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