Remove Python 2 compatibility code (cvxpy#2527)

* Remove Python 2 compatibility code

* Remove duplicate ABCs

* Fix for real abstract classes

* Some more abc fixes

* Add test for ABCs

* Remove unused imports

cvxpy/atoms/affine/affine_atom.py

@@ -13,7 +13,6 @@
 See the License for the specific language governing permissions and
 limitations under the License.
 """
-import abc
 from typing import Any, List, Tuple
 
 import scipy.sparse as sp
@@ -29,7 +28,6 @@
 
 class AffAtom(Atom):
     """ Abstract base class for affine atoms. """
-    __metaclass__ = abc.ABCMeta
     _allow_complex = True
 
     def sign_from_args(self) -> Tuple[bool, bool]:

cvxpy/atoms/affine/binary_operators.py

@@ -14,8 +14,6 @@
 limitations under the License.
 """
 
-from __future__ import division
-
 import operator as op
 from functools import reduce
 from typing import List, Tuple

cvxpy/atoms/atom.py

@@ -35,7 +35,6 @@
 
 class Atom(Expression):
     """ Abstract base class for atoms. """
-    __metaclass__ = abc.ABCMeta
     _allow_complex = False
     # args are the expressions passed into the Atom constructor.
 

cvxpy/atoms/axis_atom.py

@@ -14,7 +14,6 @@
 limitations under the License.
 """
 
-import abc
 from typing import List, Optional, Tuple
 
 import numpy as np
@@ -28,8 +27,6 @@ class AxisAtom(Atom):
     An abstract base class for atoms that can be applied along an axis.
     """
 
-    __metaclass__ = abc.ABCMeta
-
     def __init__(self, expr, axis: Optional[int] = None, keepdims: bool = False) -> None:
         self.axis = axis
         self.keepdims = keepdims

cvxpy/atoms/elementwise/elementwise.py

@@ -14,7 +14,6 @@
 limitations under the License.
 """
 
-import abc
 from typing import Tuple
 
 import numpy as np
@@ -27,7 +26,6 @@
 
 class Elementwise(Atom):
     """ Abstract base class for elementwise atoms. """
-    __metaclass__ = abc.ABCMeta
 
     def shape_from_args(self) -> Tuple[int, ...]:
         """Shape is the same as the sum of the arguments.

cvxpy/atoms/elementwise/kl_div.py

@@ -14,8 +14,6 @@
 limitations under the License.
 """
 
-from __future__ import division
-
 from typing import List, Optional, Tuple
 
 import numpy as np

cvxpy/atoms/elementwise/maximum.py

@@ -14,16 +14,13 @@
 limitations under the License.
 """
 
-import sys
+from functools import reduce
 from typing import Any, List, Tuple
 
 import numpy as np
 
 from cvxpy.atoms.elementwise.elementwise import Elementwise
 
-if sys.version_info >= (3, 0):
-    from functools import reduce
-
 
 class maximum(Elementwise):
     """Elementwise maximum of a sequence of expressions.

cvxpy/atoms/elementwise/minimum.py

@@ -13,16 +13,13 @@
 limitations under the License.
 """
 
-import sys
+from functools import reduce
 from typing import Any, List, Tuple
 
 import numpy as np
 
 from cvxpy.atoms.elementwise.elementwise import Elementwise
 
-if sys.version_info >= (3, 0):
-    from functools import reduce
-
 
 class minimum(Elementwise):
     """Elementwise minimum of a sequence of expressions.

cvxpy/atoms/elementwise/rel_entr.py

@@ -14,8 +14,6 @@
 limitations under the License.
 """
 
-from __future__ import division
-
 from typing import List, Optional, Tuple
 
 import numpy as np

cvxpy/atoms/perspective.py

@@ -136,3 +136,17 @@ def shape_from_args(self) -> Tuple[int, ...]:
         """Returns the (row, col) shape of the expression.
         """
         return self.f.shape
+
+    def _grad(self, values):
+        """Gives the (sub/super)gradient of the atom w.r.t. each argument.
+
+        Matrix expressions are vectorized, so the gradient is a matrix.
+
+        Args:
+            values: A list of numeric values for the arguments.
+
+        Returns:
+            A list of SciPy CSC sparse matrices or None.
+        """
+        # TODO
+        raise NotImplementedError()

cvxpy/atoms/quad_form.py

@@ -14,8 +14,6 @@
 limitations under the License.
 """
 
-from __future__ import division
-
 import warnings
 from typing import Tuple
 

cvxpy/constraints/cones.py

@@ -14,8 +14,6 @@
 limitations under the License.
 """
 
