[pandas 2.0] Re-factor of the DataType classes

* Changing PrimitiveType to NumericType  * Removing the macro for
declaring sub-types of NumericType and    instead using template
arguments  * Addding a static SINGLETON member to the NumericType base
class    instead of the macros for methods for creating singletons for
each numeric type  * Removing NullType's inheritance of PrimitiveType
* Removing intermediate base class between {Integer,Floating}ArrayImpl
and PrimitiveType and just making a concrete template based    class
{Integer,Floating}Array  * Changing FLOAT and DOUBLE to FLOAT32 and
FLOAT64

Author: Joshua Storck <[email protected]>

Closes #55 from joshuastorck/pandas-2.0 and squashes the following commits:

1614d44 [Joshua Storck] * Fixing native.p{xd,yx} so that it matches the latest C++ code * Changing PrimitiveType to NumericType * Removing the macro for declaring sub-types of NumericType and   instead using template arguments * Addding a static SINGLETON member to the NumericType base class   instead of the macros for methods for creating singletons for   each numeric type * Removing NullType's inheritance of PrimitiveType * Removing intermediate base class between {Integer,Floating}ArrayImpl   and PrimitiveType and just making a concrete template based   class {Integer,Floating}Array * Changing FLOAT and DOUBLE to FLOAT32 and FLOAT64

Author: Joshua Storck

pandas/native.pxd

@@ -27,7 +27,7 @@ cdef extern from "<iostream>":
     pass
 
 
-cdef extern from "pandas/status.h" namespace "pandas" nogil:
+cdef extern from "pandas/common.h" namespace "pandas" nogil:
 
     # We can later add more of the common status factory methods as needed
     cdef Status Status_OK "Status::OK"()
@@ -44,28 +44,30 @@ cdef extern from "pandas/status.h" namespace "pandas" nogil:
         c_bool IsUnknownError()
         c_bool IsNotImplemented()
 
-cdef extern from "pandas/api.h" namespace "pandas":
+cdef extern from "pandas/api.h" namespace "pandas::DataType":
 
     enum TypeId:
-        TypeId_NA " pandas::DataType::NA"
-        TypeId_UINT8 " pandas::DataType::UINT8"
-        TypeId_UINT16 " pandas::DataType::UINT16"
-        TypeId_UINT32 " pandas::DataType::UINT32"
-        TypeId_UINT64 " pandas::DataType::UINT64"
-        TypeId_INT8 " pandas::DataType::INT8"
-        TypeId_INT16 " pandas::DataType::INT16"
-        TypeId_INT32 " pandas::DataType::INT32"
-        TypeId_INT64 " pandas::DataType::INT64"
-        TypeId_BOOL " pandas::DataType::BOOL"
-        TypeId_FLOAT " pandas::DataType::FLOAT"
-        TypeId_DOUBLE " pandas::DataType::DOUBLE"
-        TypeId_PYOBJECT " pandas::DataType::PYOBJECT"
-        TypeId_CATEGORY " pandas::DataType::CATEGORY"
-        TypeId_TIMESTAMP " pandas::DataType::TIMESTAMP"
-        TypeId_TIMESTAMP_TZ " pandas::DataType::TIMESTAMP_TZ"
+        TypeId_NA " pandas::DataType::TypeId::NA"
+        TypeId_UINT8 " pandas::DataType::TypeId::UINT8"
+        TypeId_UINT16 " pandas::DataType::TypeId::UINT16"
+        TypeId_UINT32 " pandas::DataType::TypeId::UINT32"
+        TypeId_UINT64 " pandas::DataType::TypeId::UINT64"
+        TypeId_INT8 " pandas::DataType::TypeId::INT8"
+        TypeId_INT16 " pandas::DataType::TypeId::INT16"
+        TypeId_INT32 " pandas::DataType::TypeId::INT32"
+        TypeId_INT64 " pandas::DataType::TypeId::INT64"
+        TypeId_BOOL " pandas::DataType::TypeId::BOOL"
+        TypeId_FLOAT32 " pandas::DataType::TypeId::FLOAT32"
+        TypeId_FLOAT64 " pandas::DataType::TypeId::FLOAT64"
+        TypeId_PYOBJECT " pandas::DataType::TypeId::PYOBJECT"
+        TypeId_CATEGORY " pandas::DataType::TypeId::CATEGORY"
+        TypeId_TIMESTAMP " pandas::DataType::TypeId::TIMESTAMP"
+        TypeId_TIMESTAMP_TZ " pandas::DataType::TypeId::TIMESTAMP_TZ"
+
+cdef extern from "pandas/api.h" namespace "pandas":
 
     cdef cppclass DataType:
-        TypeId type
+        TypeId type()
 
