def loadKeys(self):
if not os.path.exists("keys"):
os.mkdir("keys")
return
files = get_all_files("keys/")
for key in files:
key_name = self.loadKeyFromFile(f"keys/{key}")
self.gui.keyBox.insert(0, key_name)
def calculate():
d = sys.argv[1]
files = util.get_all_files(d)
files_count = float(len(files))
avg = 0
for f in files:
h = open(f[0] + f[1])
body = h.read()
avg += average(body)
h.close()
avg /= files_count
print 'Average word length: ' + str(avg)
def main():
argument_parser = argparse.ArgumentParser()
argument_parser.add_argument('--input-dir')
argument_parser.add_argument('--output-dir', default='data')
argument_parser.add_argument('--train-size', default=0.7, type=float)
args = argument_parser.parse_args()
sub_dirs = get_subdirs(args.input_dir)
for class_dir in sub_dirs:
file_paths = get_all_files(class_dir)
# Remove the folder above the dataset
class_dir = class_dir.split('/')[-1]
train_files, test_files = train_test_split(file_paths,
train_size=args.train_size)
move_files(train_files, args.output_dir + '/train/' + class_dir)
move_files(test_files, args.output_dir + '/test/' + class_dir)
def load_all(styles, batch_size, time_steps):
"""
Loads all MIDI files as a piano roll.
(For Keras)
"""
note_data = []
beat_data = []
style_data = []
note_target = []
# TODO: Can speed this up with better parallel loading. Order gaurentee.
styles = [y for x in styles for y in x]
for style_id, style in enumerate(styles):
style_hot = one_hot(style_id, NUM_STYLES)
# Parallel process all files into a list of music sequences
seqs = Parallel(n_jobs=multiprocessing.cpu_count(),
backend='threading')(delayed(load_midi)(f)
for f in get_all_files([style]))
for seq in seqs:
if len(seq) >= time_steps:
# Clamp MIDI to note range
seq = clamp_midi(seq)
# Create training data and labels
train_data, label_data = stagger(seq, time_steps)
note_data += train_data
note_target += label_data
# beats = [compute_beat(i, NOTES_PER_BAR) for i in range(len(seq))]
# beat_data += stagger(beats, time_steps)[0]
note_data = np.array(note_data)
# beat_data = np.array(beat_data)
note_target = np.array(note_target)
#note_data[:,:,:,2] = 0 #note_data[:,:,:,0]
#note_target[:,:,:,2] = 0 #note_target[:,:,:,0]
return note_data, note_target,
def main_augumented_img(file_path, aug_img_folder):
image_paths = util.get_all_files(file_path, reg='.npy')
count = 0
total_img = len(image_paths)
for i in range(total_img):
print('processed images ratio: ' + str(int(i * 100 / total_img)) + "%")
c = gen_augumented_img(image_paths[i],
aug_img_folder,
size_label=128,
stride=64)
count += c
print('total num of images: ' + str(i))
print('total num of <quad, bayer/rgb> pairs after augumentation: ' +
str(count))
pass
def init_environment(self):
if os.path.exists(self.pages_folder_path):
util.remove_folder_contents(self.pages_folder_path)
os.makedirs(self.pages_folder_path)
if os.path.exists(self.train_path):
util.remove_folder_contents(self.train_path)
os.makedirs(self.train_path)
if os.path.exists(self.valid_path):
util.remove_folder_contents(self.valid_path)
os.makedirs(self.valid_path)
if os.path.exists(self.test_path):
util.remove_folder_contents(self.test_path)
os.makedirs(self.test_path)
if not os.path.exists(self.urls_folder_path):
print('pls provide url files in a folder named urls')
return
self.urls_file_names = util.get_all_files(self.urls_folder_path)
logger.info('init environment')
def unpack_works(config):
cv_folder = config['working'] + config['cv dir']
all_cv = util.get_all_files(cv_folder)
works = unpack_works_from_all_cv(all_cv)
return works
ids2names[idn] = set()
ids2names[idn].add(name)
if (debug) and (len(ids2names[idn]) > 1):
print('{0} cant write their name!'.format(name))
except ValueError:
print('discarded entry!')
return names2ids, ids2names
if __name__ == '__main__':
config = util.load_config(sys.argv[1])
if sys.argv[2] == 'CV':
cv_folder = config['working'] + config['cv dir']
all_cv = util.get_all_files(cv_folder)
names = get_name_frequency(all_cv, debug=True)
print('---')
print(names)
print('...')
elif sys.argv[2] == 'ID':
source = cfs.get_input(config)
names2ids, ids2names = relate_names_and_ids(source, debug=True)
print('--- # Name -> #Ids')
for name in names2ids:
print('{0}: {1}'.format(name, len(names2ids[name])))
print('--- # Id -> #names')
for idn in ids2names:
print('{0}: {1}'.format(idn, len(ids2names[idn])))
print('...')