1# -*- encoding: utf-8 -*- 

2# pylint: disable=E0203,E1101,C0111 

3""" 

4@file 

5@brief Runtime operator. 

6""" 

7import numpy 

8from onnx.defs import onnx_opset_version 

9from ._op import OpRunReduceNumpy, RuntimeTypeError, OpRun 

10 

11 

12class ReduceSum_1(OpRunReduceNumpy): 

13 

14 atts = {'axes': [], 'keepdims': 1} 

15 

16 def __init__(self, onnx_node, desc=None, **options): 

17 OpRunReduceNumpy.__init__(self, onnx_node, desc=desc, 

18 expected_attributes=ReduceSum_1.atts, 

19 **options) 

20 

21 def _run(self, data): # pylint: disable=W0221 

22 return (numpy.sum(data, axis=self.axes, 

23 keepdims=self.keepdims, 

24 dtype=data.dtype), ) 

25 

26 

27class ReduceSum_11(ReduceSum_1): 

28 

29 def __init__(self, onnx_node, desc=None, **options): 

30 ReduceSum_1.__init__(self, onnx_node, desc=desc, **options) 

31 

32 

33class ReduceSum_13(OpRunReduceNumpy): 

34 

35 atts = {'axes': [], 'keepdims': 1, 'noop_with_empty_axes': 0} 

36 

37 def __init__(self, onnx_node, desc=None, **options): 

38 OpRunReduceNumpy.__init__(self, onnx_node, desc=desc, 

39 expected_attributes=ReduceSum_13.atts, 

40 **options) 

41 

42 def run(self, data, axes=None): # pylint: disable=E0202,W0221 

43 """ 

44 Calls method ``_run``. 

45 """ 

46 res = self._run(data, axes=axes) 

47 if not self.keepdims and not isinstance(res[0], numpy.ndarray): 

48 res = (numpy.array([res[0]], dtype=res[0].dtype), ) 

49 if res[0].dtype != data.dtype: 

50 raise RuntimeTypeError( # pragma: no cover 

51 "Output type mismatch: input '{}' != output '{}' " 

52 "(operator '{}')".format( 

53 data.dtype, res[0].dtype, self.__class__.__name__)) 

54 return res 

55 

56 def _run_no_checks_(self, x, axes=None): # pylint: disable=W0221 

57 return OpRun.run(self, x, axes) 

58 

59 def _run(self, data, axes=None): # pylint: disable=W0221 

60 if axes is None and self.noop_with_empty_axes: 

61 return (data, ) 

62 if axes is not None and not isinstance(axes, int): 

63 if isinstance(axes, numpy.ndarray) and len(axes.shape) == 0: 

64 axes = int(axes) 

65 else: 

66 axes = tuple(axes) if len(axes) > 0 else None 

67 try: 

68 return (numpy.sum(data, axis=axes, 

69 keepdims=self.keepdims, 

70 dtype=data.dtype), ) 

71 except TypeError as e: # pragma: no cover 

72 raise TypeError( 

73 "Unable to reduce shape %r with axes=%r." % ( 

74 data.shape, axes)) from e 

75 

76 def infer_shapes(self, data, axes=None): # pylint: disable=E0202,W0221 

77 return self._infer_shapes(data, axes=axes) 

78 

79 def _infer_shapes(self, data, axes=None): # pylint: disable=W0221 

80 """ 

81 Returns the same shape by default. 

82 """ 

83 sh = data.reduce(axes, self.keepdims, # pylint: disable=E1101 

84 dtype=numpy.int64) # pylint: disable=E1101 

85 return (sh, ) 

86 

87 

88if onnx_opset_version() >= 13: 

89 ReduceSum = ReduceSum_13 

90elif onnx_opset_version() >= 11: # pragma: no cover 

91 ReduceSum = ReduceSum_11 

92else: # pragma: no cover 

93 ReduceSum = ReduceSum_1