ColumnNulableT

Author: den818

clickhouse/columns/lowcardinality.h

@@ -53,9 +53,16 @@ class ColumnLowCardinality : public Column {
     UniqueItems unique_items_map_;
 
 public:
+    ColumnLowCardinality(ColumnLowCardinality&& col) = default;
     // c-tor makes a deep copy of the dictionary_column.
     explicit ColumnLowCardinality(ColumnRef dictionary_column);
     explicit ColumnLowCardinality(std::shared_ptr<ColumnNullable> dictionary_column);
+
+    template <typename T>
+    explicit ColumnLowCardinality(std::shared_ptr<ColumnNullableT<T>> dictionary_column)
+        : ColumnLowCardinality(dictionary_column->template As<ColumnNullable>())
+    {}
+
     ~ColumnLowCardinality();
 
     /// Appends another LowCardinality column to the end of this one, updating dictionary.
@@ -117,16 +124,23 @@ class ColumnLowCardinalityT : public ColumnLowCardinality {
     // Type this column takes as argument of Append and returns with At() and operator[]
     using ValueType = typename DictionaryColumnType::ValueType;
 
+    explicit ColumnLowCardinalityT(ColumnLowCardinality&& col)
+        : ColumnLowCardinality(std::move(col))
+        ,  typed_dictionary_(dynamic_cast<DictionaryColumnType &>(*GetDictionary()))
+        ,  type_(GetTypeCode(typed_dictionary_))
+    {
+    }
+
     template <typename ...Args>
     explicit ColumnLowCardinalityT(Args &&... args)
         : ColumnLowCardinalityT(std::make_shared<DictionaryColumnType>(std::forward<Args>(args)...))
     {}
 
     // Create LC<T> column from existing T-column, making a deep copy of all contents.
     explicit ColumnLowCardinalityT(std::shared_ptr<DictionaryColumnType> dictionary_col)
-        : ColumnLowCardinality(dictionary_col),
-          typed_dictionary_(dynamic_cast<DictionaryColumnType &>(*GetDictionary())),
-          type_(typed_dictionary_.Type()->GetCode())
+        : ColumnLowCardinality(dictionary_col)
+        , typed_dictionary_(dynamic_cast<DictionaryColumnType &>(*GetDictionary()))
+        , type_(GetTypeCode(typed_dictionary_))
     {}
 
     /// Extended interface to simplify reading/adding individual items.
@@ -145,7 +159,15 @@ class ColumnLowCardinalityT : public ColumnLowCardinality {
     using ColumnLowCardinality::Append;
 
     inline void Append(const ValueType & value) {
-        AppendUnsafe(ItemView{type_, value});
+        if constexpr (IsNullable<WrappedColumnType>) {
+            if (value.has_value()) {
+                AppendUnsafe(ItemView{type_, *value});
+            } else {
+                AppendUnsafe(ItemView{});
+            }
+        } else {
+            AppendUnsafe(ItemView{type_, value});
+        }
     }
 
     template <typename T>
@@ -154,6 +176,41 @@ class ColumnLowCardinalityT : public ColumnLowCardinality {
             Append(item);
         }
     }
+
+    /** Create a ColumnLowCardinalityT from a ColumnLowCardinality, without copying data and offsets, but by
+     * 'stealing' those from `col`.
+     *
+     *  Ownership of column internals is transferred to returned object, original (argument) object
+     *  MUST NOT BE USED IN ANY WAY, it is only safe to dispose it.
+     *
+     *  Throws an exception if `col` is of wrong type, it is safe to use original col in this case.
+     *  This is a static method to make such conversion verbose.
+     */
+    static auto Wrap(ColumnLowCardinality&& col) {
+        return std::make_shared<ColumnLowCardinalityT<WrappedColumnType>>(std::move(col));
+    }
+
+    static auto Wrap(Column&& col) { return Wrap(std::move(dynamic_cast<ColumnLowCardinality&&>(col))); }
+
+    // Helper to simplify integration with other APIs
+    static auto Wrap(ColumnRef&& col) { return Wrap(std::move(*col->AsStrict<ColumnLowCardinality>())); }
+
+    ColumnRef Slice(size_t begin, size_t size) const override {
+        return Wrap(ColumnLowCardinality::Slice(begin, size));
+    }
+
+    ColumnRef CloneEmpty() const override { return Wrap(ColumnLowCardinality::CloneEmpty()); }
+
+private:
+
+    template <typename T>
+    static auto GetTypeCode(T& column) {
+        if constexpr (IsNullable<T>) {
+            return GetTypeCode(*column.Nested()->template AsStrict<typename T::NestedColumnType>());
+        } else {
+            return column.Type()->GetCode();
+        }
+    }
 };
 
