test/test_YoutubeDL.py

@@ -2,7 +2,6 @@
 
 # Allow direct execution
 import os
-import platform
 import sys
 import unittest
 from unittest.mock import patch
@@ -715,7 +714,7 @@ class TestYoutubeDL(unittest.TestCase):
     def test_prepare_outtmpl_and_filename(self):
         def test(tmpl, expected, *, info=None, **params):
             if 'trim_file_name' not in params:
-                params['trim_file_name'] = 'notrim'  # disable trimming
+                params['trim_file_name'] = 0  # disable trimming
             params['outtmpl'] = tmpl
             ydl = FakeYDL(params)
             ydl._num_downloads = 1
@@ -931,16 +930,12 @@ class TestYoutubeDL(unittest.TestCase):
         # --trim-filenames
         test('%(title6)s.%(ext)s', 'あ' * 10 + '.mp4')
         test('%(title6)s.%(ext)s', 'あ' * 3 + '.mp4', trim_file_name='3c')
-        if sys.getfilesystemencoding() == 'utf-8' and platform.system() != 'Windows':
-            test('%(title6)s.%(ext)s', 'あ' * 3 + '.mp4', trim_file_name='9b')
-            test('%(title6)s.%(ext)s', 'あ' * 3 + '.mp4', trim_file_name='10b')
-            test('%(title6)s.%(ext)s', 'あ' * 3 + '.mp4', trim_file_name='11b')
-            test('%(title6)s.%(ext)s', 'あ' * 4 + '.mp4', trim_file_name='12b')
-        elif platform.system() == 'Windows':
-            test('%(title6)s.%(ext)s', 'あ' * 4 + '.mp4', trim_file_name='8b')
-            test('%(title6)s.%(ext)s', 'あ' * 4 + '.mp4', trim_file_name='9b')
-            test('%(title6)s.%(ext)s', 'あ' * 5 + '.mp4', trim_file_name='10b')
-        test('folder/%(title6)s.%(ext)s', f'fol{os.path.sep}あああ.mp4', trim_file_name='3c')
+        test('%(title6)s.%(ext)s', 'あ' * 3 + '.mp4', trim_file_name='9b', filesystem_encoding='utf-8')
+        test('%(title6)s.%(ext)s', 'あ' * 3 + '.mp4', trim_file_name='10b', filesystem_encoding='utf-8')
+        test('%(title6)s.%(ext)s', 'あ' * 3 + '.mp4', trim_file_name='11b', filesystem_encoding='utf-8')
+        test('%(title6)s.%(ext)s', 'あ' * 4 + '.mp4', trim_file_name='12b', filesystem_encoding='utf-8')
+        test('%(title6)s.%(ext)s', 'あ' * 6 + '.mp4', trim_file_name='12b', filesystem_encoding='utf-16le')
+        test('folder/%(title6)s.%(ext)s', f'folder{os.path.sep}あああ.mp4', trim_file_name='3c')
 
     def test_format_note(self):
         ydl = YoutubeDL()

yt_dlp/YoutubeDL.py

@@ -12,8 +12,6 @@ import json
 import locale
 import operator
 import os
-from pathlib import Path
-import platform
 import random
 import re
 import shutil
@@ -72,6 +70,7 @@ from .update import (
 )
 from .utils import (
     DEFAULT_OUTTMPL,
+    DEFAULT_MAX_FILE_NAME,
     IDENTITY,
     LINK_TEMPLATES,
     MEDIA_EXTENSIONS,
@@ -268,6 +267,7 @@ class YoutubeDL:
     outtmpl_na_placeholder: Placeholder for unavailable meta fields.
     restrictfilenames: Do not allow "&" and spaces in file names
     trim_file_name:    Limit length of filename (extension excluded)
+    filesystem_encoding: Encoding to use when calculating filename length in bytes
     windowsfilenames:  True: Force filenames to be Windows compatible
                        False: Sanitize filenames only minimally
                        This option has no effect when running on Windows
@@ -1430,7 +1430,7 @@ class YoutubeDL:
         if not trim_filename:
             return self.escape_outtmpl(outtmpl) % info_dict
 
-        ext_suffix = '.%(ext\0s)s'
+        ext_suffix = '.%(ext\x00s)s'  # not sure why this has null char
         suffix = ''
         if outtmpl.endswith(ext_suffix):
             outtmpl = outtmpl[:-len(ext_suffix)]
@@ -1438,28 +1438,49 @@ class YoutubeDL:
         outtmpl = self.escape_outtmpl(outtmpl)
         filename = outtmpl % info_dict
 