         DataType()
 
@@ -116,8 +118,8 @@ cdef extern from "pandas/api.h" namespace "pandas":
         TypeId type_id()
         size_t length()
 
-        object GetValue(size_t i)
-        void SetValue(size_t i, object val)
+        object GetItem(size_t i)
+        void SetItem(size_t i, object val)
 
     cdef cppclass CCategoryArray" pandas::CategoryArray"(CArray):
         pass

pandas/native.pyx

@@ -209,7 +209,7 @@ cdef class Array:
     cdef inline _getitem(self, size_t i):
         if i >= self.ap.length():
             raise IndexError('Out of bounds: %d' % i)
-        return self.ap.GetValue(i)
+        return self.ap.GetItem(i)
 
     def __setitem__(self, i, val):
         cdef:
@@ -226,7 +226,7 @@ cdef class Array:
     cdef inline _setitem(self, size_t i, object val):
         if i >= self.ap.length():
             raise IndexError('Out of bounds: %d' % i)
-        self.ap.SetValue(i, val)
+        self.ap.SetItem(i, val)
 
     def slice(self, start, end):
         pass
@@ -251,7 +251,7 @@ cdef class Float32Array(FloatingArray):
 
 cdef class BooleanArray(Array):
     cdef:
-        lp.cBooleanArray* inst
+        lp.CBooleanArray* inst
 
     cdef init(self, const ArrayPtr& arr):
         Array.init(self, arr)
@@ -265,7 +265,7 @@ cdef Array wrap_array(const lp.ArrayPtr& arr):
     cdef:
         Array result
 
-    if arr.get().type_enum() == lp.TypeId_CATEGORY:
+    if arr.get().type_id() == lp.TypeId_CATEGORY:
         result = CategoryArray()
     else:
         result = Array()
@@ -280,7 +280,7 @@ cdef PandasType wrap_type(const lp.TypePtr& sp_type):
         lp.DataType* type = sp_type.get()
         PandasType result
 
-    if type.type == lp.TypeId_CATEGORY:
+    if type.type() == lp.TypeId_CATEGORY:
         result = Category()
     else:
         result = PandasType()

src/pandas/array-test.cc

@@ -40,7 +40,7 @@ class TestArray : public ::testing::Test {
 TEST_F(TestArray, Attrs) {
   DoubleType ex_type;
   ASSERT_TRUE(array_->type()->Equals(ex_type));
-  ASSERT_EQ(DataType::DOUBLE, array_->type_id());
+  ASSERT_EQ(DataType::FLOAT64, array_->type_id());
 
   ASSERT_EQ(values_.size(), array_->length());
 }

src/pandas/array.h

@@ -92,4 +92,6 @@ class ArrayView {
   int64_t length_;
 };
 
+using ArrayPtr = std::shared_ptr<Array>;
+
 }  // namespace pandas

src/pandas/dispatch.cc

@@ -22,8 +22,8 @@ Status primitive_type_from_enum(DataType::TypeId tp_enum, DataType** out) {
     MAKE_TYPE_CASE(DataType::UINT16, UInt16);
     MAKE_TYPE_CASE(DataType::UINT32, UInt32);
     MAKE_TYPE_CASE(DataType::UINT64, UInt64);
-    MAKE_TYPE_CASE(DataType::FLOAT, Float);
-    MAKE_TYPE_CASE(DataType::DOUBLE, Double);
+    MAKE_TYPE_CASE(DataType::FLOAT32, Float);
+    MAKE_TYPE_CASE(DataType::FLOAT64, Double);
     MAKE_TYPE_CASE(DataType::BOOL, Boolean);
     MAKE_TYPE_CASE(DataType::PYOBJECT, PyObject);
     default:

