octave: libinterp/octave-value/ov-range.h comparison

comparison libinterp/octave-value/ov-range.h @ 28646:e26201931ea3

new template class for octave_range objects * ov-range.h, ov-range.cc (class octave_range): Convert to template. For now, use specializations to preserve existing behavior for double-precision ranges. Change all uses. * ov.h, ov.cc: Update range constructors to create range objects for integer and float ranges. Provide value extractor functions for range<T> types that forward to virtual functions in the octave_base_value class. * ov-base.cc: Provide virtual value extractor functions for range<T> types. * ov-range-traits.h: New file. * libinterp/octave-value/module.mk: Update. * Range.h, Range.cc (Range::Range): Deprecate public constructors. * pt-eval.h, pt-eval.cc (tree_evaluator::execute_range_loop): New template function. (tree_evaluator::visit_simple_for_command): Use it to handle looping when the loop variable expression is a range.

author	John W. Eaton <jwe@octave.org>
date	Wed, 12 Aug 2020 12:14:17 -0400
parents	7ebe185e3818
children	853e60367cb6

comparison

equal deleted inserted replaced

-:175eedccc085
+:e26201931ea3
 #include "str-vec.h"
 #include "error.h"
 #include "oct-stream.h"
 #include "ov-base.h"
+#include "ov-range-traits.h"
 #include "ov-re-mat.h"
 #include "ov-typeinfo.h"
 class octave_value_list;
 // Range values.
 template <typename T>
-octave::range<double>
-make_double_range (const octave::range<T>& r)
-{
-return octave::range<double> (static_cast<double> (r.base ()),
-static_cast<double> (r.increment ()),
-static_cast<double> (r.limit ()),
-static_cast<double> (r.final_value ()),
-r.numel ());
-}
 class
-octave_range : public octave_base_value
+ov_range : public octave_base_value
 {
 public:
-octave_range (void)
+ov_range (void)
 : octave_base_value (), m_range (), m_idx_cache () { }
-octave_range (const octave::range<float>& r)
+ov_range (const octave::range<T>& r)
-: octave_base_value (), m_range (make_double_range (r)), m_idx_cache ()
+: octave_base_value (), m_range (r), m_idx_cache ()
 {
 if (numel () < 0 && numel () != -2)
 error ("invalid range");
 }
-octave_range (const octave::range<double>& r)
+ov_range (const ov_range<T>& r)
-: octave_base_value (), m_range (r), m_idx_cache ()
-{
-if (numel () < 0 && numel () != -2)
-error ("invalid range");
-}
-octave_range (const octave::range<octave_int8>& r)
-: octave_base_value (), m_range (make_double_range (r)), m_idx_cache ()
-{
-if (numel () < 0 && numel () != -2)
-error ("invalid range");
-}
-octave_range (const octave::range<octave_int16>& r)
-: octave_base_value (), m_range (make_double_range (r)), m_idx_cache ()
-{
-if (numel () < 0 && numel () != -2)
-error ("invalid range");
-}
-octave_range (const octave::range<octave_int32>& r)
-: octave_base_value (), m_range (make_double_range (r)), m_idx_cache ()
-{
-if (numel () < 0 && numel () != -2)
-error ("invalid range");
-}
-octave_range (const octave::range<octave_int64>& r)
-: octave_base_value (), m_range (make_double_range (r)), m_idx_cache ()
-{
-if (numel () < 0 && numel () != -2)
-error ("invalid range");
-}
-octave_range (const octave::range<octave_uint8>& r)
-: octave_base_value (), m_range (make_double_range (r)), m_idx_cache ()
-{
-if (numel () < 0 && numel () != -2)
-error ("invalid range");
-}
-octave_range (const octave::range<octave_uint16>& r)
-: octave_base_value (), m_range (make_double_range (r)), m_idx_cache ()
-{
-if (numel () < 0 && numel () != -2)
-error ("invalid range");
-}
-octave_range (const octave::range<octave_uint32>& r)
-: octave_base_value (), m_range (make_double_range (r)), m_idx_cache ()
-{
-if (numel () < 0 && numel () != -2)
-error ("invalid range");
-}
-octave_range (const octave::range<octave_uint64>& r)
-: octave_base_value (), m_range (make_double_range (r)), m_idx_cache ()
-{
-if (numel () < 0 && numel () != -2)
-error ("invalid range");
-}
-octave_range (double base, double limit, double increment)
-: octave_base_value (), m_range (base, limit, increment), m_idx_cache ()
-{
-if (numel () < 0)
-error ("invalid range");
-}
-octave_range (const Range& r)
-: octave_base_value (),
-m_range (r.base (), r.increment (), r.limit (), r.numel ()),
-m_idx_cache ()
-{
-if (numel () < 0 && numel () != -2)
-error ("invalid range");
-}
-octave_range (const octave_range& r)
 : octave_base_value (), m_range (r.m_range),
 m_idx_cache (r.m_idx_cache ? new idx_vector (*r.m_idx_cache) : nullptr)
 { }
-octave_range (const Range& r, const idx_vector& cache)
+ov_range (const octave::range<T>& r, const idx_vector& cache)
-: octave_base_value (),
+: octave_base_value (), m_range (r), m_idx_cache ()
-m_range (r.base (), r.increment (), r.limit (), r.numel ()),
-m_idx_cache ()
 {
 set_idx_cache (cache);
 }
-~octave_range (void) { clear_cached_info (); }
+// No assignment.
