Merge pull request #424 from pet74alex/Support_UInt128

alexey-milovidov · web-flow · commit 9c3c85786af8 · 2025-06-02T10:11:28.000-07:00
Support UInt128 type
diff --git a/README.md b/README.md
@@ -19,7 +19,7 @@ C++ client for [ClickHouse](https://clickhouse.com/).
 * LowCardinality(String) or LowCardinality(FixedString(N))
 * Tuple
 * UInt8, UInt16, UInt32, UInt64, Int8, Int16, Int32, Int64
-* Int128
+* UInt128, Int128
 * UUID
 * Map
 * Point, Ring, Polygon, MultiPolygon
diff --git a/clickhouse/base/uuid.h b/clickhouse/base/uuid.h
@@ -5,8 +5,6 @@
 
 namespace clickhouse {
 
-using UInt128 = std::pair<uint64_t, uint64_t>;
-
-using UUID = UInt128;
+using UUID = std::pair<uint64_t, uint64_t>;
 
 }
diff --git a/clickhouse/columns/factory.cpp b/clickhouse/columns/factory.cpp
@@ -67,6 +67,8 @@ static ColumnRef CreateTerminalColumn(const TypeAst& ast) {
         return std::make_shared<ColumnInt64>();
     case Type::Int128:
         return std::make_shared<ColumnInt128>();
+    case Type::UInt128:
+        return std::make_shared<ColumnUInt128>();
 
     case Type::Float32:
         return std::make_shared<ColumnFloat32>();
diff --git a/clickhouse/columns/itemview.cpp b/clickhouse/columns/itemview.cpp
@@ -82,6 +82,7 @@ void ItemView::ValidateData(Type::Code type, DataType data) {
 
         case Type::Code::IPv6:
         case Type::Code::UUID:
+        case Type::Code::UInt128:
         case Type::Code::Int128:
         case Type::Code::Decimal128:
             return AssertSize({16});
diff --git a/clickhouse/columns/itemview.h b/clickhouse/columns/itemview.h
@@ -28,7 +28,7 @@ struct ItemView {
     inline auto ConvertToStorageValue(const T& t) {
         if constexpr (std::is_same_v<std::string_view, T> || std::is_same_v<std::string, T>) {
             return std::string_view{t};
-        } else if constexpr (std::is_fundamental_v<T> || std::is_same_v<Int128, std::decay_t<T>>) {
+        } else if constexpr (std::is_fundamental_v<T> || std::is_same_v<Int128, std::decay_t<T>> || std::is_same_v<UInt128, std::decay_t<T>>) {
             return std::string_view{reinterpret_cast<const char*>(&t), sizeof(T)};
         } else {
             static_assert(!std::is_same_v<T, T>, "Unknown type, which can't be stored in ItemView");
diff --git a/clickhouse/columns/numeric.cpp b/clickhouse/columns/numeric.cpp
@@ -118,6 +118,7 @@ template class ColumnVector<uint16_t>;
 template class ColumnVector<uint32_t>;
 template class ColumnVector<uint64_t>;
 template class ColumnVector<Int128>;
+template class ColumnVector<UInt128>;
 
 template class ColumnVector<float>;
 template class ColumnVector<double>;
diff --git a/clickhouse/columns/numeric.h b/clickhouse/columns/numeric.h
@@ -67,12 +67,14 @@ class ColumnVector : public Column {
 };
 
 using Int128 = absl::int128;
+using UInt128 = absl::uint128;
 using Int64 = int64_t;
 
 using ColumnUInt8   = ColumnVector<uint8_t>;
 using ColumnUInt16  = ColumnVector<uint16_t>;
 using ColumnUInt32  = ColumnVector<uint32_t>;
 using ColumnUInt64  = ColumnVector<uint64_t>;
+using ColumnUInt128 = ColumnVector<UInt128>;
 
