ddrescue_directories.py

#!/usr/bin/env python
import os, shlex, subprocess
from tqdm import tqdm

# ddrescue directories instead of a whole drive
# creates directories recursively and ddrescues files into them

def fs (line):
	import subprocess
	PIPE = subprocess.PIPE
	return subprocess.Popen(line, shell=isinstance('', str),bufsize=-1, stdin=PIPE, stdout=PIPE,stderr=subprocess.STDOUT, close_fds=True).stdout.read().decode("utf8", "ignore")

#############################################################

# copy here
pth_dest = '/mnt/1/2/'

# list of directories
pths = '''

/mnt/2/Music

file:///mnt/2/video/
file:///mnt/2/2see
file:///mnt/2/abooks
/mnt/2/backup/

'''

# don't scrap or retry. Best for a first run
first_run = True

# save list of successfully copies files to skip on successive runs
done_files_fn = '/mnt/1/2/done_files_null_seped'

#############################################################

try:
	done_files_data = [ x for x in open(done_files_fn).read().split('\0') if len(x) ]
except:
	done_files_data = []
	
if pth_dest[-1] != '/':
	pth_dest += '/'

pths = [ x.strip() for x in pths.split('\n') if len(x.strip()) ]

for pth in pths:
	pth = pth.replace('file://', '')
	if pth[-1] != '/':
		pth += '/'

	file_list = fs('find %s -type f -print0' % shlex.quote(pth))
	
	# for errors when using `find`
	if file_list[0:5] == 'find:':
		print('Error getting file-list in %s' % pth)
		continue
	
	print(pth)
	
	file_list = [ x for x in file_list.split('\0') if len(x) ]
	tmp_count = len(file_list)
	file_list = [ x for x in file_list if x not in done_files_data ]
	tmp_count = '%d(%d)' % (len(file_list), tmp_count)
	
	if not len(file_list):
		continue
	
	for f in tqdm(file_list, desc=tmp_count):
		# ~print('\n\n\n', f)
		
		file_path = f.rsplit('/', 1)[0]
		os.makedirs(pth_dest + file_path, exist_ok=True)
		
		if first_run:
			subprocess.call(['ddrescue', '-q', '--max-read-errors=10', '-n', f, pth_dest + f, pth_dest + f + '_map_ddrescue'])
		else:
			subprocess.call(['ddrescue', '-vvvv', '-r', '5', f, pth_dest + f, pth_dest + f + '_map_ddrescue'])
			subprocess.call(['ddrescue', '-vvvv', '-r', '5', '-R', f, pth_dest + f, pth_dest + f + '_map_ddrescue'])
		
		
		# log fully copied files
		if not subprocess.call(['ddrescuelog', '-D', pth_dest + f + '_map_ddrescue']):
			with open(done_files_fn, 'a') as done_files_tmp:
				done_files_tmp.write(f + '\0')