-from __future__ import unicode_literals
-
import collections
-import io
+import contextvars
import itertools
+import json
import os
+import re
import subprocess
import time
-import re
-import json
-from .common import AudioConversionError, PostProcessor
-
-from ..compat import compat_str
+from .common import PostProcessor
+from ..compat import functools, imghdr
from ..utils import (
+ MEDIA_EXTENSIONS,
+ ISO639Utils,
+ Popen,
+ PostProcessingError,
+ _get_exe_version_output,
+ deprecation_warning,
+ detect_exe_version,
determine_ext,
dfxp2srt,
encodeArgument,
encodeFilename,
+ filter_dict,
float_or_none,
- _get_exe_version_output,
- detect_exe_version,
is_outdated_version,
- ISO639Utils,
orderedSet,
- Popen,
- PostProcessingError,
prepend_extension,
replace_extension,
shell_quote,
write_json_file,
)
-
EXT_TO_OUT_FORMATS = {
'aac': 'adts',
'flac': 'flac',
'ts': 'mpegts',
'wma': 'asf',
'wmv': 'asf',
+ 'weba': 'webm',
'vtt': 'webvtt',
}
ACODECS = {
- 'mp3': 'libmp3lame',
- 'aac': 'aac',
- 'flac': 'flac',
- 'm4a': 'aac',
- 'opus': 'libopus',
- 'vorbis': 'libvorbis',
- 'wav': None,
- 'alac': None,
+ # name: (ext, encoder, opts)
+ 'mp3': ('mp3', 'libmp3lame', ()),
+ 'aac': ('m4a', 'aac', ('-f', 'adts')),
+ 'm4a': ('m4a', 'aac', ('-bsf:a', 'aac_adtstoasc')),
+ 'opus': ('opus', 'libopus', ()),
+ 'vorbis': ('ogg', 'libvorbis', ()),
+ 'flac': ('flac', 'flac', ()),
+ 'alac': ('m4a', None, ('-acodec', 'alac')),
+ 'wav': ('wav', None, ('-f', 'wav')),
}
+def create_mapping_re(supported):
+ return re.compile(r'{0}(?:/{0})*$'.format(r'(?:\s*\w+\s*>)?\s*(?:{})\s*'.format('|'.join(supported))))
+
+
+def resolve_mapping(source, mapping):
+ """
+ Get corresponding item from a mapping string like 'A>B/C>D/E'
+ @returns (target, error_message)
+ """
+ for pair in mapping.lower().split('/'):
+ kv = pair.split('>', 1)
+ if len(kv) == 1 or kv[0].strip() == source:
+ target = kv[-1].strip()
+ if target == source:
+ return target, f'already is in target format {source}'
+ return target, None
+ return None, f'could not find a mapping for {source}'
+
+
class FFmpegPostProcessorError(PostProcessingError):
pass
class FFmpegPostProcessor(PostProcessor):
+ _ffmpeg_location = contextvars.ContextVar('ffmpeg_location', default=None)
+
def __init__(self, downloader=None):
PostProcessor.__init__(self, downloader)
- self._determine_executables()
-
- def check_version(self):
- if not self.available:
- raise FFmpegPostProcessorError('ffmpeg not found. Please install or provide the path using --ffmpeg-location')
-
- required_version = '10-0' if self.basename == 'avconv' else '1.0'
- if is_outdated_version(
- self._versions[self.basename], required_version):
- warning = 'Your copy of %s is outdated, update %s to version %s or newer if you encounter any errors.' % (
- self.basename, self.basename, required_version)
- self.report_warning(warning)
+ self._prefer_ffmpeg = self.get_param('prefer_ffmpeg', True)
+ self._paths = self._determine_executables()
@staticmethod
def get_versions_and_features(downloader=None):
@staticmethod
def get_versions(downloader=None):
- return FFmpegPostProcessor.get_version_and_features(downloader)[0]
+ return FFmpegPostProcessor.get_versions_and_features(downloader)[0]
+
+ _ffmpeg_to_avconv = {'ffmpeg': 'avconv', 'ffprobe': 'avprobe'}
def _determine_executables(self):
- programs = ['avprobe', 'avconv', 'ffmpeg', 'ffprobe']
-
- def get_ffmpeg_version(path, prog):
- out = _get_exe_version_output(path, ['-bsfs'])
- ver = detect_exe_version(out) if out else False
- if ver:
- regexs = [
- r'(?:\d+:)?([0-9.]+)-[0-9]+ubuntu[0-9.]+$', # Ubuntu, see [1]
- r'n([0-9.]+)$', # Arch Linux
- # 1. http://www.ducea.com/2006/06/17/ubuntu-package-version-naming-explanation/
- ]
- for regex in regexs:
- mobj = re.match(regex, ver)
- if mobj:
- ver = mobj.group(1)
- self._versions[prog] = ver
- if prog != 'ffmpeg' or not out:
- return
+ programs = [*self._ffmpeg_to_avconv.keys(), *self._ffmpeg_to_avconv.values()]
- mobj = re.search(r'(?m)^\s+libavformat\s+(?:[0-9. ]+)\s+/\s+(?P<runtime>[0-9. ]+)', out)
- lavf_runtime_version = mobj.group('runtime').replace(' ', '') if mobj else None
- self._features = {
- 'fdk': '--enable-libfdk-aac' in out,
- 'setts': 'setts' in out.splitlines(),