+ov_range& operator = (const ov_range&) = delete;
-octave_base_value * clone (void) const { return new octave_range (*this); }
+~ov_range (void) { clear_cached_info (); }
+octave_base_value * clone (void) const
+{
+return new ov_range (*this);
+}
 // A range is really just a special kind of real matrix object.  In
 // the places where we need to call empty_clone, it makes more sense
 // to create an empty matrix (0x0) instead of an empty range (1x0).
-octave_base_value * empty_clone (void) const { return new octave_matrix (); }
+octave_base_value * empty_clone (void) const
+{
+return new typename octave_value_range_traits<T>::matrix_type ();
+}
 type_conv_info numeric_conversion_function (void) const;
 octave_base_value * try_narrowing_conversion (void);
+builtin_type_t builtin_type (void) const { return class_to_btyp<T>::btyp; }
 // We don't need to override all three forms of subsref.  The using
 // declaration will avoid warnings about partially-overloaded virtual
 // functions.
 using octave_base_value::subsref;
 return dim_vector (n > 0, n);
 }
 octave_idx_type numel (void) const { return m_range.numel (); }
-octave_idx_type nnz (void) const { return m_range.nnz (); }
+octave_idx_type nnz (void) const
+{
+// FIXME: this is a potential waste of memory.
+octave_value tmp (raw_array_value ());
+return tmp.nnz ();
+}
 octave_value resize (const dim_vector& dv, bool fill = false) const;
-size_t byte_size (void) const { return 3 * sizeof (double); }
+size_t byte_size (void) const { return 3 * sizeof (T); }
 octave_value reshape (const dim_vector& new_dims) const
-{ return NDArray (array_value ().reshape (new_dims)); }
+{
+return raw_array_value ().reshape (new_dims);
+}
 octave_value permute (const Array<int>& vec, bool inv = false) const
-{ return NDArray (array_value ().permute (vec, inv)); }
+{
+return raw_array_value ().permute (vec, inv);
+}
 octave_value squeeze (void) const { return m_range; }
-octave_value full_value (void) const { return matrix_value (); }
+octave_value full_value (void) const { return raw_array_value (); }
 bool is_defined (void) const { return true; }
 bool is_constant (void) const { return true; }
 bool is_range (void) const { return true; }
-octave_value all (int dim = 0) const;
+bool is_double_type (void) const { return builtin_type () == btyp_double; }
-octave_value any (int dim = 0) const;
+bool is_single_type (void) const { return builtin_type () == btyp_float; }
-octave_value diag (octave_idx_type k = 0) const;
+bool isfloat (void) const { return btyp_isfloat (builtin_type ()); }
-octave_value diag (octave_idx_type m, octave_idx_type n) const;
+bool is_int8_type (void) const { return builtin_type () == btyp_int8; }
+bool is_int16_type (void) const { return builtin_type () == btyp_int16; }
+bool is_int32_type (void) const { return builtin_type () == btyp_int32; }
+bool is_int64_type (void) const { return builtin_type () == btyp_int64; }
+bool is_uint8_type (void) const { return builtin_type () == btyp_uint8; }
+bool is_uint16_type (void) const { return builtin_type () == btyp_uint16; }
+bool is_uint32_type (void) const { return builtin_type () == btyp_uint32; }
+bool is_uint64_type (void) const { return builtin_type () == btyp_uint64; }
+bool isinteger (void) const
+{
+return btyp_isinteger (builtin_type ());
+}
+bool isreal (void) const { return isfloat (); }
+bool isnumeric (void) const
+{
+return btyp_isnumeric (builtin_type ());
+}
+bool is_true (void) const { return nnz () == numel (); }
+octave_value all (int dim = 0) const
+{
+// FIXME: this is a potential waste of memory.
