task 1

task1.py

'''
Python Training
https://microsoft.sharepoint.com/teams/stca/ipe/sr/_layouts/15/WopiFrame.aspx?sourcedoc={57bcddd2-341f-4151-84f7-f332fca4d07a}&action=edit&wd=target%28LearningCorner%2Eone%7CDF7F96A9-2186-462F-A0EC-8772881176AA%2FPerl%20and%20C%23%20Training%7C1C6AD0B9-765A-4DD5-A3C7-0822927FC9D1%2F%29
'''
import xml.etree.ElementTree as ET
from operator import itemgetter
import argparse

SRC_FILE_NAME = 'ITA_Blind_R2.xml'
DES_FILE_NAME = 'task1_output.txt'

def run(src, des):
    '''
    start program
    '''
    speakers = {}
    tree = ET.ElementTree(file=src)
    for elem in tree.iter('SpeechIn'):
        speaker, session, wav = elem.attrib.get('WaveFilePath').split('\\')
        if speaker not in speakers:
            speakers[speaker] = []
        speakers[speaker].append((session, wav))

    speakers = speakers.items()

    speakers = sorted(speakers, key=itemgetter(0))

    with open(des, 'w') as fp:
        for speaker, context in speakers:
            fp.write(speaker + '\n')
            for session, wav in context:
                fp.write('\\'.join((speaker, session, wav)) + '\n')


if __name__ == '__main__':
    parser = argparse.ArgumentParser(description='process file')
    parser.add_argument('--src', type=str, default=SRC_FILE_NAME, help='source file path')
    parser.add_argument('--des', type=str, default=DES_FILE_NAME, help='destination file path')
    
    args = parser.parse_args()
    
    run(src=args.src, des=args.des)