Merge 134f604982 into 73bf102116

test/test_YoutubeDL.py

@@ -2,6 +2,7 @@
 
 # Allow direct execution
 import os
+import platform
 import sys
 import unittest
 from unittest.mock import patch
@@ -724,6 +725,7 @@ def test_add_extra_info(self):
         'title3': 'foo/bar\\test',
         'title4': 'foo "bar" test',
         'title5': 'áéí 𝐀',
+        'title6': 'あ' * 10,
         'timestamp': 1618488000,
         'duration': 100000,
         'playlist_index': 1,
@@ -739,12 +741,14 @@ def test_add_extra_info(self):
 
     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'] = 'none'  # disable trimming
             params['outtmpl'] = tmpl
             ydl = FakeYDL(params)
             ydl._num_downloads = 1
             self.assertEqual(ydl.validate_outtmpl(tmpl), None)
 
-            out = ydl.evaluate_outtmpl(tmpl, info or self.outtmpl_info)
+            out = ydl.evaluate_outtmpl(tmpl, info or self.outtmpl_info, trim_filename=True)
             fname = ydl.prepare_filename(info or self.outtmpl_info)
 
             if not isinstance(expected, (list, tuple)):
@@ -951,6 +955,20 @@ def gen():
         test('%(title3)s', ('foo/bar\\test', 'foo⧸bar⧹test'))
         test('folder/%(title3)s', ('folder/foo/bar\\test', f'folder{os.path.sep}foo⧸bar⧹test'))
 
+        # --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')
+
     def test_format_note(self):
         ydl = YoutubeDL()
         self.assertEqual(ydl._format_note({}), '')

yt_dlp/YoutubeDL.py

@@ -12,6 +12,8 @@
 import locale
 import operator
 import os
+from pathlib import Path
+import platform
 import random
 import re
 import shutil
@@ -1451,9 +1453,42 @@ def create_key(outer_mobj):
 
         return EXTERNAL_FORMAT_RE.sub(create_key, outtmpl), TMPL_DICT
 
-    def evaluate_outtmpl(self, outtmpl, info_dict, *args, **kwargs):
+    def evaluate_outtmpl(self, outtmpl, info_dict, *args, trim_filename=False, **kwargs):
         outtmpl, info_dict = self.prepare_outtmpl(outtmpl, info_dict, *args, **kwargs)
-        return self.escape_outtmpl(outtmpl) % info_dict
+        if not trim_filename:
+            return self.escape_outtmpl(outtmpl) % info_dict
+
+        ext_suffix = '.%(ext\0s)s'
+        suffix = ''
+        if outtmpl.endswith(ext_suffix):
+            outtmpl = outtmpl[:-len(ext_suffix)]
+            suffix = ext_suffix % info_dict
+        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 == 'none':
+                return 0, None
+            mobj = re.match(r'(?:(?P<length>\d+)(?P<mode>b|c)?|none)', trim_file_name)
+            return int(mobj.group('length')), mobj.group('mode') or 'c'
+
+        max_file_name, mode = parse_trim_file_name(self.params.get('trim_file_name'))
+        if max_file_name == 0:
+            # no maximum
+            return filename + suffix
+
+        encoding = sys.getfilesystemencoding() if platform.system() != 'Windows' else 'utf-16-le'
+
+        def trim_filename(name: str):
+            if mode == 'b':
+                name = name.encode(encoding)
+                name = name[:max_file_name]
+                return name.decode(encoding, 'ignore')
+            else:
+                return name[:max_file_name]
+
+        filename = os.path.join(*map(trim_filename, Path(filename).parts or '.'))
+        return filename + suffix
 
     @_catch_unsafe_extension_error
     def _prepare_filename(self, info_dict, *, outtmpl=None, tmpl_type=None):
@@ -1462,7 +1497,7 @@ def _prepare_filename(self, info_dict, *, outtmpl=None, tmpl_type=None):
             outtmpl = self.params['outtmpl'].get(tmpl_type or 'default', self.params['outtmpl']['default'])
         try:
             outtmpl = self._outtmpl_expandpath(outtmpl)
-            filename = self.evaluate_outtmpl(outtmpl, info_dict, True)
+            filename = self.evaluate_outtmpl(outtmpl, info_dict, True, trim_filename=True)
             if not filename:
                 return None
 
@@ -1474,13 +1509,6 @@ def _prepare_filename(self, info_dict, *, outtmpl=None, tmpl_type=None):
                 force_ext = OUTTMPL_TYPES[tmpl_type]
                 if force_ext:
                     filename = replace_extension(filename, force_ext, info_dict.get('ext'))
-
-            # https://github.com/blackjack4494/youtube-dlc/issues/85
-            trim_file_name = self.params.get('trim_file_name', False)
-            if trim_file_name:
-                no_ext, *ext = filename.rsplit('.', 2)
-                filename = join_nonempty(no_ext[:trim_file_name], *ext, delim='.')
-
             return filename
         except ValueError as err:
             self.report_error('Error in output template: ' + str(err) + ' (encoding: ' + repr(preferredencoding()) + ')')

yt_dlp/__init__.py

@@ -436,6 +436,8 @@ def metadataparser_actions(f):
     if opts.plugin_dirs is None:
         opts.plugin_dirs = ['default']
 
+    validate_regex('trim filenames', opts.trim_file_name, r'(?:\d+[bc]?|none)')
+
     if opts.playlist_items is not None:
         try:
             tuple(PlaylistEntries.parse_playlist_items(opts.playlist_items))

yt_dlp/options.py

@@ -1432,8 +1432,8 @@ def _preset_alias_callback(option, opt_str, value, parser):
         help='Sanitize filenames only minimally')
     filesystem.add_option(
         '--trim-filenames', '--trim-file-names', metavar='LENGTH',
-        dest='trim_file_name', default=0, type=int,
-        help='Limit the filename length (excluding extension) to the specified number of characters')
+        dest='trim_file_name', default='none',
+        help='Limit the filename length (excluding extension) to the specified number of characters or bytes')
     filesystem.add_option(
         '-w', '--no-overwrites',
         action='store_false', dest='overwrites', default=None,