dataset58 error in writing double real uneven

[cedumoul]

Hi,
I'm facing an issue with writting dataset58 double precision, real, uneven spaced:
The current implentation lead to an error when rem_vals > 0

if rem_vals > 0:
                            fmt = ['%13.5e', '%20.11e', '%13.5e', '%20.11e']
                            fh.write((''.join(fmt[rem_vals]) + '\n') % tuple(data[4 * n4_blocks:]))

It is failing since in fmt[rem_vals] is getting only the nth value in the format list, while it should be a slice until the nth value. I think the implementation should be (tested on my side)

if rem_vals > 0:
                            fmt = ['%13.5e', '%20.11e', '%13.5e', '%20.11e']
                            fh.write((''.join(fmt[:rem_vals]) + '\n') % tuple(data[4 * n4_blocks:]))

best regards,
Cédric

[jankoslavic]

Dear Cedric,

can you provide please a short example test file for this issue?
ttys,
Janko

[cedumoul]

Dear Janko,

Thank you for your reponse.

You'll find here mwe, with the current implementation, I got an error. Running into debug, I could indentify the error as explained above.

import pyuff
uffwrite = pyuff.UFF('my_file.unv')

my_set = pyuff.prepare_58(binary=None, id1='my_test', 
                 func_type=9, ver_num=0, load_case_id=0, 
                 rsp_ent_name='CH1', rsp_node=0, rsp_dir=1, 
                 ord_data_type=4, num_pts=5, abscissa_spacing=0, 
                 abscissa_spec_data_type=18,
                 ordinate_spec_data_type=12,
                 data = [1.0,2.0,2.0,4.0,1.0],
                 x = [20.0,100,300,500,2000],
                 return_full_dict=True)

for key,value in my_set.items():
    if value is None:
        my_set[key] = 0

uffwrite._write_set(my_set,'add')

[jasasonc]

Hi Cédric,

Thanks for reporting this — I recreated the issue and there really seems to be a bug

Root Cause

The issue is on line 996 of dataset_58.py in the
write path for real, double precision, uneven-spaced data.

fmt[rem_vals] indexes a single element instead of slicing the list:

# Bug: fmt[2] returns '%13.5e' (a single string), join is a no-op
fh.write((''.join(fmt[rem_vals]) + '\n') % tuple(data[4 * n4_blocks:]))

Fix

- fh.write((''.join(fmt[rem_vals]) + '\n') % tuple(data[4 * n4_blocks:]))
+ fh.write((''.join(fmt[:rem_vals]) + '\n') % tuple(data[4 * n4_blocks:]))

Verification

• Reproduced the error with the provided MWE
• Applied the fmt[rem_vals] → fmt[:rem_vals] fix
• Write/read succeeds — all 5 data points preserved
• All 55 existing tests pass

@cedumoul Your diagnosis and proposed fix are spot on. I suggest you open a PR!

[jankoslavic]

Thank you @jasasonc , thank you @cedumoul .
This was an easy fix and I have made the PR:
#108
I will now continue with a mini-release.