improvement ColumnMap

Author: den818

clickhouse/CMakeLists.txt

@@ -102,6 +102,7 @@ INSTALL(FILES base/input.h DESTINATION include/clickhouse/base/)
 INSTALL(FILES base/open_telemetry.h DESTINATION include/clickhouse/base/)
 INSTALL(FILES base/output.h DESTINATION include/clickhouse/base/)
 INSTALL(FILES base/platform.h DESTINATION include/clickhouse/base/)
+INSTALL(FILES base/projected_iterator.h DESTINATION include/clickhouse/base/)
 INSTALL(FILES base/singleton.h DESTINATION include/clickhouse/base/)
 INSTALL(FILES base/socket.h DESTINATION include/clickhouse/base/)
 INSTALL(FILES base/string_utils.h DESTINATION include/clickhouse/base/)

clickhouse/base/projected_iterator.h

@@ -0,0 +1,55 @@
+#pragma once
+
+#include <iterator>
+#include <type_traits>
+#include <utility>
+
+namespace clickhouse {
+
+template <typename UnaryFunction, typename Iterator, typename Reference = decltype(std::declval<UnaryFunction>()(std::declval<Iterator>())),
+          typename Value = std::decay_t<Reference>>
+class ProjectedIterator {
+public:
+    using value_type        = Value;
+    using reference         = Reference;
+    using pointer           = Reference;
+    using difference_type   = typename std::iterator_traits<Iterator>::difference_type;
+    using iterator_category = typename std::iterator_traits<Iterator>::iterator_category;
+
+    ProjectedIterator() = default;
+
+    inline ProjectedIterator(Iterator const& iterator, UnaryFunction functor)
+        : iterator_(iterator)
+        , functor_(std::move(functor)) {
+    }
+
+    inline UnaryFunction functor() const { return functor; }
+
+    inline Iterator const& base() const { return iterator_; }
+
+    inline reference operator*() const { return functor_(iterator_); }
+
+    inline ProjectedIterator& operator++() {
+        ++iterator_;
+        return *this;
+    }
+
+    inline ProjectedIterator& operator--() {
+        --iterator_;
+        return *this;
+    }
+
+    inline bool operator==(const ProjectedIterator& other) const {
+        return this->iterator_ == other.iterator_;
+    }
+
+    inline bool operator!=(const ProjectedIterator& other) const {
+        return !(*this == other);
+    }
+
+private:
+    Iterator iterator_;
+    UnaryFunction functor_;
+};
+
+}  // namespace clickhouse

clickhouse/columns/array.h

@@ -167,6 +167,8 @@ class ColumnArrayT : public ColumnArray {
             const size_t size_;
             size_t index_;
         public:
+            Iterator() = default;
+
             Iterator(std::shared_ptr<NestedColumnType> typed_nested_data, size_t offset, size_t size, size_t index)
                 : typed_nested_data_(typed_nested_data)
                 , offset_(offset)

clickhouse/columns/map.h

@@ -1,5 +1,6 @@
 #pragma once
 
+#include "../base/projected_iterator.h"
 #include "array.h"
 #include "column.h"
 #include "tuple.h"
@@ -122,6 +123,8 @@ class ColumnMapT : public ColumnMap {
             typename ArrayColumnType::ArrayValueView::Iterator data_iterator_;
 
         public:
+            Iterator() = default;
+
             Iterator(typename ArrayColumnType::ArrayValueView::Iterator data_iterator)
                 : data_iterator_(data_iterator) {}
 
@@ -187,13 +190,16 @@ class ColumnMapT : public ColumnMap {
             if (size() != other.size()) {
                 return false;
             }
-            using Vector = std::vector<std::pair<Key, Value>>;
-            Vector l(begin(), end());
-            Vector r(other.begin(), other.end());
-            auto comp = [](const auto& l, const auto& r) { return l.frist < r.first; };
-            std::sort(l.begin(), l.end(), comp);
-            std::sort(r.begin(), r.end(), comp);
-            return std::equal(l.begin(), l.end(), r.begin(), r.end());
+            const auto make_index = [](const auto& data) {
+                std::vector<size_t> result{data.Size()};
+                std::generate(result.begin(), result.end(), [i = 0] () mutable { return i++; });
+                std::sort(result.begin(), result.end(), [&data](size_t l, size_t r) {return data[l] < data[r];});
+                return result;
+            };
+            const auto l_index = make_index(data_);
+            const auto r_index = make_index(other.data_);
+            return std::equal(l_index.begin(), l_index.end(), r_index.begin(), r_index.end(),
+                [&l_data = data_, &r_data = other.data_](size_t l, size_t r) { return l_data[l] == r_data[r];});
             return true;
         }
 
@@ -214,13 +220,17 @@ class ColumnMapT : public ColumnMap {
 
     template <typename T>
     inline void Append(const T& value) {
-        // TODO  Refuse to copy.
-        std::vector<std::tuple<typename T::key_type, typename T::mapped_type>> container;
-        container.reserve(value.size());
-        for (const auto& i : value) {
-            container.emplace_back(i.first, i.second);
-        }
-        typed_data_->Append(container.begin(), container.end());
+        using BaseIter = typename T::const_iterator;
+        using KeyOfT = decltype(std::declval<BaseIter>()->first);
+        using ValOfT = decltype(std::declval<BaseIter>()->second);
+        using Functor = std::function<std::tuple<KeyOfT, ValOfT>(const BaseIter&)>;
+        using Iterator = ProjectedIterator<Functor, BaseIter>;
+
+        Functor functor = [](const BaseIter& i) {
+            return std::make_tuple(i->first, i->second);
+        };
+
+        typed_data_->Append(Iterator{value.begin(), functor}, Iterator{value.end(), functor});
     }
 
     static auto Wrap(ColumnMap&& col) {