[BUG] An infinite loop and a few crashes

[vyrjana]

First of all, thanks for the continued development of pyDRTtools!

I noticed a few bugs while testing out the latest version (commit 3694b9b).

Infinite loop after clicking the "Hilbert Transform" button.

This is the relevant loop:

pyDRTtools/pyDRTtools/runs.py

Lines 422 to 432 in 3694b9b

	while True:
	try:
	theta_0 = 10**(6*np.random.rand(3, 1)-3)
	out_dict_real = BHT.HT_single_est(theta_0, entry.Z_exp.real, entry.A_re, entry.A_H_im, entry.M, N_freqs, N_taus)
	theta_0 = out_dict_real['theta']
	out_dict_imag = BHT.HT_single_est(theta_0, entry.Z_exp.imag, entry.A_im, entry.A_H_re, entry.M, N_freqs, N_taus)
	break
	except:
	print('Error Occur, Try Another Inital Condition')

The reason is that theta_0 is a numpy.ndarray with more than one dimension, which causes scipy.optimize.minimize to raise a ValueError ('x0' must only have one dimension.) here:

pyDRTtools/pyDRTtools/BHT.py

Line 151 in 3694b9b

res = minimize(NMLL_fct, theta_0, args=(Z_exp, A, M, N_freqs, N_taus), callback=print_results, options={'gtol': 1E-8, 'disp': True})

Aside from making sure that theta_0 is 1D, you might also wish to narrow down the type(s) of exception(s) that are allowed to keep the loop running instead of using a bare except.

Crashes when closing a save file dialog by canceling

The functions related to the "DRT" and "EIS Regression" export buttons do not properly handle the case where the file dialog is closed by canceling:

• pyDRTtools/pyDRTtools/GUI.py

  Line 206 in 3694b9b

  path, ext = QFileDialog.getSaveFileName(None, "Please directory to save the DRT result",

• pyDRTtools/pyDRTtools/GUI.py

  Line 260 in 3694b9b

  path, ext = QFileDialog.getSaveFileName(None, "Please directory to save the EIS fitting result",

If the value assigned to path is an empty string, then a FileNotFoundError will be raised later in the function when trying to write to the path. These crashes only occur if one of the three types of runs (simple, Bayesian, or HT) have been performed due to the if-elif blocks.

The QFileDialog.getSaveFileName returns an empty string if the modal window is closed by canceling, but it can also return an invalid path if the user types in a path (e.g. /tmp/foo/bar on Linux) into the File name input field in the modal window. So, I would recommend something along the lines of:

path, ext = QFileDialog.getSaveFileName(...)
if path == "":
    # The user closed the modal window by canceling.
    return
elif not os.path.isdir(os.path.dirname(path)):
    return
elif os.path.isdir(path):
    return

There might be some edge case that I'm missing here. The user is likely to assume that the file was saved unless they are somehow made aware when the path is invalid. Qt5 has a QErrorMessage class that could be used.

The following instance, which is associated with saving a figure, does not cause a crash, but it does cause a few lines about an empty or null file name to be printed to the terminal:

pyDRTtools/pyDRTtools/GUI.py

Line 304 in 3694b9b

path, ext = QFileDialog.getSaveFileName(self,'Save file', '', file_choices)

Crash when exporting "EIS Regression" results

This crash occurs when experimental data has been loaded but none of the three types of runs have been performed. The reason for this is that the following if-else block assumes that at least a simple or Bayesian run has been performed.

pyDRTtools/pyDRTtools/GUI.py

Lines 263 to 295 in 3694b9b

	if self.data.method == 'BHT': # save result for BHT
	with open(path, 'w', newline='') as save_file:
	writer = csv.writer(save_file)
	writer.writerow(['s_res_re', self.data.out_scores['s_res_re'][0],
	self.data.out_scores['s_res_re'][1],
	self.data.out_scores['s_res_re'][2]])
	writer.writerow(['s_res_im', self.data.out_scores['s_res_im'][0],
	self.data.out_scores['s_res_im'][1],
	self.data.out_scores['s_res_im'][2]])
	writer.writerow(['s_mu_re', self.data.out_scores['s_mu_re']])
	writer.writerow(['s_mu_im', self.data.out_scores['s_mu_im']])
	writer.writerow(['s_HD_re', self.data.out_scores['s_HD_re']])
	writer.writerow(['s_HD_im', self.data.out_scores['s_HD_im']])
	writer.writerow(['s_JSD_re', self.data.out_scores['s_JSD_re']])
	writer.writerow(['s_JSD_im', self.data.out_scores['s_JSD_im']])
	writer.writerow(['freq', 'mu_Z_re','mu_Z_im','mu_H_re','mu_H_im',
	'Z_H_re_band','Z_H_im_band','Z_H_re_res','Z_H_im_res'])
	# save frequency, the fitted impedance and the residual
	for n in range(self.data.freq.shape[0]):
	writer.writerow([self.data.freq[n], self.data.mu_Z_re[n],
	self.data.mu_Z_im[n], self.data.mu_Z_H_im_agm[n],
	self.data.band_re_agm[n], self.data.band_im_agm[n],
	self.data.res_H_re[n], self.data.res_H_im[n]])
	else: # save result for simple and bayesian run
	with open(path, 'w', newline='') as save_file:
	writer = csv.writer(save_file)
	writer.writerow(['freq','mu_Z_re','mu_Z_im','Z_re_res','Z_im_res'])
	# save frequency, the fitted impedance and the residual
	for n in range(self.data.freq.shape[0]):
	writer.writerow([self.data.freq[n], self.data.mu_Z_re[n],
	self.data.mu_Z_im[n], self.data.res_re[n],
	self.data.res_im[n]])

[ciuccislab]

Hello,

Thank you for your interest in pyDRTtools. Instead of using the version from PYPI, could you please git-clone the repository directly and launch the interface accordingly, as the current version will be updated on PYPI later. We are still testing it. I just launched the version from GitHub, and everything works fine. Please let me know if you encounter any other issues."

[ciuccislab]

Crash when exporting "EIS Regression" results

Thank you for testing and calling our attention to this. This will be fixed and updated in the current version before releasing it.

Thank you

[ciuccislab]

Crash when exporting "EIS Regression" results

Could you please provide the data where you encountered this issue so we can test it? I've tested with some data on my end, and everything still works fine. Your assistance in providing the specific data would be greatly appreciated. I look forward to your response. Thank you in advance.

[vyrjana]

I did not install pyDRTtools via PyPI. I cloned this repository (see the first message for the specific commit), created and activated a new virtual environment, installed pyDRTtools as an editable package that points to the cloned repository, and installed all of the required dependencies.

Regarding the last issue (exporting "EIS regression" results); this is very much an edge case that is unlikely to affect most people if they are using the program normally. I only noticed the issue when I started looking for the causes for the other issues I reported.

If you start the program, import some data, press the button for exporting the "EIS Regression" results, and provide a valid file path, then the program should crash when it reaches the following line:

pyDRTtools/pyDRTtools/GUI.py

Line 293 in 3694b9b

writer.writerow([self.data.freq[n], self.data.mu_Z_re[n],

Following these steps, I get the following exception: AttributeError: 'EIS_object' object has no attribute 'mu_Z_re'. If you performed any of the runs (Simple, Bayesian Run, or Hilbert Transform) after importing some data, then the attributes accessed on that line should exist and an exception is not raised.