Cannot work out file type of ".nii"

[DimitriPapadopoulos]

While this is a corner case, nibabel should properly load .nii. instead of raising an exception.

>>> import nibabel
>>> nibabel.load('.nii')
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/usr/lib/python3/dist-packages/nibabel/loadsave.py", line 49, in load
    filename)
nibabel.filebasedimages.ImageFileError: Cannot work out file type of ".nii"

[DimitriPapadopoulos]

This happens because path_maybe_image() does not recognize .nii as an image file:

>>> import nibabel
>>> nibabel.filebasedimages.FileBasedImage.path_maybe_image('.nii')
(False, None)

and in turn that's because splitext_addext() fails and outputs:

>>> nibabel.filename_parser.splitext_addext('.nii')
('.nii', '', '')

instead of:

>>> nibabel.filename_parser.splitext_addext('.nii')
('', '.nii', '')

[DimitriPapadopoulos]

And finally that's because os.path.splitext() does not work as expected:

>>> os.path.splitext('.nii')
('.nii', '')

while you expect:

>>> os.path.splitext('.nii')
('', '.nii')

[effigies]

As a point of information, you can always load such a file with nibabel.Nifti1Image.from_filename. And just out of curiosity, under what circumstances are you finding a file without a name?

And finally that's because os.path.splitext() does not work as expected:

>>> os.path.splitext('.nii')
('.nii', '')

This is what I would expect. Initial dots are considered part of the filename, not an extension.

[DimitriPapadopoulos]

Happens with output of dcm2nii -d N -e N -p Y in some edge cases.

[DimitriPapadopoulos]

I know about the behavior of os.path.splitext() but I claim this is not the behavior we need in this context.

[DimitriPapadopoulos]

On the other hand I can understand why you wouldn't want to deviate from the standard os.path.splitext() behavior. However while I almost always agree with choices made in Python libraries, I find the choices made in os.path.splitext() arbitrary - an optional argument would be welcome to switch behavior in edge cases depending on the context.

[lanranranM]

As a point of information, you can always load such a file with nibabel.Nifti1Image.from_filename. And just out of curiosity, under what circumstances are you finding a file without a name?

And finally that's because os.path.splitext() does not work as expected:

>>> os.path.splitext('.nii')
('.nii', '')

This is what I would expect. Initial dots are considered part of the filename, not an extension.

Hello, I'm using nib.Nifti1Image.from_filename(file_path) to read the .nii.gz file. Referring another issue post, I renamed the file without .gz and the file_path without .gz. However, it's now rasing WrapStructError('Binary block is wrong size') nibabel.wrapstruct.WrapStructError: Binary block is wrong size. How can I sove it?