LowCardinality column implementation

.travis.yml

@@ -35,7 +35,7 @@ matrix:
         - MATRIX_EVAL="CC=clang-6.0 && CXX=clang++-6.0"
 
     - os: osx
-      osx_image: xcode8.2
+      osx_image: xcode9.4
       compiler: clang
 
 before_install:
@@ -51,4 +51,4 @@ script:
   - cd build
   - cmake .. -DBUILD_TESTS=ON && make
   - if [[ "$TRAVIS_OS_NAME" == "linux" ]]; then ./ut/clickhouse-cpp-ut ; fi
-  - if [[ "$TRAVIS_OS_NAME" == "osx" ]]; then ./ut/clickhouse-cpp-ut --gtest_filter='-Client/*' ; fi
+  - if [[ "$TRAVIS_OS_NAME" == "osx" ]]; then ./ut/clickhouse-cpp-ut --gtest_filter=-"Client/*:*Performance*" ; fi

clickhouse/CMakeLists.txt

@@ -13,11 +13,13 @@ SET ( clickhouse-cpp-lib-src
     columns/factory.cpp
     columns/ip4.cpp
     columns/ip6.cpp
+    columns/lowcardinality.cpp
     columns/nullable.cpp
     columns/numeric.cpp
     columns/string.cpp
     columns/tuple.cpp
     columns/uuid.cpp
+    columns/itemview.cpp
 
     types/type_parser.cpp
     types/types.cpp
@@ -82,12 +84,14 @@ INSTALL(FILES columns/enum.h DESTINATION include/clickhouse/columns/)
 INSTALL(FILES columns/factory.h DESTINATION include/clickhouse/columns/)
 INSTALL(FILES columns/ip4.h DESTINATION include/clickhouse/columns/)
 INSTALL(FILES columns/ip6.h DESTINATION include/clickhouse/columns/)
+INSTALL(FILES columns/lowcardinality.h DESTINATION include/clickhouse/columns/)
 INSTALL(FILES columns/nullable.h DESTINATION include/clickhouse/columns/)
 INSTALL(FILES columns/numeric.h DESTINATION include/clickhouse/columns/)
 INSTALL(FILES columns/string.h DESTINATION include/clickhouse/columns/)
 INSTALL(FILES columns/tuple.h DESTINATION include/clickhouse/columns/)
 INSTALL(FILES columns/utils.h DESTINATION include/clickhouse/columns/)
 INSTALL(FILES columns/uuid.h DESTINATION include/clickhouse/columns/)
+INSTALL(FILES columns/itemview.h DESTINATION include/clickhouse/columns/)
 
 # types
 INSTALL(FILES types/type_parser.h DESTINATION include/clickhouse/types/)

clickhouse/base/string_view.h

@@ -75,6 +75,11 @@ class StringViewImpl {
         return size_;
     }
 
+    // to mimic std::string and std::string_view
+    inline size_type length() const noexcept {
+        return size();
+    }
+
 public:
     // Returns a substring [pos, pos + count).
     // If the requested substring extends past the end of the string,

clickhouse/client.cpp

@@ -22,14 +22,19 @@
 #define DBMS_NAME                                       "ClickHouse"
 #define DBMS_VERSION_MAJOR                              1
 #define DBMS_VERSION_MINOR                              1
-#define REVISION                                        54126
+#define REVISION                                        54405
 
 #define DBMS_MIN_REVISION_WITH_TEMPORARY_TABLES         50264
 #define DBMS_MIN_REVISION_WITH_TOTAL_ROWS_IN_PROGRESS   51554
 #define DBMS_MIN_REVISION_WITH_BLOCK_INFO               51903
 #define DBMS_MIN_REVISION_WITH_CLIENT_INFO              54032
 #define DBMS_MIN_REVISION_WITH_SERVER_TIMEZONE          54058
 #define DBMS_MIN_REVISION_WITH_QUOTA_KEY_IN_CLIENT_INFO 54060
+//#define DBMS_MIN_REVISION_WITH_TABLES_STATUS            54226
+//#define DBMS_MIN_REVISION_WITH_TIME_ZONE_PARAMETER_IN_DATETIME_DATA_TYPE 54337
+#define DBMS_MIN_REVISION_WITH_SERVER_DISPLAY_NAME      54372
+#define DBMS_MIN_REVISION_WITH_VERSION_PATCH            54401
+#define DBMS_MIN_REVISION_WITH_LOW_CARDINALITY_TYPE     54405
 
 namespace clickhouse {
 
@@ -51,8 +56,10 @@ struct ClientInfo {
 struct ServerInfo {
     std::string name;
     std::string timezone;
+    std::string display_name;
     uint64_t    version_major;
     uint64_t    version_minor;
+    uint64_t    version_patch;
     uint64_t    revision;
 };
 