- 'needs_adtstoasc': is_outdated_version(lavf_runtime_version, '57.56.100', False),
- }
-
- self.basename = None
- self.probe_basename = None
- self._paths = None
- self._versions = None
- self._features = {}
-
- prefer_ffmpeg = self.get_param('prefer_ffmpeg', True)
- location = self.get_param('ffmpeg_location')
+ location = self.get_param('ffmpeg_location', self._ffmpeg_location.get())
if location is None:
- self._paths = {p: p for p in programs}
- else:
- if not os.path.exists(location):
- self.report_warning(
- 'ffmpeg-location %s does not exist! '
- 'Continuing without ffmpeg.' % (location))
- self._versions = {}
- return
- elif os.path.isdir(location):
- dirname, basename = location, None
- else:
- basename = os.path.splitext(os.path.basename(location))[0]
- basename = next((p for p in programs if basename.startswith(p)), 'ffmpeg')
- dirname = os.path.dirname(os.path.abspath(location))
- if basename in ('ffmpeg', 'ffprobe'):
- prefer_ffmpeg = True
-
- self._paths = dict(
- (p, os.path.join(dirname, p)) for p in programs)
- if basename:
- self._paths[basename] = location
-
- self._versions = {}
- for p in programs:
- get_ffmpeg_version(self._paths[p], p)
-
- if prefer_ffmpeg is False:
- prefs = ('avconv', 'ffmpeg')
+ return {p: p for p in programs}
+
+ if not os.path.exists(location):
+ self.report_warning(
+ f'ffmpeg-location {location} does not exist! Continuing without ffmpeg', only_once=True)
+ return {}
+ elif os.path.isdir(location):
+ dirname, basename, filename = location, None, None
else:
- prefs = ('ffmpeg', 'avconv')
- for p in prefs:
- if self._versions[p]:
- self.basename = p
- break
-
- if prefer_ffmpeg is False:
- prefs = ('avprobe', 'ffprobe')
+ filename = os.path.basename(location)
+ basename = next((p for p in programs if p in filename), 'ffmpeg')
+ dirname = os.path.dirname(os.path.abspath(location))
+ if basename in self._ffmpeg_to_avconv:
+ self._prefer_ffmpeg = True
+
+ paths = {p: os.path.join(dirname, p) for p in programs}
+ if basename and basename in filename:
+ for p in programs:
+ path = os.path.join(dirname, filename.replace(basename, p))
+ if os.path.exists(path):
+ paths[p] = path
+ if basename:
+ paths[basename] = location
+ return paths
+
+ _version_cache, _features_cache = {None: None}, {}
+
+ def _get_ffmpeg_version(self, prog):
+ path = self._paths.get(prog)
+ if path in self._version_cache:
+ return self._version_cache[path], self._features_cache.get(path, {})
+ out = _get_exe_version_output(path, ['-bsfs'])
+ ver = detect_exe_version(out) if out else False
+ if ver:
+ regexs = [
+ r'(?:\d+:)?([0-9.]+)-[0-9]+ubuntu[0-9.]+$', # Ubuntu, see [1]
+ r'n([0-9.]+)$', # Arch Linux
+ # 1. http://www.ducea.com/2006/06/17/ubuntu-package-version-naming-explanation/
+ ]
+ for regex in regexs:
+ mobj = re.match(regex, ver)
+ if mobj:
+ ver = mobj.group(1)
+ self._version_cache[path] = ver
+ if prog != 'ffmpeg' or not out:
+ return ver, {}
+
+ mobj = re.search(r'(?m)^\s+libavformat\s+(?:[0-9. ]+)\s+/\s+(?P<runtime>[0-9. ]+)', out)
+ lavf_runtime_version = mobj.group('runtime').replace(' ', '') if mobj else None
+ self._features_cache[path] = features = {
+ 'fdk': '--enable-libfdk-aac' in out,
+ 'setts': 'setts' in out.splitlines(),
+ 'needs_adtstoasc': is_outdated_version(lavf_runtime_version, '57.56.100', False),
+ }
+ return ver, features
+
+ @property
+ def _versions(self):
+ return filter_dict({self.basename: self._version, self.probe_basename: self._probe_version})
+
+ @functools.cached_property
+ def basename(self):
+ _ = self._version # run property
+ return self.basename
+
+ @functools.cached_property
+ def probe_basename(self):
+ _ = self._probe_version # run property
+ return self.probe_basename
+
+ def _get_version(self, kind):
+ executables = (kind, )
+ if not self._prefer_ffmpeg:
+ executables = (kind, self._ffmpeg_to_avconv[kind])
+ basename, version, features = next(filter(
+ lambda x: x[1], ((p, *self._get_ffmpeg_version(p)) for p in executables)), (None, None, {}))
+ if kind == 'ffmpeg':
+ self.basename, self._features = basename, features
else:
- prefs = ('ffprobe', 'avprobe')
- for p in prefs:
- if self._versions[p]:
- self.probe_basename = p
- break
-
- if self.basename == 'avconv':
