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