yt-dlp/yt_dlp/postprocessor/sponskrub.py

99 lines
4.0 KiB
Python
Raw Normal View History

2020-11-15 00:28:41 +00:00
import os
import shlex
2020-11-15 00:28:41 +00:00
import subprocess
from .common import PostProcessor
from ..utils import (
Popen,
PostProcessingError,
2020-11-15 00:28:41 +00:00
check_executable,
cli_option,
2020-11-15 00:28:41 +00:00
encodeArgument,
2021-01-17 23:52:47 +00:00
encodeFilename,
prepend_extension,
2020-11-15 00:28:41 +00:00
shell_quote,
str_or_none,
2020-11-15 00:28:41 +00:00
)
# Deprecated in favor of the native implementation
2020-11-15 00:28:41 +00:00
class SponSkrubPP(PostProcessor):
_temp_ext = 'spons'
_exe_name = 'sponskrub'
2021-11-29 17:46:06 +00:00
def __init__(self, downloader, path='', args=None, ignoreerror=False, cut=False, force=False, _from_cli=False):
2020-11-15 00:28:41 +00:00
PostProcessor.__init__(self, downloader)
self.force = force
self.cutout = cut
self.args = str_or_none(args) or '' # For backward compatibility
2020-11-15 00:28:41 +00:00
self.path = self.get_exe(path)
2021-11-29 17:46:06 +00:00
if not _from_cli:
self.deprecation_warning(
'yt_dlp.postprocessor.SponSkrubPP support is deprecated and may be removed in a future version. '
'Use yt_dlp.postprocessor.SponsorBlock and yt_dlp.postprocessor.ModifyChaptersPP instead')
2020-11-15 00:28:41 +00:00
if not ignoreerror and self.path is None:
if path:
raise PostProcessingError('sponskrub not found in "%s"' % path)
else:
raise PostProcessingError('sponskrub not found. Please install or provide the path using --sponskrub-path')
2020-11-15 00:28:41 +00:00
def get_exe(self, path=''):
if not path or not check_executable(path, ['-h']):
path = os.path.join(path, self._exe_name)
if not check_executable(path, ['-h']):
return None
return path
@PostProcessor._restrict_to(images=False)
2020-11-15 00:28:41 +00:00
def run(self, information):
if self.path is None:
return [], information
filename = information['filepath']
if not os.path.exists(encodeFilename(filename)): # no download
return [], information
2020-11-15 00:28:41 +00:00
if information['extractor_key'].lower() != 'youtube':
self.to_screen('Skipping sponskrub since it is not a YouTube video')
2020-11-15 00:28:41 +00:00
return [], information
if self.cutout and not self.force and not information.get('__real_download', False):
self.report_warning(
'Skipping sponskrub since the video was already downloaded. '
2020-11-15 00:28:41 +00:00
'Use --sponskrub-force to run sponskrub anyway')
return [], information
self.to_screen('Trying to %s sponsor sections' % ('remove' if self.cutout else 'mark'))
2020-11-15 00:28:41 +00:00
if self.cutout:
self.report_warning('Cutting out sponsor segments will cause the subtitles to go out of sync.')
2020-11-15 00:28:41 +00:00
if not information.get('__real_download', False):
self.report_warning('If sponskrub is run multiple times, unintended parts of the video could be cut out.')
2020-11-15 00:28:41 +00:00
2021-01-17 23:52:47 +00:00
temp_filename = prepend_extension(filename, self._temp_ext)
if os.path.exists(encodeFilename(temp_filename)):
os.remove(encodeFilename(temp_filename))
2020-11-15 00:28:41 +00:00
cmd = [self.path]
if not self.cutout:
cmd += ['-chapter']
cmd += cli_option(self._downloader.params, '-proxy', 'proxy')
cmd += shlex.split(self.args) # For backward compatibility
cmd += self._configuration_args(self._exe_name, use_compat=False)
2020-11-15 00:28:41 +00:00
cmd += ['--', information['id'], filename, temp_filename]
cmd = [encodeArgument(i) for i in cmd]
self.write_debug('sponskrub command line: %s' % shell_quote(cmd))
stdout, _, returncode = Popen.run(cmd, text=True, stdout=None if self.get_param('verbose') else subprocess.PIPE)
2020-11-15 00:28:41 +00:00
if not returncode:
os.replace(temp_filename, filename)
self.to_screen('Sponsor sections have been %s' % ('removed' if self.cutout else 'marked'))
elif returncode == 3:
self.to_screen('No segments in the SponsorBlock database')
else:
raise PostProcessingError(
stdout.strip().splitlines()[0 if stdout.strip().lower().startswith('unrecognised') else -1]
or f'sponskrub failed with error code {returncode}')
2020-11-15 00:28:41 +00:00
return [], information