- self.deprecation_warning(
- 'Support for avconv is deprecated and may be removed in a future version. Use ffmpeg instead')
- if self.probe_basename == 'avprobe':
- self.deprecation_warning(
- 'Support for avprobe is deprecated and may be removed in a future version. Use ffprobe instead')
+ self.probe_basename = basename
+ if basename == self._ffmpeg_to_avconv[kind]:
+ self.deprecated_feature(f'Support for {self._ffmpeg_to_avconv[kind]} is deprecated and '
+ f'may be removed in a future version. Use {kind} instead')
+ return version
+
+ @functools.cached_property
+ def _version(self):
+ return self._get_version('ffmpeg')
+
+ @functools.cached_property
+ def _probe_version(self):
+ return self._get_version('ffprobe')
@property
def available(self):
@property
def executable(self):
- return self._paths[self.basename]
+ return self._paths.get(self.basename)
@property
def probe_available(self):
@property
def probe_executable(self):
- return self._paths[self.probe_basename]
+ return self._paths.get(self.probe_basename)
@staticmethod
def stream_copy_opts(copy=True, *, ext=None):
yield from ('-dn', '-ignore_unknown')
if copy:
yield from ('-c', 'copy')
- # For some reason, '-c copy -map 0' is not enough to copy subtitles
- if ext in ('mp4', 'mov'):
+ if ext in ('mp4', 'mov', 'm4a'):
yield from ('-c:s', 'mov_text')
+ def check_version(self):
+ if not self.available:
+ raise FFmpegPostProcessorError('ffmpeg not found. Please install or provide the path using --ffmpeg-location')
+
+ required_version = '10-0' if self.basename == 'avconv' else '1.0'
+ if is_outdated_version(self._version, required_version):
+ self.report_warning(f'Your copy of {self.basename} is outdated, update {self.basename} '
+ f'to version {required_version} or newer if you encounter any errors')
+
def get_audio_codec(self, path):
if not self.probe_available and not self.available:
raise PostProcessingError('ffprobe and ffmpeg not found. Please install or provide the path using --ffmpeg-location')
encodeFilename(self.executable, True),
encodeArgument('-i')]
cmd.append(encodeFilename(self._ffmpeg_filename_argument(path), True))
- self.write_debug('%s command line: %s' % (self.basename, shell_quote(cmd)))
- handle = Popen(cmd, stdin=subprocess.PIPE, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
- stdout_data, stderr_data = handle.communicate_or_kill()
- expected_ret = 0 if self.probe_available else 1
- if handle.wait() != expected_ret:
+ self.write_debug(f'{self.basename} command line: {shell_quote(cmd)}')
+ stdout, stderr, returncode = Popen.run(
+ cmd, text=True, stdin=subprocess.PIPE, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
+ if returncode != (0 if self.probe_available else 1):
return None
- except (IOError, OSError):
+ except OSError:
return None
- output = (stdout_data if self.probe_available else stderr_data).decode('ascii', 'ignore')
+ output = stdout if self.probe_available else stderr
if self.probe_available:
audio_codec = None
for line in output.split('\n'):
]
cmd += opts
- cmd.append(encodeFilename(self._ffmpeg_filename_argument(path), True))
- self.write_debug('ffprobe command line: %s' % shell_quote(cmd))
- p = Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE, stdin=subprocess.PIPE)
- stdout, stderr = p.communicate()
- return json.loads(stdout.decode('utf-8', 'replace'))
+ cmd.append(self._ffmpeg_filename_argument(path))
+ self.write_debug(f'ffprobe command line: {shell_quote(cmd)}')
+ stdout, _, _ = Popen.run(cmd, text=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE, stdin=subprocess.PIPE)
+ return json.loads(stdout)
def get_stream_number(self, path, keys, value):
streams = self.get_metadata_object(path)['streams']
None)
return num, len(streams)
+ def _fixup_chapters(self, info):
+ last_chapter = traverse_obj(info, ('chapters', -1))
+ if last_chapter and not last_chapter.get('end_time'):
+ last_chapter['end_time'] = self._get_real_video_duration(info['filepath'])
+
def _get_real_video_duration(self, filepath, fatal=True):
try:
duration = float_or_none(
if fatal:
raise PostProcessingError(f'Unable to determine video duration: {e.msg}')
- def _duration_mismatch(self, d1, d2):
+ def _duration_mismatch(self, d1, d2, tolerance=2):
if not d1 or not d2:
