Compare commits

...

3 Commits

Author SHA1 Message Date
pukkandan
97a09f909e
Merge 9de472f6ea into 10fc719bc7 2024-11-17 13:02:34 -05:00
pukkandan
9de472f6ea
Add tests 2024-04-01 15:11:16 +05:30
pukkandan
0f58c5628a
Reduce filename sanitization when --no-windows-filenames is given 2024-04-01 14:46:37 +05:30
4 changed files with 28 additions and 12 deletions

View File

@ -612,8 +612,7 @@ If you fork the project on GitHub, you can run your fork's [build workflow](.git
--no-restrict-filenames Allow Unicode characters, "&" and spaces in --no-restrict-filenames Allow Unicode characters, "&" and spaces in
filenames (default) filenames (default)
--windows-filenames Force filenames to be Windows-compatible --windows-filenames Force filenames to be Windows-compatible
--no-windows-filenames Make filenames Windows-compatible only if --no-windows-filenames Sanitize filenames only minimally
using Windows (default)
--trim-filenames LENGTH Limit the filename length (excluding --trim-filenames LENGTH Limit the filename length (excluding
extension) to the specified number of extension) to the specified number of
characters characters

View File

@ -761,6 +761,13 @@ class TestYoutubeDL(unittest.TestCase):
test('%(width)06d.%%(ext)s', 'NA.%(ext)s') test('%(width)06d.%%(ext)s', 'NA.%(ext)s')
test('%%(width)06d.%(ext)s', '%(width)06d.mp4') test('%%(width)06d.%(ext)s', '%(width)06d.mp4')
# Sanitization options
test('%(title3)s', (None, 'foobartest'))
test('%(title5)s', (None, 'aei_A'), restrictfilenames=True)
test('%(title3)s', (None, 'foo_bar_test'), windowsfilenames=False, restrictfilenames=True)
if sys.platform != 'win32':
test('%(title3)s', (None, 'foobar\\test'), windowsfilenames=False)
# ID sanitization # ID sanitization
test('%(id)s', '_abcd', info={'id': '_abcd'}) test('%(id)s', '_abcd', info={'id': '_abcd'})
test('%(some_id)s', '_abcd', info={'some_id': '_abcd'}) test('%(some_id)s', '_abcd', info={'some_id': '_abcd'})

View File

@ -266,7 +266,9 @@ class YoutubeDL:
outtmpl_na_placeholder: Placeholder for unavailable meta fields. outtmpl_na_placeholder: Placeholder for unavailable meta fields.
restrictfilenames: Do not allow "&" and spaces in file names restrictfilenames: Do not allow "&" and spaces in file names
trim_file_name: Limit length of filename (extension excluded) trim_file_name: Limit length of filename (extension excluded)
windowsfilenames: Force the filenames to be windows compatible windowsfilenames: True: Force filenames to be Windows compatible
False: Sanitize filenames only minimally
This option has no effect when running on Windows
ignoreerrors: Do not stop on download/postprocessing errors. ignoreerrors: Do not stop on download/postprocessing errors.
Can be 'only_download' to ignore only download errors. Can be 'only_download' to ignore only download errors.
Default is 'only_download' for CLI, but False for API Default is 'only_download' for CLI, but False for API
@ -1192,8 +1194,7 @@ class YoutubeDL:
def prepare_outtmpl(self, outtmpl, info_dict, sanitize=False): def prepare_outtmpl(self, outtmpl, info_dict, sanitize=False):
""" Make the outtmpl and info_dict suitable for substitution: ydl.escape_outtmpl(outtmpl) % info_dict """ Make the outtmpl and info_dict suitable for substitution: ydl.escape_outtmpl(outtmpl) % info_dict
@param sanitize Whether to sanitize the output as a filename. @param sanitize Whether to sanitize the output as a filename
For backward compatibility, a function can also be passed
""" """
info_dict.setdefault('epoch', int(time.time())) # keep epoch consistent once set info_dict.setdefault('epoch', int(time.time())) # keep epoch consistent once set
@ -1309,14 +1310,23 @@ class YoutubeDL:
na = self.params.get('outtmpl_na_placeholder', 'NA') na = self.params.get('outtmpl_na_placeholder', 'NA')
def filename_sanitizer(key, value, restricted=self.params.get('restrictfilenames')): def filename_sanitizer(key, value, restricted):
return sanitize_filename(str(value), restricted=restricted, is_id=( return sanitize_filename(str(value), restricted=restricted, is_id=(
bool(re.search(r'(^|[_.])id(\.|$)', key)) bool(re.search(r'(^|[_.])id(\.|$)', key))
if 'filename-sanitization' in self.params['compat_opts'] if 'filename-sanitization' in self.params['compat_opts']
else NO_DEFAULT)) else NO_DEFAULT))
sanitizer = sanitize if callable(sanitize) else filename_sanitizer if callable(sanitize):
sanitize = bool(sanitize) self.deprecation_warning('Passing a callable "sanitize" to YoutubeDL.prepare_outtmpl is deprecated')
elif not sanitize:
pass
elif (sys.platform != 'win32' and not self.params.get('restrictfilenames')
and self.params.get('windowsfilenames') is False):
def sanitize(key, value):
return value.replace('/', '\u29F8').replace('\0', '')
else:
def sanitize(key, value):
return filename_sanitizer(key, value, restricted=self.params.get('restrictfilenames'))
def _dumpjson_default(obj): def _dumpjson_default(obj):
if isinstance(obj, (set, LazyList)): if isinstance(obj, (set, LazyList)):
@ -1399,13 +1409,13 @@ class YoutubeDL:
if sanitize: if sanitize:
# If value is an object, sanitize might convert it to a string # If value is an object, sanitize might convert it to a string
# So we convert it to repr first # So we manually convert it before sanitizing
if fmt[-1] == 'r': if fmt[-1] == 'r':
value, fmt = repr(value), str_fmt value, fmt = repr(value), str_fmt
elif fmt[-1] == 'a': elif fmt[-1] == 'a':
value, fmt = ascii(value), str_fmt value, fmt = ascii(value), str_fmt
if fmt[-1] in 'csra': if fmt[-1] in 'csra':
value = sanitizer(last_field, value) value = sanitize(last_field, value)
key = '{}\0{}'.format(key.replace('%', '%\0'), outer_mobj.group('format')) key = '{}\0{}'.format(key.replace('%', '%\0'), outer_mobj.group('format'))
TMPL_DICT[key] = value TMPL_DICT[key] = value

View File

@ -1368,12 +1368,12 @@ def create_parser():
help='Allow Unicode characters, "&" and spaces in filenames (default)') help='Allow Unicode characters, "&" and spaces in filenames (default)')
filesystem.add_option( filesystem.add_option(
'--windows-filenames', '--windows-filenames',
action='store_true', dest='windowsfilenames', default=False, action='store_true', dest='windowsfilenames', default=None,
help='Force filenames to be Windows-compatible') help='Force filenames to be Windows-compatible')
filesystem.add_option( filesystem.add_option(
'--no-windows-filenames', '--no-windows-filenames',
action='store_false', dest='windowsfilenames', action='store_false', dest='windowsfilenames',
help='Make filenames Windows-compatible only if using Windows (default)') help='Sanitize filenames only minimally')
filesystem.add_option( filesystem.add_option(
'--trim-filenames', '--trim-file-names', metavar='LENGTH', '--trim-filenames', '--trim-file-names', metavar='LENGTH',
dest='trim_file_name', default=0, type=int, dest='trim_file_name', default=0, type=int,