 }

clickhouse/columns/nullable.h

@@ -3,6 +3,8 @@
 #include "column.h"
 #include "numeric.h"
 
+#include <optional>
+
 namespace clickhouse {
 
 /**
@@ -42,7 +44,7 @@ class ColumnNullable : public Column {
 
     /// Clear column data .
     void Clear() override;
-    
+
     /// Returns count of rows in the column.
     size_t Size() const override;
 
@@ -58,4 +60,91 @@ class ColumnNullable : public Column {
     std::shared_ptr<ColumnUInt8> nulls_;
 };
 
+template <typename ColumnType>
+class ColumnNullableT : public ColumnNullable {
+public:
+    using NestedColumnType = ColumnType;
+    using ValueType = std::optional<std::decay_t<decltype(std::declval<NestedColumnType>().At(0))>>;
+
+    ColumnNullableT(std::shared_ptr<NestedColumnType> data, std::shared_ptr<ColumnUInt8> nulls)
+        : ColumnNullable(data, nulls)
+        , typed_nested_data_(data)
+    {}
+
+    explicit ColumnNullableT(std::shared_ptr<NestedColumnType> data)
+        : ColumnNullableT(data, FillNulls(data->Size()))
+    {}
+
+    template <typename ...Args>
+    explicit ColumnNullableT(Args &&... args)
+        : ColumnNullableT(std::make_shared<NestedColumnType>(std::forward<Args>(args)...))
+    {}
+
+    inline ValueType At(size_t index) const {
+        return IsNull(index) ? ValueType{} : ValueType{typed_nested_data_->At(index)};
+    }
+
+    inline ValueType operator[](size_t index) const { return At(index); }
+
+    /// Appends content of given column to the end of current one.
+    void Append(ColumnRef column) override {
+        ColumnNullable::Append(std::move(column));
+    }
+
+    inline void Append(ValueType value) {
+        ColumnNullable::Append(!value.has_value());
+        if (value.has_value()) {
+            typed_nested_data_->Append(std::move(*value));
+        } else {
+            typed_nested_data_->Append(typename ValueType::value_type{});
+        }
+    }
+
+    /** Create a ColumnNullableT from a ColumnNullable, without copying data and offsets, but by
+     * 'stealing' those from `col`.
+     *
+     *  Ownership of column internals is transferred to returned object, original (argument) object
+     *  MUST NOT BE USED IN ANY WAY, it is only safe to dispose it.
+     *
+     *  Throws an exception if `col` is of wrong type, it is safe to use original col in this case.
+     *  This is a static method to make such conversion verbose.
+     */
+    static auto Wrap(ColumnNullable&& col) {
+        return std::make_shared<ColumnNullableT<NestedColumnType>>(
+            col.Nested()->AsStrict<NestedColumnType>(),
+            col.Nulls()->AsStrict<ColumnUInt8>()) ;
+    }
+
+    static auto Wrap(Column&& col) { return Wrap(std::move(dynamic_cast<ColumnNullable&&>(col))); }
+
+    // Helper to simplify integration with other APIs
+    static auto Wrap(ColumnRef&& col) { return Wrap(std::move(*col->AsStrict<ColumnNullable>())); }
+
+    ColumnRef Slice(size_t begin, size_t size) const override {
+        return Wrap(ColumnNullable::Slice(begin, size));
+    }
+
+    ColumnRef CloneEmpty() const override { return Wrap(ColumnNullable::CloneEmpty()); }
+
+    void Swap(Column& other) override {
+        auto& col = dynamic_cast<ColumnNullableT<NestedColumnType>&>(other);
+        typed_nested_data_.swap(col.typed_nested_data_);
+        ColumnNullable::Swap(other);
+    }
+
+private:
+    static inline auto FillNulls(size_t n){
+        auto result = std::make_shared<ColumnUInt8>();
+        for (size_t i = 0; i < n; ++i) {
+            result->Append(0);
+        }
+        return result;
+    }
+
+    std::shared_ptr<NestedColumnType> typed_nested_data_;
+};
+
+template <typename T>
+constexpr bool IsNullable = std::is_base_of_v<ColumnNullable, T>;
+
 }

