Merge pull request #258 from den818/Geo

Support geo column types
ClickHouse · Dec 16, 2022 · 4251535 · 4251535
2 parents 0472da2 + bac2967
commit 4251535
Show file tree

Hide file tree

Showing 13 changed files with 371 additions and 17 deletions.
diff --git a/README.md b/README.md
@@ -22,6 +22,7 @@ C++ client for [ClickHouse](https://clickhouse.com/).
 * Int128
 * UUID
 * Map
+* Point, Ring, Polygon, MultiPolygon
 
 ## Building
 

diff --git a/clickhouse/CMakeLists.txt b/clickhouse/CMakeLists.txt
@@ -12,6 +12,7 @@ SET ( clickhouse-cpp-lib-src
     columns/decimal.cpp
     columns/enum.cpp
     columns/factory.cpp
+    columns/geo.cpp
     columns/ip4.cpp
     columns/ip6.cpp
     columns/lowcardinality.cpp
@@ -117,6 +118,7 @@ INSTALL(FILES columns/date.h DESTINATION include/clickhouse/columns/)
 INSTALL(FILES columns/decimal.h DESTINATION include/clickhouse/columns/)
 INSTALL(FILES columns/enum.h DESTINATION include/clickhouse/columns/)
 INSTALL(FILES columns/factory.h DESTINATION include/clickhouse/columns/)
+INSTALL(FILES columns/geo.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/itemview.h DESTINATION include/clickhouse/columns/)

diff --git a/clickhouse/client.h b/clickhouse/client.h
@@ -7,6 +7,7 @@
 #include "columns/date.h"
 #include "columns/decimal.h"
 #include "columns/enum.h"
+#include "columns/geo.h"
 #include "columns/ip4.h"
 #include "columns/ip6.h"
 #include "columns/lowcardinality.h"

diff --git a/clickhouse/columns/factory.cpp b/clickhouse/columns/factory.cpp
@@ -4,17 +4,19 @@
 #include "date.h"
 #include "decimal.h"
 #include "enum.h"
+#include "geo.h"
 #include "ip4.h"
 #include "ip6.h"
 #include "lowcardinality.h"
 #include "lowcardinalityadaptor.h"
+#include "map.h"
 #include "nothing.h"
 #include "nullable.h"
 #include "numeric.h"
 #include "string.h"
 #include "tuple.h"
 #include "uuid.h"
-#include "map.h"
+
 
 #include "../types/type_parser.h"
 
@@ -97,6 +99,18 @@ static ColumnRef CreateTerminalColumn(const TypeAst& ast) {
     case Type::UUID:
         return std::make_shared<ColumnUUID>();
 
+    case Type::Point:
+        return std::make_shared<ColumnPoint>();
+
+    case Type::Ring:
+        return std::make_shared<ColumnRing>();
+
+    case Type::Polygon:
+        return std::make_shared<ColumnPolygon>();
+
+    case Type::MultiPolygon:
+        return std::make_shared<ColumnMultiPolygon>();
+
     default:
         return nullptr;
     }

