data_shuffler.py
740 Bytes
from utils import *
import file_parser
import random
def shuffle_files(input, output): # pick random file and shuffle code order to output
files = [f for f in readdir(input) if is_extension(f, 'py')]
random.shuffle(files)
for p in files:
lines = read_file(p)
print("Shuffling:", p)
block = file_parser.parse_block(lines)
shuffle_block(block)
filepath = p.split(input)[1][1:]
os.makedirs(os.path.join(output, filepath.split('\\')[0]), exist_ok=True) # create the output directory if not exists
path = os.path.join(output, filepath)
write_block(path, block)
print("Done generating Shuffled Dataset")
'''
shuffle_files('data/original', 'data/shuffled')
'''