wonambi.ioeeg.wonambi module

Package to import and export common formats.

class wonambi.ioeeg.wonambi.Wonambi(filename)[source]

Bases: object

Class to read the data in Wonambi format, which is fast to write and read

Parameters:

filename (path to file) – the name of the filename with extension .won

return_dat(chan, begsam, endsam)[source]

Return the data as 2D numpy.ndarray.

Parameters:

  • chan (int or list) – index (indices) of the channels to read

  • begsam (int) – index of the first sample

  • endsam (int) – index of the last sample

Returns:

numpy.ndarray – A 2d matrix, with dimension chan X samples. To save memory, the data are memory-mapped, and you cannot change the values on disk.

Raises:

FileNotFoundError – if .dat file is not in the same directory, with the same name.

Notes

When asking for an interval outside the data boundaries, it returns NaN for those values. It then converts the memmap to a normal numpy array, I think, and so it reads the data into memory. However, I’m not 100% sure that this is what happens.

return_hdr()[source]

Return the header for further use.

Returns:

  • subj_id (str) – subject identification code

  • start_time (datetime) – start time of the dataset

  • s_freq (float) – sampling frequency

  • chan_name (list of str) – list of all the channels

  • n_samples (int) – number of samples in the dataset

  • orig (dict) – the json file

return_markers()[source]

This format doesn’t have markers.

Returns:

empty list

Raises:

FileNotFoundError – when it cannot read the events for some reason (don’t use other exceptions).

wonambi.ioeeg.wonambi.write_wonambi(data, filename, subj_id='', dtype='float64')[source]

Write file in simple Wonambi format.

Parameters:

  • data (instance of ChanTime) – data with only one trial

  • filename (path to file) – file to export to (the extensions .won and .dat will be added)

  • subj_id (str) – subject id

  • dtype (str) – numpy dtype in which you want to save the data

Notes

Wonambi format creates two files, one .won with the dataset info as json file and one .dat with the memmap recordings.

It will happily overwrite any existing file with the same name.

Memory-mapped matrices are column-major, Fortran-style, to be compatible with Matlab.