yt_dlp/postprocessor/sponskrub.py

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