einsum error with unicode input (regression in Numpy 1.14) #10369
Description
This was working before Numpy 1.14 and now fails (on Python 2):
In [1]: from __future__ import unicode_literals
...: import numpy as np
...: np.einsum('...i,...i', [1,2,3], [2,3,4])
...:
---------------------------------------------------------------------------
ValueError Traceback (most recent call last)
<ipython-input-1-dea65e271386> in <module>()
1 from __future__ import unicode_literals
2 import numpy as np
----> 3 np.einsum('...i,...i', [1,2,3], [2,3,4])
/home/conseil/miniconda3/envs/py2/lib/python2.7/site-packages/numpy/core/einsumfunc.pyc in einsum(*operands, **kwargs)
1085 # Build the contraction list and operand
1086 operands, contraction_list = einsum_path(*operands, optimize=optimize_arg,
-> 1087 einsum_call=True)
1088
1089 handle_out = False
/home/conseil/miniconda3/envs/py2/lib/python2.7/site-packages/numpy/core/einsumfunc.pyc in einsum_path(*operands, **kwargs)
686
687 # Python side parsing
--> 688 input_subscripts, output_subscript, operands = _parse_einsum_input(operands)
689 subscripts = input_subscripts + '->' + output_subscript
690
/home/conseil/miniconda3/envs/py2/lib/python2.7/site-packages/numpy/core/einsumfunc.pyc in _parse_einsum_input(operands)
529 if char not in input_subscripts:
530 raise ValueError("Output character %s did not appear in the input"
--> 531 % char)
532
533 # Make sure number operands is equivalent to the number of terms
ValueError: Output character e did not appear in the input