src/pandas/numpy_interop.cc

@@ -50,8 +50,8 @@ Status numpy_type_num_to_pandas(int type_num, DataType::TypeId* pandas_type) {
     TYPE_MAP_CASE(UINT16, UINT16);
     TYPE_MAP_CASE(UINT32, UINT32);
     TYPE_MAP_CASE(UINT64, UINT64);
-    TYPE_MAP_CASE(FLOAT32, FLOAT);
-    TYPE_MAP_CASE(FLOAT64, DOUBLE);
+    TYPE_MAP_CASE(FLOAT32, FLOAT32);
+    TYPE_MAP_CASE(FLOAT64, FLOAT64);
     TYPE_MAP_CASE(BOOL, BOOL);
     TYPE_MAP_CASE(OBJECT, PYOBJECT);
     default:

src/pandas/type.cc

@@ -28,4 +28,17 @@ std::string TimestampType::ToString() const {
   return ss.str();
 }
 
+// Constexpr numeric type names
+constexpr const char* UInt8Type::NAME;
+constexpr const char* Int8Type::NAME;
+constexpr const char* UInt16Type::NAME;
+constexpr const char* Int16Type::NAME;
+constexpr const char* UInt32Type::NAME;
+constexpr const char* Int32Type::NAME;
+constexpr const char* UInt64Type::NAME;
+constexpr const char* Int64Type::NAME;
+constexpr const char* FloatType::NAME;
+constexpr const char* DoubleType::NAME;
+constexpr const char* BooleanType::NAME;
+
 }  // namespace pandas

src/pandas/type.h

@@ -32,10 +32,10 @@ class DataType {
     BOOL = 9,
 
     // 4-byte floating point value
-    FLOAT = 10,
+    FLOAT32 = 10,
 
     // 8-byte floating point value
-    DOUBLE = 11,
+    FLOAT64 = 11,
 
     // PyObject*
     PYOBJECT = 12,
@@ -94,72 +94,96 @@ class PANDAS_EXPORT PyObjectType : public DataType {
   std::string ToString() const override;
 };
 
-template <typename Derived>
-class PANDAS_EXPORT PrimitiveType : public DataType {
+template <typename DERIVED, typename C_TYPE, DataType::TypeId TYPE_ID,
+    std::size_t SIZE = sizeof(C_TYPE)>
+class PANDAS_EXPORT NumericType : public DataType {
  public:
-  PrimitiveType() : DataType(Derived::type_enum) {}
+  using c_type = C_TYPE;
+  static constexpr DataType::TypeId type_id = TYPE_ID;
+  static constexpr size_t size = SIZE;
 
-  std::string ToString() const override {
-    return std::string(static_cast<const Derived*>(this)->name());
-  }
+  NumericType() : DataType(type_id) {}
+
+  std::string ToString() const override { return std::string(DERIVED::NAME); }
+
+  static std::shared_ptr<DERIVED> SINGLETON;
 };
 
-#define PRIMITIVE_DECL(TYPENAME, C_TYPE, ENUM, SIZE, NAME)      \
- public:                                                        \
-  typedef C_TYPE c_type;                                        \
-  static constexpr DataType::TypeId type_enum = DataType::ENUM; \
-  static constexpr size_t size = SIZE;                          \
-                                                                \
-  explicit TYPENAME() : PrimitiveType<TYPENAME>() {}            \
-                                                                \
-  static const char* name() { return NAME; }
+template <typename DERIVED, typename C_TYPE, DataType::TypeId TYPE_ID, std::size_t SIZE>
+std::shared_ptr<DERIVED> NumericType<DERIVED, C_TYPE, TYPE_ID, SIZE>::SINGLETON(
+    std::move(std::make_shared<DERIVED>()));
+
+class PANDAS_EXPORT NullType : public DataType {
+ public:
+  NullType() : DataType(DataType::TypeId::NA) {}
 
-class PANDAS_EXPORT NullType : public PrimitiveType<NullType> {
-  PRIMITIVE_DECL(NullType, void, NA, 0, "null");
+  std::string ToString() const override { return std::string("null"); }
 };
 
-class PANDAS_EXPORT UInt8Type : public PrimitiveType<UInt8Type> {
-  PRIMITIVE_DECL(UInt8Type, uint8_t, UINT8, 1, "uint8");
+class PANDAS_EXPORT UInt8Type
+    : public NumericType<UInt8Type, std::uint8_t, DataType::TypeId::UINT8> {
+ public:
+  constexpr static const char* NAME = "uint8";
 };
 