return None
# The duration is often only known to nearest second. So there can be <1sec disparity natually.
# Further excuse an additional <1sec difference.
- return abs(d1 - d2) > 2
+ return abs(d1 - d2) > tolerance
def run_ffmpeg_multiple_files(self, input_paths, out_path, opts, **kwargs):
return self.real_run_ffmpeg(
cmd += [encodeArgument('-loglevel'), encodeArgument('repeat+info')]
def make_args(file, args, name, number):
- keys = ['_%s%d' % (name, number), '_%s' % name]
+ keys = [f'_{name}{number}', f'_{name}']
if name == 'o':
args += ['-movflags', '+faststart']
if number == 1:
make_args(path, list(opts), arg_type, i + 1)
for i, (path, opts) in enumerate(path_opts) if path)
- self.write_debug('ffmpeg command line: %s' % shell_quote(cmd))
- p = Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE, stdin=subprocess.PIPE)
- stdout, stderr = p.communicate_or_kill()
- if p.returncode not in variadic(expected_retcodes):
- stderr = stderr.decode('utf-8', 'replace').strip()
+ self.write_debug(f'ffmpeg command line: {shell_quote(cmd)}')
+ _, stderr, returncode = Popen.run(
+ cmd, text=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE, stdin=subprocess.PIPE)
+ if returncode not in variadic(expected_retcodes):
self.write_debug(stderr)
- raise FFmpegPostProcessorError(stderr.split('\n')[-1])
+ raise FFmpegPostProcessorError(stderr.strip().splitlines()[-1])
for out_path, _ in output_path_opts:
if out_path:
self.try_utime(out_path, oldest_mtime, oldest_mtime)
- return stderr.decode('utf-8', 'replace')
+ return stderr
def run_ffmpeg(self, path, out_path, opts, **kwargs):
return self.run_ffmpeg_multiple_files([path], out_path, opts, **kwargs)
"""
concat_file = f'{out_file}.concat'
self.write_debug(f'Writing concat spec to {concat_file}')
- with open(concat_file, 'wt', encoding='utf-8') as f:
+ with open(concat_file, 'w', encoding='utf-8') as f:
f.writelines(self._concat_spec(in_files, concat_opts))
out_flags = list(self.stream_copy_opts(ext=determine_ext(out_file)))
self.real_run_ffmpeg(
[(concat_file, ['-hide_banner', '-nostdin', '-f', 'concat', '-safe', '0'])],
[(out_file, out_flags)])
- os.remove(concat_file)
+ self._delete_downloaded_files(concat_file)
@classmethod
def _concat_spec(cls, in_files, concat_opts=None):
class FFmpegExtractAudioPP(FFmpegPostProcessor):
- COMMON_AUDIO_EXTS = ('wav', 'flac', 'm4a', 'aiff', 'mp3', 'ogg', 'mka', 'opus', 'wma')
- SUPPORTED_EXTS = ('aac', 'flac', 'mp3', 'm4a', 'opus', 'vorbis', 'wav', 'alac')
+ COMMON_AUDIO_EXTS = (*MEDIA_EXTENSIONS.common_audio, 'wma')
+ SUPPORTED_EXTS = tuple(ACODECS.keys())
+ FORMAT_RE = create_mapping_re(('best', *SUPPORTED_EXTS))
def __init__(self, downloader=None, preferredcodec=None, preferredquality=None, nopostoverwrites=False):
FFmpegPostProcessor.__init__(self, downloader)
- self._preferredcodec = preferredcodec or 'best'
+ self.mapping = preferredcodec or 'best'
self._preferredquality = float_or_none(preferredquality)
self._nopostoverwrites = nopostoverwrites
acodec_opts = []
else:
acodec_opts = ['-acodec', codec]
- opts = ['-vn'] + acodec_opts + more_opts
+ opts = ['-vn', *acodec_opts, *more_opts]
try:
FFmpegPostProcessor.run_ffmpeg(self, path, out_path, opts)
except FFmpegPostProcessorError as err:
- raise AudioConversionError(err.msg)
+ raise PostProcessingError(f'audio conversion failed: {err.msg}')
@PostProcessor._restrict_to(images=False)
def run(self, information):
orig_path = path = information['filepath']
- orig_ext = information['ext']
-
- if self._preferredcodec == 'best' and orig_ext in self.COMMON_AUDIO_EXTS:
- self.to_screen('Skipping audio extraction since the file is already in a common audio format')
