Skip to content
Advertisement

Load a npz in numpy from bytes

I have a npz file saved from numpy that I can load by using numpy.load(mynpzfile). However, I would like to save this file as a part of a binary file, packed with another file. Something like:

import numpy as np
from PIL import Image
from io import BytesIO

# Saving file jointly
input1 = open('animagefile.png', 'rb').read()
input2 = open('npzfile.npz', 'rb').read()
filesize = len(input1).to_bytes(4, 'big')

output = filesize + input1 + input2

with open('Output.bin', 'wb') as fp:
    fp.write(output)

# Open them
input1 = open('Output.bin', 'rb').read()
filesize2 = int.from_bytes(input1[:4], "big")

segmentation = Image.open(BytesIO(input1[4:4+filesize2])) 
# THIS LINE GIVES ME AN ERROR
layouts = np.frombuffer(input2[4+filesize2:])

However, when reading back the npz I get an error. I have tried load and frombuffer, and both give me an error:

  • frombuffer: ValueError: buffer size must be a multiple of element size
  • load: ValueError: embedded null byte

What can I do?

Advertisement

Answer

The last line should be input1, not input2. Also, guessing from the error message, you forgot to put BytesIO.

layouts = np.load(BytesIO(input1[4+filesize2:]))
                  ^^^^^^^      ^
User contributions licensed under: CC BY-SA
6 People found this is helpful
Advertisement