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