@@ -563,6 +570,8 @@ void Client::Impl::SendQuery(const std::string& query) {
 
         if (server_info_.revision >= DBMS_MIN_REVISION_WITH_QUOTA_KEY_IN_CLIENT_INFO)
             WireFormat::WriteString(&output_, info.quota_key);
+        if (server_info_.revision >= DBMS_MIN_REVISION_WITH_VERSION_PATCH)
+            WireFormat::WriteUInt64(&output_, info.client_revision);
     }
 
     /// Per query settings.
@@ -697,6 +706,18 @@ bool Client::Impl::ReceiveHello() {
             }
         }
 
+        if (server_info_.revision >= DBMS_MIN_REVISION_WITH_SERVER_DISPLAY_NAME) {
+            if (!WireFormat::ReadString(&input_, &server_info_.display_name)) {
+                return false;
+            }
+        }
+
+        if (server_info_.revision >= DBMS_MIN_REVISION_WITH_VERSION_PATCH) {
+            if (!WireFormat::ReadUInt64(&input_, &server_info_.version_patch)) {
+                return false;
+            }
+        }
+
         return true;
     } else if (packet_type == ServerCodes::Exception) {
         ReceiveException(true);

clickhouse/client.h

@@ -9,6 +9,7 @@
 #include "columns/enum.h"
 #include "columns/ip4.h"
 #include "columns/ip6.h"
+#include "columns/lowcardinality.h"
 #include "columns/nullable.h"
 #include "columns/numeric.h"
 #include "columns/string.h"

clickhouse/columns/array.cpp

@@ -78,6 +78,13 @@ size_t ColumnArray::Size() const {
     return offsets_->Size();
 }
 
+void ColumnArray::Swap(Column& other) {
+    if (auto col = dynamic_cast<ColumnArray*>(&other)) {
+        data_.swap(col->data_);
+        offsets_.swap(col->offsets_);
+    }
+}
+
 void ColumnArray::OffsetsIncrease(size_t n) {
     offsets_->Append(n);
 }

clickhouse/columns/array.h

@@ -38,6 +38,8 @@ class ColumnArray : public Column {
     /// Makes slice of the current column.
     ColumnRef Slice(size_t, size_t) override;
 
+    void Swap(Column&) override;
+
     void OffsetsIncrease(size_t);
 
 private:

clickhouse/columns/column.h

@@ -3,6 +3,9 @@
 #include "../base/coded.h"
 #include "../base/input.h"
 #include "../types/types.h"
+#include "../columns/itemview.h"
+
+#include <memory>
 
 namespace clickhouse {
 
@@ -50,6 +53,19 @@ class Column : public std::enable_shared_from_this<Column> {
     /// Makes slice of the current column.
     virtual ColumnRef Slice(size_t begin, size_t len) = 0;
 
+    virtual void Swap(Column&) = 0;
+
+    /// Get a view on raw item data if it is supported by column, will throw an exception if index is out of range.
+    /// Please note that view is invalidated once column is items are added or deleted, column is loaded from strean or destroyed.
+    virtual ItemView GetItem(size_t) const {
+        throw std::runtime_error("GetItem() is not supported for column of " + type_->GetName());
+    }
+
+    friend void swap(Column& left, Column& right)
+    {
+        left.Swap(right);
+    }
+
 protected:
     TypeRef type_;
 };

clickhouse/columns/date.cpp

@@ -48,6 +48,17 @@ ColumnRef ColumnDate::Slice(size_t begin, size_t len) {
     return result;
 }
 
+void ColumnDate::Swap(Column& other) {
+    if (auto col = dynamic_cast<ColumnDate*>(&other)) {
+        data_.swap(col->data_);
+    }
+}
+
+ItemView ColumnDate::GetItem(size_t index) const {
+    return data_->GetItem(index);
+}
+
+
 
 ColumnDateTime::ColumnDateTime()
     : Column(Type::CreateDateTime())
@@ -94,4 +105,14 @@ ColumnRef ColumnDateTime::Slice(size_t begin, size_t len) {
     return result;
 }
 
+void ColumnDateTime::Swap(Column& other) {
+    if (auto col = dynamic_cast<ColumnDateTime*>(&other)) {
+        data_.swap(col->data_);
+    }
+}
+
+ItemView ColumnDateTime::GetItem(size_t index) const {
+    return data_->GetItem(index);
+}
+
 }

clickhouse/columns/date.h

@@ -37,6 +37,10 @@ class ColumnDate : public Column {
     /// Makes slice of the current column.
     ColumnRef Slice(size_t begin, size_t len) override;
 
+    void Swap(Column& other) override;
+
+    ItemView GetItem(size_t index) const override;
+
 private:
     std::shared_ptr<ColumnUInt16> data_;
 };
@@ -70,6 +74,10 @@ class ColumnDateTime : public Column {
     /// Makes slice of the current column.
     ColumnRef Slice(size_t begin, size_t len) override;
 
+    void Swap(Column& other) override;
+
+    ItemView GetItem(size_t index) const override;
+
 private:
     std::shared_ptr<ColumnUInt32> data_;
 };

clickhouse/columns/decimal.cpp

@@ -120,4 +120,14 @@ ColumnRef ColumnDecimal::Slice(size_t begin, size_t len) {
     return slice;
 }
 
