AttributeError ('NoneType' object has no attribute 'read') after update to 2021.1.0

Builder
1

An experiment that used to run without error in a 2020 version* now crashes, after updating to v2021.1.0. Data collection has not been started yet, as the experiment is still under work.
*: Sadly I no longer know the exact previous version.

This happens at the end of a routine that displays some text and some progress dots appearing progressively on the screen (updated at every frame, as per a variable built inside of a code component).

The traceback states:

  File "d:\Experiments\hSs_behav_lastrun.py", line 738, in <module>
    with contextlib.closing(wave.open(file,'r')) as f:
  File "C:\Program Files\PsychoPy3\lib\wave.py", line 499, in open
    return Wave_read(f)
  File "C:\Program Files\PsychoPy3\lib\wave.py", line 163, in __init__
    self.initfp(f)
  File "C:\Program Files\PsychoPy3\lib\wave.py", line 128, in initfp
    self._file = Chunk(file, bigendian = 0)
  File "C:\Program Files\PsychoPy3\lib\chunk.py", line 61, in __init__
    self.chunkname = file.read(4)
AttributeError: 'NoneType' object has no attribute 'read'

At line 738, the following chunk of code starts (all within the thisPractice loop):

with contextlib.closing(wave.open(file,'r')) as f:
    frames = f.getnframes()
    rate = f.getframerate()
    WAVduration = frames / float(rate)
# keep track of which components have finished
choiceComponents = [cross, notes, phrase, rating_text, choiceRating, punctele_2, brackets_2]
for thisComponent in choiceComponents:
    thisComponent.tStart = None
    thisComponent.tStop = None
    thisComponent.tStartRefresh = None
    thisComponent.tStopRefresh = None
    if hasattr(thisComponent, 'status'):
        thisComponent.status = NOT_STARTED
# reset timers
t = 0
_timeToFirstFrame = win.getFutureFlipTime(clock="now")
choiceClock.reset(-_timeToFirstFrame)  # t0 is time of first possible flip
frameN = -1

I know (as per previous threads 1 and 2) that AttributeError means the output is not the expected object, but I could not figure out where in that code chain/traceback this happens.

I can upload the entire experiment if needed for debugging. Thanks in advance!

OS (e.g. Win10): Win10
PsychoPy version (e.g. 1.84.x): 2021.1.0
Standard Standalone? (y/n) y

2

If you could upload the full experiment that would be great, I’m not 100% sure where this is happening so it would be useful to run it myself so I can click through the error stack

3

Thank you very much Todd, please find the files for the full expt here.

4

Forgot to mention that it’s the ‘feedback’ routine where the error happens.

5

I think the problem is when reading in your conditions file - it seems to think that there’s 16 rows when there’s only 1, but the last 15 are all blank (hence the AttributeError - it’s expecting a filename and trying to access the attributes of a string, but it’s getting None).

As a quick fix, go to your loop and delete the contents of the Conditions field. Then click the Open In Excel button to make a new Excel file and paste the contents (as “Text Only”) of the old practice.xlsx file into it, then save over practice.xlsx. As a more long term solution, I’ll look into why the read function is no longer filtering out blank rows.

6

Good catch, that was it indeed! Funnily enough, the practice.xlsx file I sent was in fact a debug version of the full file, with 1 practice trial instead of 16, which I had created while trying to solve a different problem (to which sadly I got no reaction). It seems a change in the number of rows of the conditions file makes psychoPy stil expect the same number of rows, with this error coming up if that number has changed.

Incidentally, I would point out it’s quite confusing that psychoPy reports the number of rows in a conditions file as “conditions”, since as I understand it, what you normally have across rows are raw trials, a number of which are typically repeated within each (actual) condition.