diff --git a/clickhouse/columns/geo.cpp b/clickhouse/columns/geo.cpp
@@ -0,0 +1,108 @@
+#include "geo.h"
+
+#include "utils.h"
+
+namespace {
+using namespace ::clickhouse;
+
+template <Type::Code type_code>
+TypeRef CreateGeoType() {
+    if constexpr (type_code == Type::Code::Point) {
+        return Type::CreatePoint();
+    } else if constexpr (type_code == Type::Code::Ring) {
+        return Type::CreateRing();
+    } else if constexpr (type_code == Type::Code::Polygon) {
+        return Type::CreatePolygon();
+    } else if constexpr (type_code == Type::Code::MultiPolygon) {
+        return Type::CreateMultiPolygon();
+    }
+}
+
+template <typename ColumnType>
+std::shared_ptr<ColumnType> CreateColumn() {
+    if constexpr (std::is_same_v<ColumnType, ColumnTupleT<ColumnFloat64, ColumnFloat64>>) {
+        return std::make_shared<ColumnTupleT<ColumnFloat64, ColumnFloat64>>(
+            std::make_tuple(std::make_shared<ColumnFloat64>(), std::make_shared<ColumnFloat64>()));
+    } else {
+        return std::make_shared<ColumnType>();
+    }
+}
+
+}  // namespace
+
+namespace clickhouse {
+
+template <typename NestedColumnType, Type::Code type_code>
+ColumnGeo<NestedColumnType, type_code>::ColumnGeo()
+    : Column(std::move(CreateGeoType<type_code>())),
+      data_(std::move(CreateColumn<NestedColumnType>())) {
+}
+
+template <typename NestedColumnType, Type::Code type_code>
+ColumnGeo<NestedColumnType, type_code>::ColumnGeo(ColumnRef data)
+    : Column(std::move(CreateGeoType<type_code>()))
+    , data_(std::move(WrapColumn<NestedColumnType>(std::move(data)))) {
+}
+
+template <typename NestedColumnType, Type::Code type_code>
+void ColumnGeo<NestedColumnType, type_code>::Clear() {
+    data_->Clear();
+}
+
+template <typename NestedColumnType, Type::Code type_code>
+const typename ColumnGeo<NestedColumnType, type_code>::ValueType ColumnGeo<NestedColumnType, type_code>::At(size_t n) const {
+    return data_->At(n);
+}
+
+template <typename NestedColumnType, Type::Code type_code>
+const typename ColumnGeo<NestedColumnType, type_code>::ValueType ColumnGeo<NestedColumnType, type_code>::operator[](size_t n) const {
+    return data_->At(n);
+}
+
+template <typename NestedColumnType, Type::Code type_code>
+void ColumnGeo<NestedColumnType, type_code>::Append(ColumnRef column) {
+    if (auto col = column->template As<ColumnGeo>()) {
+        data_->Append(col->data_->template As<Column>());
+    }
+}
+
+template <typename NestedColumnType, Type::Code type_code>
+bool ColumnGeo<NestedColumnType, type_code>::LoadBody(InputStream* input, size_t rows) {
+    return data_->LoadBody(input, rows);
+}
+
+template <typename NestedColumnType, Type::Code type_code>
+void ColumnGeo<NestedColumnType, type_code>::SaveBody(OutputStream* output) {
+    data_->SaveBody(output);
+}
+
+template <typename NestedColumnType, Type::Code type_code>
+size_t ColumnGeo<NestedColumnType, type_code>::Size() const {
+    return data_->Size();
+}
+
+template <typename NestedColumnType, Type::Code type_code>
