    ASSERT_EQ(res - data, expected_offset);
  };

  auto CheckTruncated = [](const std::string& s) -> void {
    const uint8_t* data = reinterpret_cast<const uint8_t*>(s.data());
    ASSERT_RAISES(Invalid, SkipUTF8BOM(data, static_cast<int64_t>(s.size())));
  };

  CheckOk("", 0);
  CheckOk("a", 0);
  CheckOk("ab", 0);
  CheckOk("abc", 0);
  CheckOk("abcd", 0);
  CheckOk("\xc3\xa9", 0);
  CheckOk("\xee", 0);
  CheckOk("\xef\xbc", 0);
  CheckOk("\xef\xbb\xbe", 0);
  CheckOk("\xef\xbb\xbf", 3);
  CheckOk("\xef\xbb\xbfx", 3);

  CheckTruncated("\xef");
  CheckTruncated("\xef\xbb");
}

// Currently a known issue with Valgrind and wide string AVX2 instructions
// https://sourceware.org/bugzilla/show_bug.cgi?id=22954