为什么在PIL.Image上调整大小/缩略图不起作用？ 蟒蛇

Question

I am trying to resize a PNG image. Worth noting that before the image becomes PNG it is converted from SVG (working without problems).

This is the code:

        if format == 'png':
            output = BytesIO()
            svg2png(bytestring=monkey_image.to_str(), write_to=output)
            contents = output.getvalue()
            size = 15, 15
            img = Image.open(BytesIO(contents))
         #  tried both resize and thumbnail, same result
         #  img.thumbnail(size, Image.ANTIALIAS) 
            img = img.resize(size, Image.ANTIALIAS)
            img.save(output, format="PNG")
            contents = output.getvalue()
            output.close()
        return contents

The result I am seeing is that it returns the image in the original size (retures the first defined contents)

Maybe I am not saving it correctly to the output in the second to?

Answer 1

You are appending your data to an existing BytesIO object. Additional writes to such an object do not replace existing data; writing adds more data to the end of the file:

>>> from io import BytesIO
>>> out = BytesIO()
>>> out.write(b'123')
3
>>> out.getvalue()
b'123'
>>> out.write(b'456')
3
>>> out.getvalue()
b'123456'

You now have binary data for two images in a single file, but compliant decoders will ignore trailing data in an image file.

Use a new, empty BytesIO() object:

resized = BytesIO()
img.save(resized, format="PNG")
contents = resized.getvalue()

You could also seek to the start of the file and truncate:

output.seek(0)
output.truncate()

before writing to it again, but just creating a new in-memory file object is clearer and less error-prone.