-import abc
-
 from cvxpy.constraints.constraint import Constraint
 
 
@@ -40,8 +38,6 @@ class Cone(Constraint):
         A unique id for the constraint.
     """
 
-    __metaclass__ = abc.ABCMeta
-
     def __init__(self, args, constr_id=None) -> None:
         super(Cone, self).__init__(args, constr_id)
 

cvxpy/constraints/constraint.py

@@ -38,8 +38,6 @@ class Constraint(u.Canonical):
         A unique id for the constraint.
     """
 
-    __metaclass__ = abc.ABCMeta
-
     def __init__(self, args, constr_id=None) -> None:
         # TODO cast constants.
         # self.args = [cvxtypes.expression().cast_to_const(arg) for arg in args]
@@ -129,7 +127,8 @@ def is_dpp(self, context='dcp') -> bool:
         else:
             raise ValueError("Unsupported context ", context)
 
-    @abc.abstractproperty
+    @property
+    @abc.abstractmethod
     def residual(self):
         """The residual of the constraint.
 
@@ -213,16 +212,6 @@ def get_data(self):
         """
         return [self.id]
 
-    def __nonzero__(self):
-        """Raises an exception when called.
-
-        Python 2 version.
-
-        Called when evaluating the truth value of the constraint.
-        Raising an error here prevents writing chained constraints.
-        """
-        return self._chain_constraints()
-
     def _chain_constraints(self):
         """Raises an error due to chained constraints.
         """

cvxpy/expressions/expression.py

@@ -119,8 +119,6 @@ class Expression(u.Canonical):
     expressions (e.g., the sum of two expressions) and constraints.
     """
 
-    __metaclass__ = abc.ABCMeta
-
     # Handles arithmetic operator overloading with Numpy.
     __array_priority__ = 100
 

cvxpy/expressions/leaf.py

@@ -15,7 +15,6 @@
 """
 from __future__ import annotations
 
-import abc
 from typing import TYPE_CHECKING, Iterable
 
 if TYPE_CHECKING:
@@ -95,8 +94,6 @@ class Leaf(expression.Expression):
         An iterable of length two specifying lower and upper bounds.
     """
 
-    __metaclass__ = abc.ABCMeta
-
     def __init__(
         self, shape: int | tuple[int, ...], value=None, nonneg: bool = False,
         nonpos: bool = False, complex: bool = False, imag: bool = False,

cvxpy/expressions/variable.py

@@ -29,7 +29,7 @@ class Variable(Leaf):
     """
 
     def __init__(
-        self, shape: int | Iterable[int, ...] = (), name: str | None = None,
+        self, shape: int | Iterable[int] = (), name: str | None = None,
         var_id: int | None = None, **kwargs: Any
     ):
         if var_id is None:

cvxpy/interface/base_matrix_interface.py

@@ -21,12 +21,11 @@
 import cvxpy.interface.matrix_utilities
 
 
-class BaseMatrixInterface:
+class BaseMatrixInterface(metaclass=abc.ABCMeta):
     """
     An interface between constants' internal values
     and the target matrix used internally.
     """
-    __metaclass__ = abc.ABCMeta
 
     @abc.abstractmethod
     def const_to_matrix(self, value, convert_scalars: bool = False):

cvxpy/performance_tests/test_warmstart.py

@@ -18,8 +18,6 @@
 DO NOT CALL THESE FUNCTIONS IN YOUR CODE!
 """
 
-from __future__ import print_function
-
 import time
 import unittest
 

cvxpy/problems/param_prob.py

@@ -16,20 +16,19 @@
 import abc
 
 
-class ParamProb:
+class ParamProb(metaclass=abc.ABCMeta):
     """An abstract base class for parameterized problems.
 
     Parameterized problems are produced during the first canonicalization
     and allow canonicalization to be short-circuited for future solves.
     """
-    __metaclass__ = abc.ABCMeta
 
-    @abc.abstractproperty
+    @abc.abstractmethod
     def is_mixed_integer(self) -> bool:
         """Is the problem mixed-integer?"""
         raise NotImplementedError()
 
-    @abc.abstractproperty
+    @abc.abstractmethod
     def apply_parameters(self, id_to_param_value=None, zero_offset: bool = False,
                          keep_zeros: bool = False):
         """Returns A, b after applying parameters (and reshaping).

cvxpy/reductions/matrix_stuffing.py

@@ -15,7 +15,6 @@
 """
 
 
-import abc
 from typing import List, Optional, Tuple
 
 import numpy as np
@@ -106,8 +105,6 @@ def ravel_multi_index(multi_index, x, vert_offset):
 class MatrixStuffing(Reduction):
     """Stuffs a problem into a standard form for a family of solvers."""
 