-        def parse_trim_file_name(trim_file_name):
-            if trim_file_name is None or trim_file_name == 'notrim':
-                return 0, None
-            mobj = re.match(r'(?:(?P<length>\d+)(?P<mode>b|c)?|notrim)', trim_file_name)
-            return int(mobj.group('length')), mobj.group('mode') or 'c'
+        def parse_max_file_name(max_file_name: str):
+            # old --trim-filenames format
+            try:
+                return 'c', int(max_file_name)
+            except ValueError:
+                pass
 
-        max_file_name, mode = parse_trim_file_name(self.params.get('trim_file_name'))
+            try:
+                max_length = int(max_file_name[:-1])
+            except ValueError:
+                raise ValueError('Invalid --trim-filenames specified')
+
+            if max_file_name[-1].lower() == 'c':
+                return 'c', max_length
+            elif max_file_name[-1].lower() == 'b':
+                return 'b', max_length
+            else:
+                raise ValueError("--trim-filenames must end with 'b' or 'c'")
+
+        max_file_name = self.params.get('trim_file_name')
+        if max_file_name is None:
+            max_file_name = DEFAULT_MAX_FILE_NAME
+        mode, max_file_name = parse_max_file_name(max_file_name)
+        if max_file_name < 0:
+            raise ValueError('Invalid --trim-filenames specified')
         if max_file_name == 0:
             # no maximum
             return filename + suffix
 
-        encoding = sys.getfilesystemencoding() if platform.system() != 'Windows' else 'utf-16-le'
+        encoding = self.params.get('filesystem_encoding') or sys.getfilesystemencoding()
 
-        def trim_filename(name: str):
+        def trim_filename(name: str, length: int):
             if mode == 'b':
                 name = name.encode(encoding)
-                name = name[:max_file_name]
+                name = name[:length]
                 return name.decode(encoding, 'ignore')
             else:
-                return name[:max_file_name]
+                return name[:length]
 
-        filename = os.path.join(*map(trim_filename, Path(filename).parts))
+        # only trim last component of path - assume the directories are valid names
+        head, tail = os.path.split(filename)
+        tail = trim_filename(tail, max_file_name)
+        filename = os.path.join(head, tail)
         return filename + suffix
 
     @_catch_unsafe_extension_error

yt_dlp/__init__.py

@@ -429,8 +429,6 @@ def validate_options(opts):
     }
 
     # Other options
-    validate_regex('trim filenames', opts.trim_file_name, r'(?:\d+[bc]?|notrim)')
-
     if opts.playlist_items is not None:
         try:
             tuple(PlaylistEntries.parse_playlist_items(opts.playlist_items))
@@ -888,6 +886,7 @@ def parse_options(argv=None):
         'max_downloads': opts.max_downloads,
         'prefer_free_formats': opts.prefer_free_formats,
         'trim_file_name': opts.trim_file_name,
+        'filesystem_encoding': opts.filesystem_encoding,
         'verbose': opts.verbose,
         'dump_intermediate_pages': opts.dump_intermediate_pages,
         'write_pages': opts.write_pages,

yt_dlp/options.py

@@ -1378,8 +1378,12 @@ def create_parser():
         help='Sanitize filenames only minimally')
     filesystem.add_option(
         '--trim-filenames', '--trim-file-names', metavar='LENGTH',
-        dest='trim_file_name', default='notrim',
+        dest='trim_file_name',
         help='Limit the filename length (excluding extension) to the specified number of characters or bytes')
+    filesystem.add_option(
+        '--filesystem-encoding', metavar='ENCODING',
+        dest='filesystem_encoding',
+        help='Override filesystem encoding used when calculating filename length in bytes')
     filesystem.add_option(
         '-w', '--no-overwrites',
         action='store_false', dest='overwrites', default=None,

yt_dlp/utils/_utils.py

@@ -2852,6 +2852,12 @@ OUTTMPL_TYPES = {
     'pl_infojson': 'info.json',
 }
 
+# https://en.m.wikipedia.org/wiki/Comparison_of_file_systems#Limits
+if platform.system() in ('Darwin', 'Windows'):
+    DEFAULT_MAX_FILE_NAME = f'{255 - len(".annotations.xml")}c'
+else:
+    DEFAULT_MAX_FILE_NAME = f'{255 - len(".annotations.xml".encode(sys.getfilesystemencoding()))}b'
+
 # As of [1] format syntax is:
 #  %[mapping_key][conversion_flags][minimum_width][.precision][length_modifier]type
 # 1. https://docs.python.org/2/library/stdtypes.html#string-formatting