read_ipf_str(): Function fails with ValueError

[dpmayerUSGS]

plio version(s) affected: 1.5.3 - 1.5.4

Description

The normal way to parse the contents of a list of one or more Socet Set IPF files is with plio is via read_ipf(). plio does some magic with singledispatch to parse the input list before passing the name of each IPF file to read_ipf_str() which does the actual parsing. However, read_ipf() balks at any known valid input and emits a ValueError that traces back to read_ipf_str().

How to reproduce

from plio.io.io_bae import read_ipf,
input_data = ['example.ipf']
read_ipf(input_data)

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/usgs/cpkgs/appl/miniconda3_linux/envs/APPL-Tools/lib/python3.9/functools.py", line 888, in wrapper
    return dispatch(args[0].__class__)(*args, **kw)
  File "/usgs/cpkgs/appl/miniconda3_linux/envs/APPL-Tools/lib/python3.9/site-packages/plio/io/io_bae.py", line 103, in read_ipf_str
    d = np.hstack(np.array(d[3:]))
ValueError: setting an array element with a sequence. The requested array has an inhomogeneous shape after 1 dimensions. The detected shape was (3299,) + inhomogeneous part.

Possible Solution

Change this line to d = np.hstack(np.array(d[3:], dtype=object))

I've tested this by writing a custom function that is equivalent to plio's read_ipf_str() but with the above change, and then passing the path of a single IPF to it.

Additional context

It's been a couple years since I've called read_ipf() so I'm not sure why this function suddenly stopped working. Perhaps NumPy got more strict about converting lists to arrays?

Any of the IPFs in https://github.com/DOI-USGS/plio/tree/main/plio/examples/SocetSet should be suitable for testing. I discovered the issue using IPFs associated with a particular project, and reproduced the issue with https://github.com/DOI-USGS/plio/blob/cd781053658934adcf3972b762f314e36a964131/plio/examples/SocetSet/P01_001540_1889_XI_08N204W.ipf