I think there is a small bug in local_force.py that raises an error due to an unforeseen parsing situation in def _parse_gaussian_log(self, file: str | PathLike) -> None: # noqa: C901.

The check in line 873 :

        if (
            not np.array_equal(input_coordinates, standard_coordinates)
            and standard_coordinates.size > 0
        ):

evaluates as True in some logfiles in which standard_coordinates are filled but input_coordinates are not, because the input coordinates are effectively standard coordinates and gaussian output skips the input coordinates block (as far as I understand). I can confirm that we have logfiles for which this happens. Thus, valid gaussian logfiles can be discarded even though there are coordinates available!

The issue is fixed by checking, also, whether input_coordinates is zero, i.e.:

        if (
            not np.array_equal(input_coordinates, standard_coordinates)
            and standard_coordinates.size > 0
            and input_coordinates.size > 0
        ):

but the issue may be present elsewhere. Notably, I suspect that input coordinates must be set to standard coordinates in such cases. Can set up a pull request, but I wanted to raise attention because the issue (if it is indeed an issue) might be apparent elsewhere.

P.d.: I attached an exemplary log file that is compressed because it was larger than 25 mb.

test.tar.gz