studyforrest-ppa-analysis/code/onsets2subfolders.py

#!/usr/bin/python3
'''
created on Fri Feb 14 2020
author: Christian Olaf Haeusler
'''

from glob import glob
import argparse
import os
import re
import shutil


def parse_arguments():
    '''
    '''
    parser = argparse.ArgumentParser(
        description='copy confound files to subject-specific directories')

    parser.add_argument('-fmri',
                        default='inputs/studyforrest-data-aligned/sub-??/in_bold3Tp2/sub-??_task-aomovie_run-1_bold.nii.gz',
                        help='''pattern of pathes/files that contains 4D fmri files
                        (e.g. 'inputs/studyforrest-data-aligned/sub-??/in_bold3Tp2/sub-??_task-aomovie_run-1_bold.nii.gz');
                        needed to get subjects for which used 4D data exist''')

    parser.add_argument('-onsets',
                        default='./events/onsets/aomovie/run-?/*.txt',
                        help='''pattern for input path/filenames of\n
                        onsets files (e.g. './events/onsets/aomovie/run-?/*.txt')''')

    parser.add_argument('-o',
                        default='./',
                        help='the directory in which individual subject directories will be created'
                        )



    args = parser.parse_args()

    fmriPattern = args.fmri
    onsetPattern = args.onsets
    outDir = args.o

    return fmriPattern, onsetPattern, outDir


def get_subjects(pattern):
    '''
    searches a given directory (e.g. a directory with functional data) for
    subdirs to get name / number of available subjects

    Output:
        list of subdirectory (str; format 'sub-xy')
    '''
    pathes = sorted(glob(pattern))

    subs = []
    for path in pathes:
        x = re.search(r'sub-\d{2}', path)
        sub = x.group()[-2:]
        subs.append(int(sub))

    return subs


def copy_sources_2_destinations(outDir, sub, stim, sources):
    '''
    '''
    # create the list of destinations
    dests = [os.path.join(outDir, sub, 'onsets', stim, 'run'+ x.split('run')[1])
             for x in sources]
    sourcesDests = zip(sources, dests)

    for source, dest in sourcesDests:
        # create folders if necessary
        try:
            os.makedirs(os.path.dirname(dest))
        except OSError:
            if not os.path.isdir(os.path.dirname(dest)):
                raise
        # copy that shit
        shutil.copyfile(source, dest)

    return None


if __name__ == "__main__":
    fmriPattern, onsetPattern, outDir = parse_arguments()
    # get subject numbers from available subject folders
    subs = get_subjects(fmriPattern)

    regex = re.search(r'[a-z]{2}movie', onsetPattern)
    stim = regex.group()

    onsetFiles = sorted(glob(onsetPattern))

    for sub in subs:
        sub = 'sub-' + str(sub).zfill(2)
        # copy event files for the audio-only movie
        copy_sources_2_destinations(outDir, sub, stim, onsetFiles)