numpy fromstring behavior (v1.12.1,v1.13.1)

[tonymonk007]

Hi,

I am trying to use np.fromstring to read Hex code into ascii array. However, I get the following result when I expect only 170. Does anybody happen to know where the number "194" is from?

>>>import numpy as np
>>>np.fromstring('\xaa',np.uint8)
array([194, 170], dtype=uint8)

This is the example from the document.

>>> np.fromstring('\x01\x02\x03\x04\x05', dtype=np.uint8, count=3)
array([1, 2, 3], dtype=uint8)

Thanks

[jaimefrio]

It works properly (returns 170 only) with 1.11.1 on Linux, which is what I have available right now.

[tonymonk007]

I tried both 1.12 and 1.13 in Linux. They all give [194,170] instead of 170.

[eric-wieser]

This is python 3, right? There, '\xaa' is a unicode object, so is being utf8-encoded in order to obtain bytes

This doesn't happen on python 2, because quotes create bytes objects

[eric-wieser]

IMO, np.fromstring(unicode) should just throw an error - we should have been using y, not s, in PyArg_ParseTupleAndKeywords

Can we do that, or do we need to deprecate it first?