 using ColumnInt8    = ColumnVector<int8_t>;
 using ColumnInt16   = ColumnVector<int16_t>;
diff --git a/clickhouse/types/type_parser.cpp b/clickhouse/types/type_parser.cpp
@@ -54,7 +54,7 @@ static const std::unordered_map<std::string, Type::Code> kTypeCode = {
     { "IPv4",        Type::IPv4 },
     { "IPv6",        Type::IPv6 },
     { "Int128",      Type::Int128 },
-//    { "UInt128",      Type::UInt128 },
+    { "UInt128",     Type::UInt128 },
     { "Decimal",     Type::Decimal },
     { "Decimal32",   Type::Decimal32 },
     { "Decimal64",   Type::Decimal64 },
diff --git a/clickhouse/types/types.cpp b/clickhouse/types/types.cpp
@@ -39,6 +39,7 @@ const char* Type::TypeName(Type::Code code) {
         case Type::Code::IPv4:           return "IPv4";
         case Type::Code::IPv6:           return "IPv6";
         case Type::Code::Int128:         return "Int128";
+        case Type::Code::UInt128:        return "UInt128";
         case Type::Code::Decimal:        return "Decimal";
         case Type::Code::Decimal32:      return "Decimal32";
         case Type::Code::Decimal64:      return "Decimal64";
@@ -68,6 +69,7 @@ std::string Type::GetName() const {
         case UInt16:
         case UInt32:
         case UInt64:
+        case UInt128:
         case UUID:
         case Float32:
         case Float64:
@@ -126,6 +128,7 @@ uint64_t Type::GetTypeUniqueId() const {
         case UInt16:
         case UInt32:
         case UInt64:
+        case UInt128:
         case UUID:
         case Float32:
         case Float64:
diff --git a/clickhouse/types/types.h b/clickhouse/types/types.h
@@ -12,6 +12,7 @@
 namespace clickhouse {
 
 using Int128 = absl::int128;
+using UInt128 = absl::uint128;
 using Int64 = int64_t;
 
 using TypeRef = std::shared_ptr<class Type>;
@@ -43,6 +44,7 @@ class Type {
         IPv4,
         IPv6,
         Int128,
+        UInt128,
         Decimal,
         Decimal32,
         Decimal64,
@@ -339,6 +341,11 @@ inline TypeRef Type::CreateSimple<Int128>() {
     return TypeRef(new Type(Int128));
 }
 
+template <>
+inline TypeRef Type::CreateSimple<UInt128>() {
+    return TypeRef(new Type(UInt128));
+}
+
 template <>
 inline TypeRef Type::CreateSimple<uint8_t>() {
     return TypeRef(new Type(UInt8));
diff --git a/ut/Column_ut.cpp b/ut/Column_ut.cpp
@@ -207,6 +207,7 @@ using TestCases = ::testing::Types<
     GenericColumnTestCase<ColumnIPv6, &makeColumn<ColumnIPv6>, in6_addr, &MakeIPv6s>,
 
     GenericColumnTestCase<ColumnInt128, &makeColumn<ColumnInt128>, clickhouse::Int128, &MakeInt128s>,
+    GenericColumnTestCase<ColumnUInt128, &makeColumn<ColumnUInt128>, clickhouse::UInt128, &MakeUInt128s>,
     GenericColumnTestCase<ColumnUUID, &makeColumn<ColumnUUID>, clickhouse::UUID, &MakeUUIDs>,
 
     DecimalColumnTestCase<ColumnDecimal, 18, 0>,
@@ -286,7 +287,7 @@ inline auto convertValueForGetItem(const ColumnType& col, ValueType&& t) {
         // Since ColumnDecimal can hold 32, 64, 128-bit wide data and there is no way telling at run-time.
         const ItemView item = col.GetItem(0);
         return std::string_view(reinterpret_cast<const char*>(&t), item.data.size());
-    } else if constexpr (std::is_same_v<T, clickhouse::UInt128>
+    } else if constexpr (std::is_same_v<T, clickhouse::UInt128> || std::is_same_v<T, clickhouse::UUID>
             || std::is_same_v<T, clickhouse::Int128>) {
         return std::string_view{reinterpret_cast<const char*>(&t), sizeof(T)};
     } else if constexpr (std::is_same_v<T, in_addr>) {
diff --git a/ut/CreateColumnByType_ut.cpp b/ut/CreateColumnByType_ut.cpp
@@ -76,7 +76,7 @@ INSTANTIATE_TEST_SUITE_P(Basic, CreateColumnByTypeWithName, ::testing::Values(
     "Int8", "Int16", "Int32", "Int64",
     "UInt8", "UInt16", "UInt32", "UInt64",
     "String", "Date", "DateTime",
-    "UUID", "Int128"
+    "UUID", "Int128", "UInt128"
 ));
 