+ target_format, _skip_msg = resolve_mapping(information['ext'], self.mapping)
+ if target_format == 'best' and information['ext'] in self.COMMON_AUDIO_EXTS:
+ target_format, _skip_msg = None, 'the file is already in a common audio format'
+ if not target_format:
+ self.to_screen(f'Not converting audio {orig_path}; {_skip_msg}')
return [], information
filecodec = self.get_audio_codec(path)
if filecodec is None:
raise PostProcessingError('WARNING: unable to obtain file audio codec with ffprobe')
- more_opts = []
- if self._preferredcodec == 'best' or self._preferredcodec == filecodec or (self._preferredcodec == 'm4a' and filecodec == 'aac'):
- if filecodec == 'aac' and self._preferredcodec in ['m4a', 'best']:
- # Lossless, but in another container
- acodec = 'copy'
- extension = 'm4a'
- more_opts = ['-bsf:a', 'aac_adtstoasc']
- elif filecodec in ['aac', 'flac', 'mp3', 'vorbis', 'opus']:
- # Lossless if possible
- acodec = 'copy'
- extension = filecodec
- if filecodec == 'aac':
- more_opts = ['-f', 'adts']
- if filecodec == 'vorbis':
- extension = 'ogg'
- elif filecodec == 'alac':
- acodec = None
- extension = 'm4a'
- more_opts += ['-acodec', 'alac']
- else:
- # MP3 otherwise.
- acodec = 'libmp3lame'
- extension = 'mp3'
- more_opts = self._quality_args(acodec)
+ if filecodec == 'aac' and target_format in ('m4a', 'best'):
+ # Lossless, but in another container
+ extension, _, more_opts, acodec = *ACODECS['m4a'], 'copy'
+ elif target_format == 'best' or target_format == filecodec:
+ # Lossless if possible
+ try:
+ extension, _, more_opts, acodec = *ACODECS[filecodec], 'copy'
+ except KeyError:
+ extension, acodec, more_opts = ACODECS['mp3']
else:
# We convert the audio (lossy if codec is lossy)
- acodec = ACODECS[self._preferredcodec]
+ extension, acodec, more_opts = ACODECS[target_format]
if acodec == 'aac' and self._features.get('fdk'):
- acodec = 'libfdk_aac'
- extension = self._preferredcodec
+ acodec, more_opts = 'libfdk_aac', []
+
+ more_opts = list(more_opts)
+ if acodec != 'copy':
more_opts = self._quality_args(acodec)
- if self._preferredcodec == 'aac':
- more_opts += ['-f', 'adts']
- elif self._preferredcodec == 'm4a':
- more_opts += ['-bsf:a', 'aac_adtstoasc']
- elif self._preferredcodec == 'vorbis':
- extension = 'ogg'
- elif self._preferredcodec == 'wav':
- extension = 'wav'
- more_opts += ['-f', 'wav']
- elif self._preferredcodec == 'alac':
- extension = 'm4a'
- more_opts += ['-acodec', 'alac']
-
- prefix, sep, ext = path.rpartition('.') # not os.path.splitext, since the latter does not work on unicode in all setups
- temp_path = new_path = prefix + sep + extension
+
+ temp_path = new_path = replace_extension(path, extension, information['ext'])
if new_path == path:
+ if acodec == 'copy':
+ self.to_screen(f'Not converting audio {orig_path}; file is already in target format {target_format}')
+ return [], information
orig_path = prepend_extension(path, 'orig')
temp_path = prepend_extension(path, 'temp')
if (self._nopostoverwrites and os.path.exists(encodeFilename(new_path))
and os.path.exists(encodeFilename(orig_path))):
- self.to_screen('Post-process file %s exists, skipping' % new_path)
+ self.to_screen(f'Post-process file {new_path} exists, skipping')
return [], information
- try:
- self.to_screen(f'Destination: {new_path}')
- self.run_ffmpeg(path, temp_path, acodec, more_opts)
- except AudioConversionError as e:
- raise PostProcessingError(
- 'audio conversion failed: ' + e.msg)
- except Exception:
- raise PostProcessingError('error running ' + self.basename)
+ self.to_screen(f'Destination: {new_path}')
+ self.run_ffmpeg(path, temp_path, acodec, more_opts)
os.replace(path, orig_path)
os.replace(temp_path, new_path)
