Pyedflib error reading signal

Question

I have a collection of EDF files containing EEG signal data. I am using pyedflib to access the files, however I am frequently having difficulty reading signals from some of the files. Basically, there are a number of files for which I get arrays of all 0s when I try to read the signal values. Given:

def get_sig(fname):
  import pyedflib
  f=pyedflib.EdfReader(fname)
  file_dur=f.getFileDuration()
  fs=int(inFile.getSignalHeader(0)['sample_rate'])
  chan_names=f.getSignalLabels()
  #note... channel names and file duration are captured correctly

  sig=f.readSignal(4)
  return sig

This returns an array of length 'file_dur' * 'fs', however the result is an array of all 0's and the following warning is displayed:

read -1, less than 8965120 requested!!!

Does anyone have any ideas what could cause a problem like this? Unfortunately, I can't share any of the data as it is PHI, but if there is any additional information that could be helpful just ask.

A couple extra notes:

1. I have also tried sig=f.readSignal(4,start=0,n=100) to make sure that it isn't a problem with the beginning of the file or anything of the sort and it returns an array of all 0s of length 100.
2. The values are correct when I examine them in Matlab (i.e. non-zero).
3. The problem appears to be file specific (some are processed correctly while others are not), however I can't find any differences between the files other than the actual signal values
4. The index 4 used in f.readSignal(4) is not hardcoded in my full application, this is simply for demonstration purposes (in my example files, 4 corresponds to EEG channel F4).

Thanks!

P.S. I'm adding this to the pyedflib wiki as well.

Answer 1

Turns out pyedflib doesn't clean up in between files if the same handle is used. The above code will run fine, however when implemented as a loop inside of a driver program, I wasn't explicitly closing the pyedflib file handle. Once I added f.close() after each file it worked fine.