yt_dlp/postprocessor/sponskrub.py

   1 import os
   2 import shlex
   3 import subprocess
   4
   5 from .common import PostProcessor
   6 from ..utils import (
   7     Popen,
   8     PostProcessingError,
   9     check_executable,
  10     cli_option,
  11     encodeArgument,
  12     encodeFilename,
  13     prepend_extension,
  14     shell_quote,
  15     str_or_none,
  16 )
  17
  18
  19 # Deprecated in favor of the native implementation
  20 class SponSkrubPP(PostProcessor):
  21     _temp_ext = 'spons'
  22     _exe_name = 'sponskrub'
  23
  24     def __init__(self, downloader, path='', args=None, ignoreerror=False, cut=False, force=False, _from_cli=False):
  25         PostProcessor.__init__(self, downloader)
  26         self.force = force
  27         self.cutout = cut
  28         self.args = str_or_none(args) or ''  # For backward compatibility
  29         self.path = self.get_exe(path)
  30
  31         if not _from_cli:
  32             self.deprecation_warning(
  33                 'yt_dlp.postprocessor.SponSkrubPP support is deprecated and may be removed in a future version. '
  34                 'Use yt_dlp.postprocessor.SponsorBlock and yt_dlp.postprocessor.ModifyChaptersPP instead')
  35
  36         if not ignoreerror and self.path is None:
  37             if path:
  38                 raise PostProcessingError('sponskrub not found in "%s"' % path)
  39             else:
  40                 raise PostProcessingError('sponskrub not found. Please install or provide the path using --sponskrub-path')
  41
  42     def get_exe(self, path=''):
  43         if not path or not check_executable(path, ['-h']):
  44             path = os.path.join(path, self._exe_name)
  45             if not check_executable(path, ['-h']):
  46                 return None
  47         return path
  48
  49     @PostProcessor._restrict_to(images=False)
  50     def run(self, information):
  51         if self.path is None:
  52             return [], information
  53
  54         filename = information['filepath']
  55         if not os.path.exists(encodeFilename(filename)):  # no download
  56             return [], information
  57
  58         if information['extractor_key'].lower() != 'youtube':
  59             self.to_screen('Skipping sponskrub since it is not a YouTube video')
  60             return [], information
  61         if self.cutout and not self.force and not information.get('__real_download', False):
  62             self.report_warning(
  63                 'Skipping sponskrub since the video was already downloaded. '
  64                 'Use --sponskrub-force to run sponskrub anyway')
  65             return [], information
  66
  67         self.to_screen('Trying to %s sponsor sections' % ('remove' if self.cutout else 'mark'))
  68         if self.cutout:
  69             self.report_warning('Cutting out sponsor segments will cause the subtitles to go out of sync.')
  70             if not information.get('__real_download', False):
  71                 self.report_warning('If sponskrub is run multiple times, unintended parts of the video could be cut out.')
  72
  73         temp_filename = prepend_extension(filename, self._temp_ext)
  74         if os.path.exists(encodeFilename(temp_filename)):
  75             os.remove(encodeFilename(temp_filename))
  76
  77         cmd = [self.path]
  78         if not self.cutout:
  79             cmd += ['-chapter']
  80         cmd += cli_option(self._downloader.params, '-proxy', 'proxy')
  81         cmd += shlex.split(self.args)  # For backward compatibility
  82         cmd += self._configuration_args(self._exe_name, use_compat=False)
  83         cmd += ['--', information['id'], filename, temp_filename]
  84         cmd = [encodeArgument(i) for i in cmd]
  85
  86         self.write_debug('sponskrub command line: %s' % shell_quote(cmd))
  87         stdout, _, returncode = Popen.run(cmd, text=True, stdout=None if self.get_param('verbose') else subprocess.PIPE)
  88
  89         if not returncode:
  90             os.replace(temp_filename, filename)
  91             self.to_screen('Sponsor sections have been %s' % ('removed' if self.cutout else 'marked'))
  92         elif returncode == 3:
  93             self.to_screen('No segments in the SponsorBlock database')
  94         else:
  95             raise PostProcessingError(
  96                 stdout.strip().splitlines()[0 if stdout.strip().lower().startswith('unrecognised') else -1]
  97                 or f'sponskrub failed with error code {returncode}')
  98         return [], information