OverflowError while loading EDAX SPD datasets from an old µXRF system

[kleinerELM]

Describe the bug

I Get the error "OverflowError: Python int too large to convert to C long" using hyperspy to load an EDAX SPD file under windows
I have a very large (14Gig) EDAX *.SPD map. Furthermore it is propably incomplete, since the measurement stopped inbetween. But it is readable without an issue using the EDAX software. And the same error occurs while loading other big maps.
The error occurs at hs.load(), no matter if I set lazy to True or False

To Reproduce

import hyperspy.api as hs
s = hs.load("ABCD.spd", lazy=True)

If you are interested in the datasets, you can download them here:

https://nextcloud.uni-weimar.de/s/A564WdQD4AZRT7z
pass: hyperspy

Expected behavior

it should load the dataset (at least the complete one) without error.

Python environment:

Python: 3.13.11 | packaged by Anaconda, Inc. | (main, Dec 10 2025, 21:21:58) [MSC v.1929 64 bit (AMD64)]
Architektur: ('64bit', 'WindowsPE')
sys.maxsize: 9223372036854775807
HyperSpy-Version: 2.4.0

[ericpre]

Can you please provide the dimension of the map? This will be useful to figure out what is going on.

[kleinerELM]

The 2D dimension for the incomplete map is 1179x771 px
736 lines (in y direction) were scanned completly one incomplete and the rest is empty - you can see it in the Zn-map:
ABCD_ZnK.BMP

The complete dataset has a resolution of 1068x687 px:

Map_ImgTCnt.Bmp

I am not shure about the energy range - I think it is 40 keV for both maps.

[kleinerELM]

I just tried to open the datasets under a linux installation and get similar errors:
OverflowError: Python integer 14507137070 out of bounds for uint32 and
OverflowError: Python integer 11739457070 out of bounds for uint32
These integers seem to point to the file sizes of the dataset.

Here I used the following system:

Python: 3.14.3 | packaged by conda-forge | (main, Feb  5 2026, 05:49:06) [GCC 14.3.0]
Architektur: ('64bit', 'ELF')
sys.maxsize: 9223372036854775807
HyperSpy-Version: 2.4.0

It seems, that the error message points to the numpy memmap function:

  File "***/.conda/envs/hyperspy/lib/python3.14/site-packages/rsciio/edax/_api.py", line 1003, in file_reader
    return spd_reader(
        filename,
    ...<5 lines>...
        **kwds,
    )
  File "***/.conda/envs/hyperspy/lib/python3.14/site-packages/rsciio/edax/_api.py", line 819, in spd_reader
    np.memmap(f, mode=mode, offset=data_offset, dtype=data_type, **kwds)
    ~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "***/.conda/envs/hyperspy/lib/python3.14/site-packages/numpy/_core/memmap.py", line 248, in __new__
    bytes = flen - offset

[ericpre]

This issue was that the type for the offset parameter was incorrect. This is fixed in #481.