+void ColumnDecimal::Swap(Column& other) {
+    if (auto col = dynamic_cast<ColumnDecimal*>(&other)) {
+        data_.swap(col->data_);
+    }
+}
+
+ItemView ColumnDecimal::GetItem(size_t index) const {
+    return data_->GetItem(index);
+}
+
 }

clickhouse/columns/decimal.h

@@ -24,6 +24,8 @@ class ColumnDecimal : public Column {
     void Clear() override;
     size_t Size() const override;
     ColumnRef Slice(size_t begin, size_t len) override;
+    void Swap(Column& other) override;
+    ItemView GetItem(size_t index) const override;
 
 private:
     /// Depending on a precision it can be one of:

clickhouse/columns/enum.cpp

@@ -90,6 +90,19 @@ ColumnRef ColumnEnum<T>::Slice(size_t begin, size_t len) {
     return std::make_shared<ColumnEnum<T>>(type_, SliceVector(data_, begin, len));
 }
 
+template <typename T>
+void ColumnEnum<T>::Swap(Column& other) {
+    if (auto col = dynamic_cast<ColumnEnum<T>*>(&other)) {
+        data_.swap(col->data_);
+    }
+}
+
+template <typename T>
+ItemView ColumnEnum<T>::GetItem(size_t index) const {
+    const T value = data_[index];
+    return ItemView{type_->GetCode(), value};
+}
+
 template class ColumnEnum<int8_t>;
 template class ColumnEnum<int16_t>;
 

clickhouse/columns/enum.h

@@ -45,6 +45,10 @@ class ColumnEnum : public Column {
     /// Makes slice of the current column.
     ColumnRef Slice(size_t begin, size_t len) override;
 
+    void Swap(Column& other) override;
+
+    ItemView GetItem(size_t index) const override;
+
 private:
     std::vector<T> data_;
 };

clickhouse/columns/factory.cpp

@@ -9,6 +9,7 @@
 #include "nothing.h"
 #include "nullable.h"
 #include "numeric.h"
+#include "lowcardinality.h"
 #include "string.h"
 #include "tuple.h"
 #include "uuid.h"
@@ -134,6 +135,18 @@ static ColumnRef CreateColumnFromAst(const TypeAst& ast) {
             }
             break;
         }
+        case TypeAst::LowCardinality: {
+            const auto nested = ast.elements.front();
+            switch (nested.code) {
+                // TODO (nemkov): update this to maximize code reuse.
+                case Type::String:
+                    return std::make_shared<ColumnLowCardinalityT<ColumnString>>();
+                case Type::FixedString:
+                    return std::make_shared<ColumnLowCardinalityT<ColumnFixedString>>(nested.elements.front().value);
+                default:
+                    throw std::runtime_error("LowCardinality(" + nested.name + ") is not supported");
+            }
+        }
 
         case TypeAst::Null:
         case TypeAst::Number:

clickhouse/columns/ip4.cpp

@@ -78,4 +78,14 @@ ColumnRef ColumnIPv4::Slice(size_t begin, size_t len) {
     return std::make_shared<ColumnIPv4>(data_->Slice(begin, len));
 }
 
+void ColumnIPv4::Swap(Column& other) {
+    if (auto col = dynamic_cast<ColumnIPv4*>(&other)) {
+        data_.swap(col->data_);
+    }
+}
+
+ItemView ColumnIPv4::GetItem(size_t index) const {
+    return data_->GetItem(index);
+}
+
 }

clickhouse/columns/ip4.h

@@ -43,6 +43,10 @@ class ColumnIPv4 : public Column {
     /// Makes slice of the current column.
     ColumnRef Slice(size_t begin, size_t len) override;
 
+    void Swap(Column& other) override;
+
+    ItemView GetItem(size_t index) const override;
+
 private:
     std::shared_ptr<ColumnUInt32> data_;
 };

clickhouse/columns/ip6.cpp

@@ -78,4 +78,14 @@ ColumnRef ColumnIPv6::Slice(size_t begin, size_t len) {
     return std::make_shared<ColumnIPv6>(data_->Slice(begin, len));
 }
 
+void ColumnIPv6::Swap(Column& other) {
+    if (auto col = dynamic_cast<ColumnIPv6*>(&other)) {
+        data_.swap(col->data_);
+    }
+}
+
+ItemView ColumnIPv6::GetItem(size_t index) const {
+    return data_->GetItem(index);
+}
+
 }

clickhouse/columns/ip6.h

@@ -41,6 +41,8 @@ class ColumnIPv6 : public Column{
 
     /// Makes slice of the current column.
     ColumnRef Slice(size_t begin, size_t len) override;
+    void Swap(Column& other) override;
+    ItemView GetItem(size_t index) const override;
 
 private:
     std::shared_ptr<ColumnFixedString> data_;