diff --git a/test/test_utils.py b/test/test_utils.py index 64fad58ad..5ebb8d498 100644 --- a/test/test_utils.py +++ b/test/test_utils.py @@ -38,6 +38,7 @@ from youtube_dl.utils import ( parse_iso8601, read_batch_urls, sanitize_filename, + sanitize_path, shell_quote, smuggle_url, str_to_int, @@ -131,6 +132,26 @@ class TestUtil(unittest.TestCase): self.assertEqual(sanitize_filename('_BD_eEpuzXw', is_id=True), '_BD_eEpuzXw') self.assertEqual(sanitize_filename('N0Y__7-UOdI', is_id=True), 'N0Y__7-UOdI') + def test_sanitize_path(self): + if sys.platform != 'win32': + return + + self.assertEqual(sanitize_path('abc'), 'abc') + self.assertEqual(sanitize_path('abc/def'), 'abc\\def') + self.assertEqual(sanitize_path('abc\\def'), 'abc\\def') + self.assertEqual(sanitize_path('abc|def'), 'abc#def') + self.assertEqual(sanitize_path('<>:"|?*'), '#######') + self.assertEqual(sanitize_path('C:/abc/def'), 'C:\\abc\\def') + self.assertEqual(sanitize_path('C?:/abc/def'), 'C##\\abc\\def') + + self.assertEqual(sanitize_path('\\\\?\\UNC\\ComputerName\\abc'), '\\\\?\\UNC\\ComputerName\\abc') + self.assertEqual(sanitize_path('\\\\?\\UNC/ComputerName/abc'), '\\\\?\\UNC\\ComputerName\\abc') + + self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc') + self.assertEqual(sanitize_path('\\\\?\\C:/abc'), '\\\\?\\C:\\abc') + self.assertEqual(sanitize_path('\\\\?\\C:\\ab?c\\de:f'), '\\\\?\\C:\\ab#c\\de#f') + self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc') + def test_ordered_set(self): self.assertEqual(orderedSet([1, 1, 2, 3, 4, 4, 5, 6, 7, 3, 5]), [1, 2, 3, 4, 5, 6, 7]) self.assertEqual(orderedSet([]), []) diff --git a/youtube_dl/utils.py b/youtube_dl/utils.py index 7426e2a1f..0f49d602e 100644 --- a/youtube_dl/utils.py +++ b/youtube_dl/utils.py @@ -311,6 +311,24 @@ def sanitize_filename(s, restricted=False, is_id=False): return result +def sanitize_path(s): + """Sanitizes and normalizes path on Windows""" + if sys.platform != 'win32': + return s + drive, _ = os.path.splitdrive(s) + unc, _ = os.path.splitunc(s) + unc_or_drive = unc or drive + norm_path = os.path.normpath(remove_start(s, unc_or_drive)).split(os.path.sep) + if unc_or_drive: + norm_path.pop(0) + sanitized_path = [ + re.sub('[/<>:"\\|\\\\?\\*]', '#', path_part) + for path_part in norm_path] + if unc_or_drive: + sanitized_path.insert(0, unc_or_drive + os.path.sep) + return os.path.join(*sanitized_path) + + def orderedSet(iterable): """ Remove all duplicates from the input iterable """ res = []