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