ut/CMakeLists.txt

@@ -20,6 +20,7 @@ SET ( clickhouse-cpp-ut-src
     CreateColumnByType_ut.cpp
     Column_ut.cpp
     roundtrip_column.cpp
+    roundtrip_tests.cpp
 
     utils.cpp
     value_generators.cpp

ut/Column_ut.cpp

@@ -106,6 +106,28 @@ class GenericColumnTest : public testing::Test {
 
         return std::tuple{column, values};
     }
+
+    static std::optional<std::string> SkipTest(clickhouse::Client& client) {
+        if constexpr (std::is_same_v<ColumnType, ColumnDate32>) {
+            // Date32 first appeared in v21.9.2.17-stable
+            const auto server_info = client.GetServerInfo();
+            if (versionNumber(server_info) < versionNumber(21, 9)) {
+                std::stringstream buffer;
+                buffer << "Date32 is available since v21.9.2.17-stable and can't be tested against server: " << server_info;
+                return buffer.str();
+            }
+        }
+
+        if constexpr (std::is_same_v<ColumnType, ColumnInt128>) {
+            const auto server_info = client.GetServerInfo();
+            if (versionNumber(server_info) < versionNumber(21, 7)) {
+                std::stringstream buffer;
+                buffer <<  "ColumnInt128 is available since v21.7.2.7-stable and can't be tested against server: " << server_info;
+                return buffer.str();
+            }
+        }
+        return std::nullopt;
+    }
 };
 
 using ValueColumns = ::testing::Types<
@@ -279,21 +301,33 @@ TYPED_TEST(GenericColumnTest, RoundTrip) {
 
     clickhouse::Client client(LocalHostEndpoint);
 
-    if constexpr (std::is_same_v<typename TestFixture::ColumnType, ColumnDate32>) {
-        // Date32 first appeared in v21.9.2.17-stable
-        const auto server_info = client.GetServerInfo();
-        if (versionNumber(server_info) < versionNumber(21, 9)) {
-            GTEST_SKIP() << "Date32 is available since v21.9.2.17-stable and can't be tested against server: " << server_info;
-        }
+    if (auto message = this->SkipTest(client)) {
+        GTEST_SKIP() << *message;
     }
 
-    if constexpr (std::is_same_v<typename TestFixture::ColumnType, ColumnInt128>) {
-        const auto server_info = client.GetServerInfo();
-        if (versionNumber(server_info) < versionNumber(21, 7)) {
-            GTEST_SKIP() << "ColumnInt128 is available since v21.7.2.7-stable and can't be tested against server: " << server_info;
+    auto result_typed = RoundtripColumnValues(client, column)->template AsStrict<typename TestFixture::ColumnType>();
+    EXPECT_TRUE(CompareRecursive(*column, *result_typed));
+}
+
+TYPED_TEST(GenericColumnTest, NulableT_RoundTrip) {
+    using NullableType = ColumnNullableT<TypeParam>;
+    auto column = std::make_shared<NullableType>(this->MakeColumn());
+    auto values = this->GenerateValues(100);
+    FromVectorGenerator<bool> is_null({true, false});
+    for (size_t i = 0; i < values.size(); ++i) {
+        if (is_null(i)) {
+            column->Append(std::nullopt);
+        } else {
+            column->Append(values[i]);
         }
     }
 
-    auto result_typed = RoundtripColumnValues(client, column)->template AsStrict<typename TestFixture::ColumnType>();
+    clickhouse::Client client(LocalHostEndpoint);
+
+    if (auto message = this->SkipTest(client)) {
+        GTEST_SKIP() << *message;
+    }
+
+    auto result_typed = WrapColumn<NullableType>(RoundtripColumnValues(client, column));
     EXPECT_TRUE(CompareRecursive(*column, *result_typed));
 }