-    __metaclass__ = abc.ABCMeta
-
     def apply(self, problem) -> None:
         """Returns a stuffed problem.
 

cvxpy/reductions/reduction.py

@@ -17,7 +17,7 @@
 from abc import ABCMeta, abstractmethod
 
 
-class Reduction:
+class Reduction(metaclass=ABCMeta):
     """Abstract base class for reductions.
 
     A reduction is an actor that transforms a problem into an
@@ -46,8 +46,6 @@ class Reduction:
         A problem owned by this reduction; possibly None.
     """
 
-    __metaclass__ = ABCMeta
-
     def __init__(self, problem=None) -> None:
         """Construct a reduction for reducing `problem`.
 

cvxpy/reductions/solvers/conic_solvers/conic_solver.py

@@ -165,7 +165,8 @@ def psd_format_mat(constr):
         # Default is identity.
         return sp.eye(constr.size, format='csc')
 
-    def format_constraints(self, problem, exp_cone_order):
+    @classmethod
+    def format_constraints(cls, problem, exp_cone_order):
         """
         Returns a ParamConeProg whose problem data tensors will yield the
         coefficient "A" and offset "b" for the constraint in the following
@@ -249,7 +250,7 @@ def format_constraints(self, problem, exp_cone_order):
                     arg_mats.append(space_mat)
                 restruct_mat.append(sp.hstack(arg_mats))
             elif type(constr) == PSD:
-                restruct_mat.append(self.psd_format_mat(constr))
+                restruct_mat.append(cls.psd_format_mat(constr))
             else:
                 raise ValueError("Unsupported constraint type.")
 

cvxpy/reductions/solvers/solver.py

@@ -30,8 +30,6 @@ class Solver(Reduction):
     #   There are separate ConeDims classes for cone programs vs QPs.
     #   See cone_matrix_stuffing.py and qp_matrix_stuffing.py for details.
 
-    __metaclass__ = abc.ABCMeta
-
     # Solver capabilities.
     MIP_CAPABLE = False
     BOUNDED_VARIABLES = False

cvxpy/tests/test_base_classes.py

@@ -0,0 +1,27 @@
+import inspect
+
+import pytest
+
+from cvxpy.atoms.affine.affine_atom import AffAtom
+from cvxpy.atoms.atom import Atom
+from cvxpy.constraints.constraint import Constraint
+from cvxpy.expressions.expression import Expression
+from cvxpy.expressions.leaf import Leaf
+from cvxpy.interface.base_matrix_interface import BaseMatrixInterface
+from cvxpy.problems.param_prob import ParamProb
+from cvxpy.reductions.reduction import Reduction
+from cvxpy.reductions.solvers.conic_solvers.conic_solver import ConicSolver
+from cvxpy.reductions.solvers.solver import Solver
+from cvxpy.utilities.canonical import Canonical
+
+
+@pytest.mark.parametrize("expected_abc", [
+    Canonical,
+    Expression, Atom, AffAtom, Leaf,
+    Constraint,
+    Reduction, Solver, ConicSolver,
+    ParamProb,
+    BaseMatrixInterface,
+])
+def test_is_abstract(expected_abc):
+    assert inspect.isabstract(expected_abc)

cvxpy/tests/test_cone2cone.py

@@ -49,7 +49,7 @@ def simulate_chain(in_prob):
 
         # Dualize the problem, reconstruct a high-level cvxpy problem for the dual.
         # Solve the problem, invert the dualize reduction.
-        cone_prog = ConicSolver().format_constraints(cone_prog, exp_cone_order=[0, 1, 2])
+        cone_prog = ConicSolver.format_constraints(cone_prog, exp_cone_order=[0, 1, 2])
         data, inv_data = a2d.Dualize.apply(cone_prog)
         A, b, c, K_dir = data[s.A], data[s.B], data[s.C], data['K_dir']
         y = cp.Variable(shape=(A.shape[1],))
@@ -205,7 +205,7 @@ def simulate_chain(in_prob, affine, **solve_kwargs):
 
         # apply the Slacks reduction, reconstruct a high-level problem,
         # solve the problem, invert the reduction.
-        cone_prog = ConicSolver().format_constraints(cone_prog, exp_cone_order=[0, 1, 2])
+        cone_prog = ConicSolver.format_constraints(cone_prog, exp_cone_order=[0, 1, 2])
         data, inv_data = a2d.Slacks.apply(cone_prog, affine)
         G, h, f, K_dir, K_aff = data[s.A], data[s.B], data[s.C], data['K_dir'], data['K_aff']
         G = sp.sparse.csc_matrix(G)