Bug in read_txrm

[Skielex]

Hey,

I came across a bug in the read_txrm function related to slicing, as well as a type issue. I'll start with the bug. If you try to read out a portion of the images you get an exception due to a bug in the code.

Example:

reader.read_txm(file_path, slice_range=(slice(10, 11, None), slice(None), slice(None)))

Throws:

---------------------------------------------------------------------------
IndexError                                Traceback (most recent call last)
<ipython-input-4-7cc0a3c63f99> in <module>
----> 1 reader.read_txm(file_path, slice_range=(slice(10, 11, None), slice(None), slice(None)))

C:\Anaconda\envs\py37\lib\site-packages\dxchange\reader.py in read_txm(file_name, slice_range)
    357     """
    358 
--> 359     return read_txrm(file_name, slice_range)
    360 
    361 

C:\Anaconda\envs\py37\lib\site-packages\dxchange\reader.py in read_txrm(file_name, slice_range)
    322         img_string = "ImageData{}/Image{}".format(
    323             int(np.ceil((i + 1) / 100.0)), int(i + 1))
--> 324         array_of_images[i] = _read_ole_image(ole, img_string, metadata)[slice_range[1:]]
    325 
    326     reference = metadata['reference']

IndexError: index 10 is out of bounds for axis 0 with size 1

This issue is that the index i is 10 while the size of the array_of_images is just one in this case. Because of this it is not possible to use slices on the first axis, except if you start at 0 and step with 1.

There's also a smaller issue related to the data type. The array_of_images is always created using float32, so even though the _read_ole_image function correctly returns data as uint16 in my case, it is changed changed to float32 before being returned, which is actually very significant when working with +30GB data files.