+ColumnRef ColumnGeo<NestedColumnType, type_code>::Slice(size_t begin, size_t len) const {
+    return std::make_shared<ColumnGeo>(data_->Slice(begin, len));
+}
+
+template <typename NestedColumnType, Type::Code type_code>
+ColumnRef ColumnGeo<NestedColumnType, type_code>::CloneEmpty() const {
+    return std::make_shared<ColumnGeo>();
+}
+
+template <typename NestedColumnType, Type::Code type_code>
+void ColumnGeo<NestedColumnType, type_code>::Swap(Column& other) {
+    auto& col = dynamic_cast<ColumnGeo&>(other);
+    data_.swap(col.data_);
+}
+
+template class ColumnGeo<ColumnTupleT<ColumnFloat64, ColumnFloat64>, Type::Code::Point>;
+
+template class ColumnGeo<ColumnArrayT<ColumnPoint>, Type::Code::Ring>;
+
+template class ColumnGeo<ColumnArrayT<ColumnRing>, Type::Code::Polygon>;
+
+template class ColumnGeo<ColumnArrayT<ColumnPolygon>, Type::Code::MultiPolygon>;
+
+}  // namespace clickhouse
diff --git a/clickhouse/columns/geo.h b/clickhouse/columns/geo.h
@@ -0,0 +1,76 @@
+#pragma once
+
+#include "array.h"
+#include "column.h"
+#include "numeric.h"
+#include "tuple.h"
+
+namespace clickhouse {
+
+template <typename NestedColumnType, Type::Code type_code>
+class ColumnGeo : public Column {
+public:
+    using ValueType = typename NestedColumnType::ValueType;
+
+    ColumnGeo();
+
+    explicit ColumnGeo(ColumnRef data);
+
+    /// Appends one element to the end of column.
+    template <typename T = ValueType>
+    void Append(const T& value) {
+        data_->Append(value);
+    }
+
+    /// Returns element at given row number.
+    const ValueType At(size_t n) const;
+
+    /// Returns element at given row number.
+    const ValueType operator[](size_t n) const;
+
+public:
+    /// Appends content of given column to the end of current one.
+    void Append(ColumnRef column) override;
+
+    /// Loads column data from input stream.
+    bool LoadBody(InputStream* input, size_t rows) override;
+
+    /// Saves column data to output stream.
+    void SaveBody(OutputStream* output) override;
+
+    /// Clear column data .
+    void Clear() override;
+
+    /// Returns count of rows in the column.
+    size_t Size() const override;
+
+    /// Makes slice of the current column.
+    ColumnRef Slice(size_t begin, size_t len) const override;
+    ColumnRef CloneEmpty() const override;
+    void Swap(Column& other) override;
+
+private:
+    std::shared_ptr<NestedColumnType> data_;
+};
+
+// /**
+//  * Represents a Point column.
+//  */
+using ColumnPoint = ColumnGeo<ColumnTupleT<ColumnFloat64, ColumnFloat64>, Type::Code::Point>;
+
+/**
+ * Represents a Ring column.
+ */
+using ColumnRing = ColumnGeo<ColumnArrayT<ColumnPoint>, Type::Code::Ring>;
+
+/**
+ * Represents a Polygon column.
+ */
+using ColumnPolygon = ColumnGeo<ColumnArrayT<ColumnRing>, Type::Code::Polygon>;
+
+/**
+ * Represents a MultiPolygon column.
+ */
+using ColumnMultiPolygon = ColumnGeo<ColumnArrayT<ColumnPolygon>, Type::Code::MultiPolygon>;
+
+}  // namespace clickhouse
diff --git a/clickhouse/columns/tuple.h b/clickhouse/columns/tuple.h
@@ -76,16 +76,11 @@ class ColumnTupleT : public ColumnTuple {
 