# Try to update the date time for extracted audio file.
if information.get('filetime') is not None:
self.try_utime(
- new_path, time.time(), information['filetime'],
- errnote='Cannot update utime of audio file')
+ new_path, time.time(), information['filetime'], errnote='Cannot update utime of audio file')
return [orig_path], information
class FFmpegVideoConvertorPP(FFmpegPostProcessor):
- SUPPORTED_EXTS = ('mp4', 'mkv', 'flv', 'webm', 'mov', 'avi', 'mka', 'ogg', *FFmpegExtractAudioPP.SUPPORTED_EXTS)
- FORMAT_RE = re.compile(r'{0}(?:/{0})*$'.format(r'(?:\w+>)?(?:%s)' % '|'.join(SUPPORTED_EXTS)))
+ SUPPORTED_EXTS = (
+ *sorted((*MEDIA_EXTENSIONS.common_video, 'gif')),
+ *sorted((*MEDIA_EXTENSIONS.common_audio, 'aac', 'vorbis')),
+ )
+ FORMAT_RE = create_mapping_re(SUPPORTED_EXTS)
_ACTION = 'converting'
def __init__(self, downloader=None, preferedformat=None):
- super(FFmpegVideoConvertorPP, self).__init__(downloader)
- self._preferedformats = preferedformat.lower().split('/')
-
- def _target_ext(self, source_ext):
- for pair in self._preferedformats:
- kv = pair.split('>')
- if len(kv) == 1 or kv[0].strip() == source_ext:
- return kv[-1].strip()
+ super().__init__(downloader)
+ self.mapping = preferedformat
@staticmethod
def _options(target_ext):
+ yield from FFmpegPostProcessor.stream_copy_opts(False)
if target_ext == 'avi':
- return ['-c:v', 'libxvid', '-vtag', 'XVID']
- return []
+ yield from ('-c:v', 'libxvid', '-vtag', 'XVID')
@PostProcessor._restrict_to(images=False)
def run(self, info):
filename, source_ext = info['filepath'], info['ext'].lower()
- target_ext = self._target_ext(source_ext)
- _skip_msg = (
- f'could not find a mapping for {source_ext}' if not target_ext
- else f'already is in target format {source_ext}' if source_ext == target_ext
- else None)
+ target_ext, _skip_msg = resolve_mapping(source_ext, self.mapping)
if _skip_msg:
self.to_screen(f'Not {self._ACTION} media file "{filename}"; {_skip_msg}')
return [], info
class FFmpegEmbedSubtitlePP(FFmpegPostProcessor):
+ SUPPORTED_EXTS = ('mp4', 'mov', 'm4a', 'webm', 'mkv', 'mka')
+
def __init__(self, downloader=None, already_have_subtitle=False):
- super(FFmpegEmbedSubtitlePP, self).__init__(downloader)
+ super().__init__(downloader)
self._already_have_subtitle = already_have_subtitle
@PostProcessor._restrict_to(images=False)
def run(self, info):
- if info['ext'] not in ('mp4', 'webm', 'mkv'):
- self.to_screen('Subtitles can only be embedded in mp4, webm or mkv files')
+ if info['ext'] not in self.SUPPORTED_EXTS:
+ self.to_screen(f'Subtitles can only be embedded in {", ".join(self.SUPPORTED_EXTS)} files')
return [], info
subtitles = info.get('requested_subtitles')
if not subtitles:
filename = info['filepath']
- # Disabled temporarily. There needs to be a way to overide this
+ # Disabled temporarily. There needs to be a way to override this
# in case of duration actually mismatching in extractor
# See: https://github.com/yt-dlp/yt-dlp/issues/1870, https://github.com/yt-dlp/yt-dlp/issues/1385
'''
