Issue with some uncompressed TRX files

[neurolabusc]

Even though the scripts in this library create uncompressed files, it seems to have problems with some uncompressed files created y other tools, while being fine with compressed files. Note in this example, it succeeds with the compressed "dpsv_9", but has issues with the uncompressed "dpsv_0". NiiVue handles both, and zipinfo suggests identical contents.

$ python tff_convert_tractogram.py dpsv_9.trx dpsv9.tck

$ python tff_convert_tractogram.py dpsv_0.trx dpsv0.tck
Traceback (most recent call last):
  File "/Users/chris/src/trx-python/scripts/tff_convert_tractogram.py", line 59, in <module>
    main()
    ~~~~^^
  File "/Users/chris/src/trx-python/scripts/tff_convert_tractogram.py", line 53, in main
    convert_tractogram(args.in_tractogram, args.out_tractogram, args.reference,
    ~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                       pos_dtype=args.positions_dtype,
                       ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                       offsets_dtype=args.offsets_dtype)
                       ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/chris/py3/lib/python3.13/site-packages/trx/workflows.py", line 97, in convert_tractogram
    sft = trx.to_sft()
  File "/Users/chris/py3/lib/python3.13/site-packages/trx/trx_file_memmap.py", line 1719, in to_sft
    sft = StatefulTractogram(
        deepcopy(self.streamlines),
    ...<3 lines>...
        data_per_streamline=deepcopy(self.data_per_streamline),
    )
  File "/Users/chris/py3/lib/python3.13/site-packages/dipy/testing/decorators.py", line 201, in wrapper
    return convert_positional_to_keyword(func, args, kwargs)
  File "/Users/chris/py3/lib/python3.13/site-packages/dipy/testing/decorators.py", line 192, in convert_positional_to_keyword
    return func(*args, **kwargs)
  File "/Users/chris/py3/lib/python3.13/site-packages/dipy/io/stateful_tractogram.py", line 117, in __init__
    streamlines = streamlines.copy()
  File "/Users/chris/py3/lib/python3.13/site-packages/nibabel/streamlines/array_sequence.py", line 352, in copy
    seq._data[next_offset : next_offset + length] = chunk
    ~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
ValueError: could not broadcast input array from shape (0,3) into shape (192094477,3)

$ zipinfo dpsv_9.trx 
Archive:  dpsv_9.trx
Zip file size: 625723 bytes, number of entries: 5
-rw-r--r--  3.0 unx   383460 bx defN 22-Apr-12 08:44 dpv/z.float32
-rw-r--r--  3.0 unx     1836 bx defN 25-Feb-15 09:29 dps/DataSetID.float32
-rw-r--r--  3.0 unx      195 tx defN 25-Feb-15 09:22 header.json
-rw-r--r--  3.0 unx     3680 bx defN 22-Apr-08 08:47 offsets.uint64
-rw-r--r--  3.0 unx   575190 bx defN 22-Apr-08 08:47 positions.3.float16
5 files, 964361 bytes uncompressed, 624925 bytes compressed:  35.2%

$ zipinfo dpsv_0.trx
Archive:  dpsv_0.trx
Zip file size: 965159 bytes, number of entries: 5
-rw-r--r--  3.0 unx   383460 bx stor 22-Apr-12 08:44 dpv/z.float32
-rw-r--r--  3.0 unx     1836 bx stor 25-Feb-15 09:29 dps/DataSetID.float32
-rw-r--r--  3.0 unx      195 tx stor 25-Feb-15 09:22 header.json
-rw-r--r--  3.0 unx     3680 bx stor 22-Apr-08 08:47 offsets.uint64
-rw-r--r--  3.0 unx   575190 bx stor 22-Apr-08 08:47 positions.3.float16
5 files, 964361 bytes uncompressed, 964361 bytes compressed:  0.0%

TRK.zip