 INSTANTIATE_TEST_SUITE_P(Parametrized, CreateColumnByTypeWithName, ::testing::Values(
diff --git a/ut/columns_ut.cpp b/ut/columns_ut.cpp
@@ -483,6 +483,28 @@ TEST(ColumnsCase, Int128) {
     EXPECT_EQ(0, col->At(4));
 }
 
+TEST(ColumnsCase, UInt128) {
+    auto col = std::make_shared<ColumnUInt128>(std::vector<UInt128>{
+            absl::MakeUint128(0xffffffffffffffffll, 0xffffffffffffffffll), // 2^128 - 1
+            absl::MakeUint128(0, 0xffffffffffffffffll),  // 2^64 - 1
+            absl::MakeUint128(0xffffffffffffffffll, 0),  // 2^128 - 2^64
+            absl::MakeUint128(0x8000000000000000ll, 0),
+            UInt128(0)
+    });
+
+    EXPECT_EQ(absl::MakeUint128(0xffffffffffffffffll, 0xffffffffffffffffll), col->At(0));
+
+    EXPECT_EQ(absl::MakeUint128(0, 0xffffffffffffffffll), col->At(1));
+    EXPECT_EQ(0ull,                  absl::Uint128High64(col->At(1)));
+    EXPECT_EQ(0xffffffffffffffffull, absl::Uint128Low64(col->At(1)));
+
+    EXPECT_EQ(absl::MakeUint128(0xffffffffffffffffll, 0), col->At(2));
+    EXPECT_EQ(static_cast<uint64_t>(0xffffffffffffffffull),  absl::Uint128High64(col->At(2)));
+    EXPECT_EQ(0ull,                  absl::Uint128Low64(col->At(2)));
+
+    EXPECT_EQ(0, col->At(4));
+}
+
 TEST(ColumnsCase, ColumnIPv4)
 {
     // TODO: split into proper method-level unit-tests
diff --git a/ut/value_generators.cpp b/ut/value_generators.cpp
@@ -104,13 +104,23 @@ std::vector<clickhouse::Int64> MakeDateTimes() {
 std::vector<clickhouse::Int128> MakeInt128s() {
     return {
         absl::MakeInt128(0xffffffffffffffffll, 0xffffffffffffffffll), // -1
-        absl::MakeInt128(0, 0xffffffffffffffffll),  // 2^64
+        absl::MakeInt128(0, 0xffffffffffffffffll),  // 2^64 - 1
         absl::MakeInt128(0xffffffffffffffffll, 0),
         absl::MakeInt128(0x8000000000000000ll, 0),
         Int128(0)
     };
 }
 
+std::vector<clickhouse::UInt128> MakeUInt128s() {
+    return {
+        absl::MakeUint128(0xffffffffffffffffll, 0xffffffffffffffffll), // 2^128 - 1
+        absl::MakeUint128(0, 0xffffffffffffffffll),  // 2^64 - 1
+        absl::MakeUint128(0xffffffffffffffffll, 0),  // 2^128 - 2^64
+        absl::MakeUint128(0x8000000000000000ll, 0),
+        UInt128(0)
+    };
+}
+
 std::vector<clickhouse::Int128> MakeDecimals(size_t /*precision*/, size_t scale) {
     const auto scale_multiplier = static_cast<size_t>(std::pow(10, scale));
     const long long int rhs_value = 12345678910;
diff --git a/ut/value_generators.h b/ut/value_generators.h
@@ -39,6 +39,7 @@ std::vector<in_addr> MakeIPv4s();
 std::vector<in6_addr> MakeIPv6s();
 std::vector<clickhouse::UUID> MakeUUIDs();
 std::vector<clickhouse::Int128> MakeInt128s();
+std::vector<clickhouse::UInt128> MakeUInt128s();
 std::vector<clickhouse::Int128> MakeDecimals(size_t precision, size_t scale);
 
 template <typename T, std::enable_if_t<std::is_integral<T>::value, bool> = true>

Original file line number	Diff line number	Diff line change
`@@ -5,8 +5,6 @@`
`5`	`5`
`6`	`6`	`namespace clickhouse {`
`7`	`7`
`8`		`-using UInt128 = std::pair<uint64_t, uint64_t>;`
`9`		`-`
`10`		`-using UUID = UInt128;`
	`8`	`+using UUID = std::pair<uint64_t, uint64_t>;`
`11`	`9`
`12`	`10`	`}`