+typedef typename octave_value_range_traits<T>::matrix_type ov_mx_type;
+typename ov_mx_type::object_type m (raw_array_value ());
+return m.all (dim);
+}
+octave_value any (int dim = 0) const
+{
+// FIXME: this is a potential waste of memory.
+typedef typename octave_value_range_traits<T>::matrix_type ov_mx_type;
+typename ov_mx_type::object_type m (raw_array_value ());
+return m.any (dim);
+}
+octave_value diag (octave_idx_type k = 0) const
+{
+// FIXME: this is a potential waste of memory.
+return m_range.diag (k);
+}
+octave_value diag (octave_idx_type nr, octave_idx_type nc) const
+{
+// FIXME: this is a potential waste of memory.
+typedef typename octave_value_range_traits<T>::matrix_type ov_mx_type;
+typename ov_mx_type::object_type m (raw_array_value ());
+return m.diag (nr, nc);
+}
 octave_value sort (octave_idx_type dim = 0, sortmode mode = ASCENDING) const
 {
-Matrix tmp = matrix_value ();
+Array<T> tmp = raw_array_value ();
 return tmp.sort (dim, mode);
 }
 octave_value sort (Array<octave_idx_type>& sidx, octave_idx_type dim = 0,
 sortmode mode = ASCENDING) const
 {
-Matrix tmp = matrix_value ();
+Array<T> tmp = raw_array_value ();
 return tmp.sort (sidx, dim, mode);
 }
 sortmode issorted (sortmode mode = UNSORTED) const
-{ return m_range.issorted (mode); }
+{
+return m_range.issorted (mode);
+}
 Array<octave_idx_type> sort_rows_idx (sortmode) const
-{ return Array<octave_idx_type> (dim_vector (1, 0)); }
+{
+return Array<octave_idx_type> (dim_vector (1, 0));
+}
 sortmode is_sorted_rows (sortmode mode = UNSORTED) const
-{ return (mode == UNSORTED) ? ASCENDING : mode; }
+{
+return (mode == UNSORTED) ? ASCENDING : mode;
-builtin_type_t builtin_type (void) const { return btyp_double; }
+}
-bool isreal (void) const { return true; }
+Array<T> raw_array_value (void) const { return m_range.array_value (); }
-bool is_double_type (void) const { return true; }
-bool isfloat (void) const { return true; }
-bool isnumeric (void) const { return true; }
-bool is_true (void) const;
 double double_value (bool = false) const;
 float float_value (bool = false) const;
 double scalar_value (bool frc_str_conv = false) const
-{ return double_value (frc_str_conv); }
+{
+return double_value (frc_str_conv);
+}
 float float_scalar_value (bool frc_str_conv = false) const
-{ return float_value (frc_str_conv); }
+{
+return float_value (frc_str_conv);
+}
 Matrix matrix_value (bool = false) const
-{ return m_range.array_value (); }
+{
+return raw_array_value ();
+}
 FloatMatrix float_matrix_value (bool = false) const
-{ return matrix_value (); }
+{
+return raw_array_value ();
+}
 NDArray array_value (bool = false) const
-{ return matrix_value (); }
+{
+return raw_array_value ();
+}
 FloatNDArray float_array_value (bool = false) const
-{ return FloatMatrix (matrix_value ()); }
+{
+return raw_array_value ();
+}
 charNDArray char_array_value (bool = false) const;
 // FIXME: it would be better to have Range::intXNDArray_value
 // functions to avoid the intermediate conversion to a matrix
 // object.