-class PANDAS_EXPORT Int8Type : public PrimitiveType<Int8Type> {
-  PRIMITIVE_DECL(Int8Type, int8_t, INT8, 1, "int8");
+class PANDAS_EXPORT Int8Type
+    : public NumericType<Int8Type, std::int8_t, DataType::TypeId::INT8> {
+ public:
+  constexpr static const char* NAME = "int8";
 };
 
-class PANDAS_EXPORT UInt16Type : public PrimitiveType<UInt16Type> {
-  PRIMITIVE_DECL(UInt16Type, uint16_t, UINT16, 2, "uint16");
+class PANDAS_EXPORT UInt16Type
+    : public NumericType<UInt16Type, std::uint16_t, DataType::TypeId::UINT16> {
+ public:
+  constexpr static const char* NAME = "uint16";
 };
 
-class PANDAS_EXPORT Int16Type : public PrimitiveType<Int16Type> {
-  PRIMITIVE_DECL(Int16Type, int16_t, INT16, 2, "int16");
+class PANDAS_EXPORT Int16Type
+    : public NumericType<Int16Type, std::int16_t, DataType::TypeId::INT16> {
+ public:
+  constexpr static const char* NAME = "int16";
 };
 
-class PANDAS_EXPORT UInt32Type : public PrimitiveType<UInt32Type> {
-  PRIMITIVE_DECL(UInt32Type, uint32_t, UINT32, 4, "uint32");
+class PANDAS_EXPORT UInt32Type
+    : public NumericType<UInt32Type, std::uint32_t, DataType::TypeId::UINT32> {
+ public:
+  constexpr static const char* NAME = "uint32";
 };
 
-class PANDAS_EXPORT Int32Type : public PrimitiveType<Int32Type> {
-  PRIMITIVE_DECL(Int32Type, int32_t, INT32, 4, "int32");
+class PANDAS_EXPORT Int32Type
+    : public NumericType<Int32Type, std::int32_t, DataType::TypeId::INT32> {
+ public:
+  constexpr static const char* NAME = "int32";
 };
 
-class PANDAS_EXPORT UInt64Type : public PrimitiveType<UInt64Type> {
-  PRIMITIVE_DECL(UInt64Type, uint64_t, UINT64, 8, "uint64");
+class PANDAS_EXPORT UInt64Type
+    : public NumericType<UInt64Type, std::uint64_t, DataType::TypeId::UINT64> {
+ public:
+  constexpr static const char* NAME = "uint64";
 };
 
-class PANDAS_EXPORT Int64Type : public PrimitiveType<Int64Type> {
-  PRIMITIVE_DECL(Int64Type, int64_t, INT64, 8, "int64");
+class PANDAS_EXPORT Int64Type
+    : public NumericType<Int64Type, std::int64_t, DataType::TypeId::INT64> {
+ public:
+  constexpr static const char* NAME = "int64";
 };
 
-class PANDAS_EXPORT FloatType : public PrimitiveType<FloatType> {
-  PRIMITIVE_DECL(FloatType, float, FLOAT, 4, "float");
+class PANDAS_EXPORT FloatType
+    : public NumericType<FloatType, float, DataType::TypeId::FLOAT32> {
+ public:
+  constexpr static const char* NAME = "float32";
 };
 
-class PANDAS_EXPORT DoubleType : public PrimitiveType<DoubleType> {
-  PRIMITIVE_DECL(DoubleType, double, DOUBLE, 8, "double");
+class PANDAS_EXPORT DoubleType
+    : public NumericType<DoubleType, double, DataType::TypeId::FLOAT64> {
+ public:
+  constexpr static const char* NAME = "float64";
 };
 
-class PANDAS_EXPORT BooleanType : public PrimitiveType<BooleanType> {
-  PRIMITIVE_DECL(BooleanType, uint8_t, BOOL, 1, "bool");
+class PANDAS_EXPORT BooleanType
+    : public NumericType<BooleanType, std::uint8_t, DataType::TypeId::BOOL> {
+ public:
+  constexpr static const char* NAME = "bool";
 };
 
 }  // namespace pandas