if not sub_langs:
return [], info
- input_files = [filename] + sub_filenames
+ input_files = [filename, *sub_filenames]
opts = [
*self.stream_copy_opts(ext=info['ext']),
'-map', '-0:s',
]
for i, (lang, name) in enumerate(zip(sub_langs, sub_names)):
- opts.extend(['-map', '%d:0' % (i + 1)])
+ opts.extend(['-map', f'{i + 1}:0'])
lang_code = ISO639Utils.short2long(lang) or lang
- opts.extend(['-metadata:s:s:%d' % i, 'language=%s' % lang_code])
+ opts.extend([f'-metadata:s:s:{i}', f'language={lang_code}'])
if name:
- opts.extend(['-metadata:s:s:%d' % i, 'handler_name=%s' % name,
- '-metadata:s:s:%d' % i, 'title=%s' % name])
+ opts.extend([f'-metadata:s:s:{i}', f'handler_name={name}',
+ f'-metadata:s:s:{i}', f'title={name}'])
temp_filename = prepend_extension(filename, 'temp')
- self.to_screen('Embedding subtitles in "%s"' % filename)
+ self.to_screen(f'Embedding subtitles in "{filename}"')
self.run_ffmpeg_multiple_files(input_files, temp_filename, opts)
os.replace(temp_filename, filename)
@PostProcessor._restrict_to(images=False)
def run(self, info):
+ self._fixup_chapters(info)
filename, metadata_filename = info['filepath'], None
files_to_delete, options = [], []
if self._add_chapters and info.get('chapters'):
return [], info
temp_filename = prepend_extension(filename, 'temp')
- self.to_screen('Adding metadata to "%s"' % filename)
+ self.to_screen(f'Adding metadata to "{filename}"')
self.run_ffmpeg_multiple_files(
(filename, metadata_filename), temp_filename,
itertools.chain(self._options(info['ext']), *options))
- for file in filter(None, files_to_delete):
- os.remove(file) # Don't obey --keep-files
+ self._delete_downloaded_files(*files_to_delete)
os.replace(temp_filename, filename)
return [], info
@staticmethod
def _get_chapter_opts(chapters, metadata_filename):
- with io.open(metadata_filename, 'wt', encoding='utf-8') as f:
+ with open(metadata_filename, 'w', encoding='utf-8') as f:
def ffmpeg_escape(text):
return re.sub(r'([\\=;#\n])', r'\\\1', text)
metadata_file_content += 'END=%d\n' % (chapter['end_time'] * 1000)
chapter_title = chapter.get('title')
if chapter_title:
- metadata_file_content += 'title=%s\n' % ffmpeg_escape(chapter_title)
+ metadata_file_content += f'title={ffmpeg_escape(chapter_title)}\n'
f.write(metadata_file_content)
yield ('-map_metadata', '1')
def add(meta_list, info_list=None):
value = next((
- str(info[key]) for key in [f'{meta_prefix}_'] + list(variadic(info_list or meta_list))
+ info[key] for key in [f'{meta_prefix}_', *variadic(info_list or meta_list)]
if info.get(key) is not None), None)
if value not in ('', None):
+ value = ', '.join(map(str, variadic(value)))
+ value = value.replace('\0', '') # nul character cannot be passed in command line
metadata['common'].update({meta_f: value for meta_f in variadic(meta_list)})
- # See [1-4] for some info on media metadata/metadata supported
- # by ffmpeg.