-int8NDArray
+int8NDArray int8_array_value (void) const
-int8_array_value (void) const { return int8NDArray (array_value ()); }
+{
+return raw_array_value ();
-int16NDArray
+}
-int16_array_value (void) const { return int16NDArray (array_value ()); }
+int16NDArray int16_array_value (void) const
-int32NDArray
+{
-int32_array_value (void) const { return int32NDArray (array_value ()); }
+return raw_array_value ();
+}
-int64NDArray
-int64_array_value (void) const { return int64NDArray (array_value ()); }
+int32NDArray int32_array_value (void) const
+{
-uint8NDArray
+return raw_array_value ();
-uint8_array_value (void) const { return uint8NDArray (array_value ()); }
+}
-uint16NDArray
+int64NDArray int64_array_value (void) const
-uint16_array_value (void) const { return uint16NDArray (array_value ()); }
+{
+return raw_array_value ();
-uint32NDArray
+}
-uint32_array_value (void) const { return uint32NDArray (array_value ()); }
+uint8NDArray uint8_array_value (void) const
-uint64NDArray
+{
-uint64_array_value (void) const { return uint64NDArray (array_value ()); }
+return raw_array_value ();
+}
+uint16NDArray uint16_array_value (void) const
+{
+return raw_array_value ();
+}
+uint32NDArray uint32_array_value (void) const
+{
+return raw_array_value ();
+}
+uint64NDArray uint64_array_value (void) const
+{
+return raw_array_value ();
+}
 SparseMatrix sparse_matrix_value (bool = false) const
-{ return SparseMatrix (matrix_value ()); }
+{
+return SparseMatrix (matrix_value ());
+}
 SparseComplexMatrix sparse_complex_matrix_value (bool = false) const
-{ return SparseComplexMatrix (sparse_matrix_value ()); }
+{
+return SparseComplexMatrix (complex_matrix_value ());
+}
 Complex complex_value (bool = false) const;
 FloatComplex float_complex_value (bool = false) const;
 boolNDArray bool_array_value (bool warn = false) const;
 ComplexMatrix complex_matrix_value (bool = false) const
-{ return ComplexMatrix (matrix_value ()); }
+{
+return raw_array_value ();
+}
 FloatComplexMatrix float_complex_matrix_value (bool = false) const
-{ return FloatComplexMatrix (matrix_value ()); }
+{
+return raw_array_value ();
+}
 ComplexNDArray complex_array_value (bool = false) const
-{ return ComplexMatrix (matrix_value ()); }
+{
+return raw_array_value ();
+}
 FloatComplexNDArray float_complex_array_value (bool = false) const
-{ return FloatComplexMatrix (matrix_value ()); }
+{
+return raw_array_value ();
-octave::range<double> range_value (void) const { return m_range; }
+}
+octave::range<float> float_range_value (void) const;
+octave::range<double> range_value (void) const;
+octave::range<octave_int8> int8_range_value (void) const;
+octave::range<octave_int16> int16_range_value (void) const;
+octave::range<octave_int32> int32_range_value (void) const;
+octave::range<octave_int64> int64_range_value (void) const;
+octave::range<octave_uint8> uint8_range_value (void) const;
+octave::range<octave_uint16> uint16_range_value (void) const;
+octave::range<octave_uint32> uint32_range_value (void) const;
+octave::range<octave_uint64> uint64_range_value (void) const;
 octave_value convert_to_str_internal (bool pad, bool force, char type) const;
 octave_value as_double (void) const;
 octave_value as_single (void) const;
 bool save_binary (std::ostream& os, bool save_as_floats);
 bool load_binary (std::istream& is, bool swap,
 octave::mach_info::float_format fmt);
-bool save_hdf5 (octave_hdf5_id loc_id, const char *name, bool save_as_floats);
+bool save_hdf5 (octave_hdf5_id loc_id, const char *name, bool flag);
 bool load_hdf5 (octave_hdf5_id loc_id, const char *name);
 int write (octave::stream& os, int block_size,
 oct_data_conv::data_type output_type, int skip,
 mxArray * as_mxArray (bool interleaved) const;
 octave_value map (unary_mapper_t umap) const
 {
-octave_matrix m (matrix_value ());
+octave_value tmp (raw_array_value ());
-return m.map (umap);
+return tmp.map (umap);
 }
 octave_value fast_elem_extract (octave_idx_type n) const;
-private:
+protected:
-octave::range<double> m_range;
+octave::range<T> m_range;
 idx_vector set_idx_cache (const idx_vector& idx) const
 {
 delete m_idx_cache;
 m_idx_cache = (idx ? new idx_vector (idx) : nullptr);
 delete m_idx_cache; m_idx_cache = nullptr;
 }
 mutable idx_vector *m_idx_cache;
-// No assignment.
