CODEWITHSUNDEEP
pythonarrayspython-2.7numpypython-3.7
Maurizio Mastrofini
Maurizio Mastrofini

Reputation: 67

NumPy `fromstring` function works fine in Python 2.7 but returns error in Python 3.7

I use this syntax to convert the byte array dataword (2 bytes for each sample):

data = numpy.fromstring(dataword, dtype=numpy.int16)

same instruction in Python 3.7 returns the error:

TypeError: fromstring() argument 1 must be read-only bytes-like object, not memoryview

dataword = scope.ReadBinary(rlen-4) #dataword is a byte array, each 2 byte is an integer
data = numpy.fromstring(dataword, dtype=numpy.int16)# data is the int16 array

This is the content of data in Python 2.7.14:

[-1.41601562 -1.42382812 -1.42578125 ...,  1.66992188  1.65234375  1.671875  ]

I am expecting to get the same result with Python 3.7.

How am I supposed to use numpy.fromstring() in 3.7?

Upvotes: 1

Views: 1035

Answers (3)

Maurizio Mastrofini
Maurizio Mastrofini

Reputation: 67

fromstring doesn't work because dataword is a memeoryview not a string, then frombuffer musty be used

data = numpy.frombuffer(dataword.tobytes(), dtype=numpy.int16)

Upvotes: 0

Maurizio Mastrofini
Maurizio Mastrofini

Reputation: 67

simple solution... found reading numpy manual: replace fromstring with frombuffer

data = numpy.frombuffer(dataword, dtype=numpy.int16) works perfectly

Upvotes: 1

mossymountain
mossymountain

Reputation: 183

The TypeError tries to tell you that dataword is of the non-supported type memoryview.
It needs to be passed as an unmutable type, like bytes:

data = numpy.fromstring(dataword.tobytes(), dtype=numpy.int16)

Even better; it seems like scope is a file-like object, so this could also work:

data = numpy.fromfile(scope, dtype=numpy.int16, count=rlen//2-4)

Upvotes: 1

Related Questions