youtube_dlc/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     encodeArgument,
  10     shell_quote,
  11     PostProcessingError,
  12 )
  13
  14
  15 class SponSkrubPP(PostProcessor):
  16     _temp_ext = 'spons'
  17     _def_args = []
  18     _exe_name = 'sponskrub'
  19
  20     def __init__(self, downloader, path='', args=None, ignoreerror=False, cut=False, force=False):
  21         PostProcessor.__init__(self, downloader)
  22         self.force = force
  23         self.cutout = cut
  24         self.args = ['-chapter'] if not cut else []
  25         self.args += self._configuration_args(self._def_args) if args is None else compat_shlex_split(args)
  26         self.path = self.get_exe(path)
  27
  28         if not ignoreerror and self.path is None:
  29             if path:
  30                 raise PostProcessingError('sponskrub not found in "%s"' % path)
  31             else:
  32                 raise PostProcessingError('sponskrub not found. Please install or provide the path using --sponskrub-path.')
  33
  34     def get_exe(self, path=''):
  35         if not path or not check_executable(path, ['-h']):
  36             path = os.path.join(path, self._exe_name)
  37             if not check_executable(path, ['-h']):
  38                 return None
  39         return path
  40
  41     def run(self, information):
  42         if self.path is None:
  43             return [], information
  44
  45         if information['extractor_key'].lower() != 'youtube':
  46             self.to_screen('Skipping sponskrub since it is not a YouTube video')
  47             return [], information
  48         if self.cutout and not self.force and not information.get('__real_download', False):
  49             self.report_warning(
  50                 'Skipping sponskrub since the video was already downloaded. '
  51                 'Use --sponskrub-force to run sponskrub anyway')
  52             return [], information
  53
  54         self.to_screen('Trying to %s sponsor sections' % ('remove' if self.cutout else 'mark'))
  55         if self.cutout:
  56             self.report_warning('Cutting out sponsor segments will cause the subtitles to go out of sync.')
  57             if not information.get('__real_download', False):
  58                 self.report_warning('If sponskrub is run multiple times, unintended parts of the video could be cut out.')
  59
  60         filename = information['filepath']
  61         temp_filename = filename + '.' + self._temp_ext + os.path.splitext(filename)[1]
  62         if os.path.exists(temp_filename):
  63             os.remove(temp_filename)
  64
  65         cmd = [self.path]
  66         if self.args:
  67             cmd += self.args
  68         cmd += ['--', information['id'], filename, temp_filename]
  69         cmd = [encodeArgument(i) for i in cmd]
  70
  71         self.write_debug('sponskrub command line: %s' % shell_quote(cmd))
  72         p = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE, stdin=subprocess.PIPE)
  73         stdout, stderr = p.communicate()
  74
  75         if p.returncode == 0:
  76             os.remove(filename)
  77             os.rename(temp_filename, filename)
  78             self.to_screen('Sponsor sections have been %s' % ('removed' if self.cutout else 'marked'))
  79         elif p.returncode == 3:
  80             self.to_screen('No segments in the SponsorBlock database')
  81         else:
  82             stderr = stderr.decode('utf-8', 'replace')
  83             msg = stderr.strip().split('\n')[-1]
  84             raise PostProcessingError(msg if msg else 'sponskrub failed with error code %s!' % p.returncode)
  85         return [], information