+static octave_hdf5_id hdf5_save_type;
-octave_range& operator = (const octave_range&);
 DECLARE_OV_TYPEID_FUNCTIONS_AND_DATA
 };
-typedef octave_range octave_float_range;
+DECLARE_TEMPLATE_OV_TYPEID_SPECIALIZATIONS (ov_range, float)
-typedef octave_range octave_double_range;
+DECLARE_TEMPLATE_OV_TYPEID_SPECIALIZATIONS (ov_range, double)
+DECLARE_TEMPLATE_OV_TYPEID_SPECIALIZATIONS (ov_range, octave_int8)
-typedef octave_range octave_int8_range;
+DECLARE_TEMPLATE_OV_TYPEID_SPECIALIZATIONS (ov_range, octave_int16)
-typedef octave_range octave_int16_range;
+DECLARE_TEMPLATE_OV_TYPEID_SPECIALIZATIONS (ov_range, octave_int32)
-typedef octave_range octave_int32_range;
+DECLARE_TEMPLATE_OV_TYPEID_SPECIALIZATIONS (ov_range, octave_int64)
-typedef octave_range octave_int64_range;
+DECLARE_TEMPLATE_OV_TYPEID_SPECIALIZATIONS (ov_range, octave_uint8)
+DECLARE_TEMPLATE_OV_TYPEID_SPECIALIZATIONS (ov_range, octave_uint16)
-typedef octave_range octave_uint8_range;
+DECLARE_TEMPLATE_OV_TYPEID_SPECIALIZATIONS (ov_range, octave_uint32)
-typedef octave_range octave_uint16_range;
+DECLARE_TEMPLATE_OV_TYPEID_SPECIALIZATIONS (ov_range, octave_uint64)
-typedef octave_range octave_uint32_range;
-typedef octave_range octave_uint64_range;
+// Specializations.
+template <>
+octave::range<float>
+ov_range<float>::float_range_value (void) const;
+template <>
+octave::range<double>
+ov_range<double>::range_value (void) const;
+template <>
+octave::range<octave_int8>
+ov_range<octave_int8>::int8_range_value (void) const;
+template <>
+octave::range<octave_int16>
+ov_range<octave_int16>::int16_range_value (void) const;
+template <>
+octave::range<octave_int32>
+ov_range<octave_int32>::int32_range_value (void) const;
+template <>
+octave::range<octave_int64>
+ov_range<octave_int64>::int64_range_value (void) const;
+template <>
+octave::range<octave_uint8>
+ov_range<octave_uint8>::uint8_range_value (void) const;
+template <>
+octave::range<octave_uint16>
+ov_range<octave_uint16>::uint16_range_value (void) const;
+template <>
+octave::range<octave_uint32>
+ov_range<octave_uint32>::uint32_range_value (void) const;
+template <>
+octave::range<octave_uint64>
+ov_range<octave_uint64>::uint64_range_value (void) const;
+// The following specializations are here to preserve previous Range
+// performance until solutions can be generalized for other types.
+template <>
+idx_vector
+ov_range<double>::index_vector (bool require_integers) const;
+template <>
+octave_idx_type
+ov_range<double>::nnz (void) const;
+// The following specialization is also historical baggage.  For double
+// ranges, we can produce special double-valued diagnoal matrix objects
+// but Octave currently provides only double and Complex diagonal matrix
+// objects.
+template <>
+octave_value
+ov_range<double>::diag (octave_idx_type k) const;
+template <>
+octave_value
+ov_range<double>::diag (octave_idx_type nr, octave_idx_type nc) const;
+template <>
+void
+ov_range<double>::print_raw (std::ostream& os, bool pr_as_read_syntax) const;
+typedef ov_range<float> octave_float_range;
+typedef ov_range<double> octave_double_range;
+typedef ov_range<octave_int8> octave_int8_range;
+typedef ov_range<octave_int16> octave_int16_range;
+typedef ov_range<octave_int32> octave_int32_range;
+typedef ov_range<octave_int64> octave_int64_range;
+typedef ov_range<octave_uint8> octave_uint8_range;
+typedef ov_range<octave_uint16> octave_uint16_range;
+typedef ov_range<octave_uint32> octave_uint32_range;
+typedef ov_range<octave_uint64> octave_uint64_range;
+typedef octave_double_range octave_range;
 #endif

Mercurial > octave

comparison libinterp/octave-value/ov-range.h @ 28646:e26201931ea3