     inline ValueType operator[](size_t index) const { return GetTupleOfValues(index); }
 
-    template <typename T, size_t index = std::tuple_size_v<T>>
-    inline void Append([[maybe_unused]] T value) {
-        static_assert(index <= std::tuple_size_v<T>);
-        static_assert(std::tuple_size_v<TupleOfColumns> == std::tuple_size_v<T>);
-        if constexpr (index == 0) {
-            return;
-        } else {
-            std::get<index - 1>(typed_columns_)->Append(std::move(std::get<index - 1>(value)));
-            Append<T, index - 1>(std::move(value));
-        }
+    using ColumnTuple::Append;
+
+    template <typename... T>
+    inline void Append(std::tuple<T...> value) {
+        AppendTuple(std::move(value));
     }
 
     /** Create a ColumnTupleT from a ColumnTuple, without copying data and offsets, but by
@@ -122,6 +117,18 @@ class ColumnTupleT : public ColumnTuple {
     }
 
 private:
+    template <typename T, size_t index = std::tuple_size_v<T>>
+    inline void AppendTuple([[maybe_unused]] T value) {
+        static_assert(index <= std::tuple_size_v<T>);
+        static_assert(std::tuple_size_v<TupleOfColumns> == std::tuple_size_v<T>);
+        if constexpr (index == 0) {
+            return;
+        } else {
+            std::get<index - 1>(typed_columns_)->Append(std::move(std::get<index - 1>(value)));
+            AppendTuple<T, index - 1>(std::move(value));
+        }
+    }
+
     template <typename T, size_t index = std::tuple_size_v<T>>
     inline static std::vector<ColumnRef> TupleToVector([[maybe_unused]] const T& value) {
         static_assert(index <= std::tuple_size_v<T>);

diff --git a/clickhouse/types/type_parser.cpp b/clickhouse/types/type_parser.cpp
@@ -46,7 +46,11 @@ static const std::unordered_map<std::string, Type::Code> kTypeCode = {
     { "Decimal64",   Type::Decimal64 },
     { "Decimal128",  Type::Decimal128 },
     { "LowCardinality", Type::LowCardinality },
-    { "Map", Type::Map},
+    { "Map",         Type::Map},
+    { "Point",       Type::Point},
+    { "Ring",        Type::Ring},
+    { "Polygon",     Type::Polygon},
+    { "MultiPolygon", Type::MultiPolygon},
 };
 
 static Type::Code GetTypeCode(const std::string& name) {

diff --git a/clickhouse/types/types.cpp b/clickhouse/types/types.cpp
@@ -47,6 +47,10 @@ const char* Type::TypeName(Type::Code code) {
         case Type::Code::DateTime64:     return "DateTime64";
         case Type::Code::Date32:         return "Date32";
         case Type::Code::Map:            return "Map";
+        case Type::Code::Point:          return "Point";
+        case Type::Code::Ring:           return "Ring";
+        case Type::Code::Polygon:        return "Polygon";
+        case Type::Code::MultiPolygon:   return "MultiPolygon";
     }
 
     return "Unknown type";
@@ -72,6 +76,10 @@ std::string Type::GetName() const {
         case IPv6:
         case Date:
         case Date32:
+        case Point:
+        case Ring:
+        case Polygon:
+        case MultiPolygon:
             return TypeName(code_);
         case FixedString:
             return As<FixedStringType>()->GetName();
@@ -126,6 +134,10 @@ uint64_t Type::GetTypeUniqueId() const {
         case IPv6:
         case Date:
         case Date32:
+        case Point:
+        case Ring:
+        case Polygon:
+        case MultiPolygon:
             // For simple types, unique ID is the same as Type::Code
             return code_;
 
@@ -233,6 +245,22 @@ TypeRef Type::CreateMap(TypeRef key_type, TypeRef value_type) {
     return std::make_shared<MapType>(key_type, value_type);
 }
 
+TypeRef Type::CreatePoint() {
+    return TypeRef(new Type(Type::Point));
+}
+
+TypeRef Type::CreateRing() {
+    return TypeRef(new Type(Type::Ring));
+}
+
+TypeRef Type::CreatePolygon() {
+    return TypeRef(new Type(Type::Polygon));
+}
+
+TypeRef Type::CreateMultiPolygon() {
+    return TypeRef(new Type(Type::MultiPolygon));
+}
+
 /// class ArrayType
 
 ArrayType::ArrayType(TypeRef item_type) : Type(Array), item_type_(item_type) {

diff --git a/clickhouse/types/types.h b/clickhouse/types/types.h
@@ -50,7 +50,11 @@ class Type {
         LowCardinality,
         DateTime64,
         Date32,
-        Map
+        Map,
+        Point,
+        Ring,
+        Polygon,
+        MultiPolygon
     };
 
     using EnumItem = std::pair<std::string /* name */, int16_t /* value */>;
@@ -128,6 +132,14 @@ class Type {
 
     static TypeRef CreateMap(TypeRef key_type, TypeRef value_type);
 
+    static TypeRef CreatePoint();
+
+    static TypeRef CreateRing();
+
+    static TypeRef CreatePolygon();
+
+    static TypeRef CreateMultiPolygon();
+
 private:
     uint64_t GetTypeUniqueId() const;