Coverage for mlprodict/onnxrt/ops_cpu/op_einsum.py: 100%

1# -*- encoding: utf-8 -*- 

2# pylint: disable=E0203,E1101,C0111 

3""" 

4@file 

5@brief Runtime operator. 

6""" 

7import numpy 

8from ._op import OpRun 

9from ..shape_object import ShapeObject 

10 

11 

12class Einsum(OpRun): 

13 

14 atts = {'equation': ''} 

15 python_inputs = ['*inputs'] 

16 

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

18 OpRun.__init__(self, onnx_node, desc=desc, 

19 expected_attributes=Einsum.atts, 

20 **options) 

21 if not isinstance(self.equation, (str, bytes)): 

22 raise TypeError( # pragma: no cover 

23 "equation must be string but is %r." % type(self.equation)) 

24 self.equation = self.equation.strip() 

25 if len(self.equation) == 0: 

26 raise TypeError("equation is empty.") # pragma: no cover 

27 

28 def _run(self, *args): # pylint: disable=W0221 

29 try: 

30 return (numpy.einsum(self.equation, *args, optimize=True), ) 

31 except TypeError: 

32 return (numpy.einsum(self.equation, *args), ) 

33 

34 def _infer_shapes(self, *args): # pylint: disable=W0221 

35 try: 

36 return (ShapeObject.einsum_shape(self.equation, *args), ) 

37 except RuntimeError: # pragma: no cover 

38 return (ShapeObject(None, dtype=args[0].dtype), ) 

39 

40 def _infer_types(self, *args): # pylint: disable=W0221 

41 return (args[0], ) 

42 

43 def _infer_sizes(self, *args): # pylint: disable=W0221 

44 res = self.run(*args) 

45 maxi = max(a.size for a in args) 

46 return (dict(temp=maxi * 3 * args[0].dtype.itemsize), ) + res 

47 

48 def to_python(self, inputs): 

49 return ("import numpy", 

50 "return numpy.einsum(equation, *inputs)")