Merge pull request #192 from Enmk/Enum_memory_improvement

Better ColumnEnum - less memory

Author: Enmk

clickhouse/columns/enum.cpp

@@ -44,7 +44,7 @@ const T& ColumnEnum<T>::At(size_t n) const {
 }
 
 template <typename T>
-const std::string ColumnEnum<T>::NameAt(size_t n) const {
+std::string_view ColumnEnum<T>::NameAt(size_t n) const {
     return type_->As<EnumType>()->GetEnumName(data_.at(n));
 }
 

clickhouse/columns/enum.h

@@ -19,7 +19,7 @@ class ColumnEnum : public Column {
 
     /// Returns element at given row number.
     const T& At(size_t n) const;
-    const std::string NameAt(size_t n) const;
+    std::string_view NameAt(size_t n) const;
 
     /// Returns element at given row number.
     const T& operator[] (size_t n) const;
@@ -38,7 +38,7 @@ class ColumnEnum : public Column {
     /// Saves column data to output stream.
     void SaveBody(OutputStream* output) override;
 
-    /// Clear column data .
+    /// Clear column data.
     void Clear() override;
 
     /// Returns count of rows in the column.

clickhouse/types/types.cpp

@@ -259,8 +259,8 @@ std::string DecimalType::GetName() const {
 
 EnumType::EnumType(Type::Code type, const std::vector<EnumItem>& items) : Type(type) {
     for (const auto& item : items) {
-        value_to_name_[item.second] = item.first;
-        name_to_value_[item.first]  = item.second;
+        auto result = name_to_value_.insert(item);
+        value_to_name_[item.second] = result.first->first;
     }
 }
 
@@ -291,7 +291,7 @@ std::string EnumType::GetName() const {
     return result;
 }
 
-const std::string& EnumType::GetEnumName(int16_t value) const {
+std::string_view EnumType::GetEnumName(int16_t value) const {
     return value_to_name_.at(value);
 }
 

clickhouse/types/types.h

@@ -212,13 +212,13 @@ class EnumType : public Type {
     std::string GetName() const;
 
     /// Methods to work with enum types.
-    const std::string& GetEnumName(int16_t value) const;
+    std::string_view GetEnumName(int16_t value) const;
     int16_t GetEnumValue(const std::string& name) const;
     bool HasEnumName(const std::string& name) const;
     bool HasEnumValue(int16_t value) const;
 
 private:
-    using ValueToNameType     = std::map<int16_t, std::string>;
+    using ValueToNameType     = std::map<int16_t, std::string_view>;
     using NameToValueType     = std::map<std::string, int16_t>;
     using ValueToNameIterator = ValueToNameType::const_iterator;