- # 1. https://kdenlive.org/en/project/adding-meta-data-to-mp4-video/
- # 2. https://wiki.multimedia.cx/index.php/FFmpeg_Metadata
- # 3. https://kodi.wiki/view/Video_file_tagging
+ # Info on media metadata/metadata supported by ffmpeg:
+ # https://wiki.multimedia.cx/index.php/FFmpeg_Metadata
+ # https://kdenlive.org/en/project/adding-meta-data-to-mp4-video/
+ # https://kodi.wiki/view/Video_file_tagging
add('title', ('track', 'title'))
add('date', 'upload_date')
add(('description', 'synopsis'), 'description')
add(('purl', 'comment'), 'webpage_url')
add('track', 'track_number')
- add('artist', ('artist', 'creator', 'uploader', 'uploader_id'))
- add('genre')
+ add('artist', ('artist', 'artists', 'creator', 'creators', 'uploader', 'uploader_id'))
+ add('composer', ('composer', 'composers'))
+ add('genre', ('genre', 'genres'))
add('album')
- add('album_artist')
+ add('album_artist', ('album_artist', 'album_artists'))
add('disc', 'disc_number')
add('show', 'series')
add('season_number')
for key, value in info.items():
mobj = re.fullmatch(meta_regex, key)
if value is not None and mobj:
- metadata[mobj.group('i') or 'common'][mobj.group('key')] = value
+ metadata[mobj.group('i') or 'common'][mobj.group('key')] = value.replace('\0', '')
+
+ # Write id3v1 metadata also since Windows Explorer can't handle id3v2 tags
+ yield ('-write_id3v1', '1')
for name, value in metadata['common'].items():
yield ('-metadata', f'{name}={value}')
stream_idx = 0
- for fmt in info.get('requested_formats') or []:
+ for fmt in info.get('requested_formats') or [info]:
stream_count = 2 if 'none' not in (fmt.get('vcodec'), fmt.get('acodec')) else 1
lang = ISO639Utils.short2long(fmt.get('language') or '') or fmt.get('language')
for i in range(stream_idx, stream_idx + stream_count):
old_stream, new_stream = self.get_stream_number(info['filepath'], ('tags', 'mimetype'), 'application/json')
if old_stream is not None:
- yield ('-map', '-0:%d' % old_stream)
+ yield ('-map', f'-0:{old_stream}')
new_stream -= 1
- yield ('-attach', infofn,
- '-metadata:s:%d' % new_stream, 'mimetype=application/json')
+ yield (
+ '-attach', self._ffmpeg_filename_argument(infofn),
+ f'-metadata:s:{new_stream}', 'mimetype=application/json',
+ f'-metadata:s:{new_stream}', 'filename=info.json',
+ )
class FFmpegMergerPP(FFmpegPostProcessor):
+ SUPPORTED_EXTS = MEDIA_EXTENSIONS.common_video
+
@PostProcessor._restrict_to(images=False)
def run(self, info):
filename = info['filepath']
args.extend([f'-bsf:a:{audio_streams}', 'aac_adtstoasc'])
audio_streams += 1
if fmt.get('vcodec') != 'none':
- args.extend(['-map', '%u:v:0' % (i)])
- self.to_screen('Merging formats into "%s"' % filename)
+ args.extend(['-map', f'{i}:v:0'])
+ self.to_screen(f'Merging formats into "{filename}"')
self.run_ffmpeg_multiple_files(info['__files_to_merge'], temp_filename, args)
os.rename(encodeFilename(temp_filename), encodeFilename(filename))
return info['__files_to_merge'], info
required_version = '10-0'
if is_outdated_version(
self._versions[self.basename], required_version):
- warning = ('Your copy of %s is outdated and unable to properly mux separate video and audio files, '
+ warning = (f'Your copy of {self.basename} is outdated and unable to properly mux separate video and audio files, '
'yt-dlp will download single file media. '
- 'Update %s to version %s or newer to fix this.') % (
- self.basename, self.basename, required_version)
+ f'Update {self.basename} to version {required_version} or newer to fix this.')
self.report_warning(warning)
return False
return True
stretched_ratio = info.get('stretched_ratio')
if stretched_ratio not in (None, 1):
self._fixup('Fixing aspect ratio', info['filepath'], [
- *self.stream_copy_opts(), '-aspect', '%f' % stretched_ratio])
+ *self.stream_copy_opts(), '-aspect', f'{stretched_ratio:f}'])
return [], info
@PostProcessor._restrict_to(images=False)
def run(self, info):
if all(self._needs_fixup(info)):
+ args = ['-f', 'mp4']
+ if self.get_audio_codec(info['filepath']) == 'aac':
+ args.extend(['-bsf:a', 'aac_adtstoasc'])
self._fixup('Fixing MPEG-TS in MP4 container', info['filepath'], [
- *self.stream_copy_opts(), '-f', 'mp4', '-bsf:a', 'aac_adtstoasc'])
+ *self.stream_copy_opts(), *args])
return [], info
def __init__(self, downloader=None, trim=0.001):
# "trim" should be used when the video contains unintended packets
- super(FFmpegFixupTimestampPP, self).__init__(downloader)
+ super().__init__(downloader)
assert isinstance(trim, (int, float))
self.trim = str(trim)
opts = ['-vf', 'setpts=PTS-STARTPTS']
else:
opts = ['-c', 'copy', '-bsf', 'setts=ts=TS-STARTPTS']
- self._fixup('Fixing frame timestamp', info['filepath'], opts + [*self.stream_copy_opts(False), '-ss', self.trim])
+ self._fixup('Fixing frame timestamp', info['filepath'], [*opts, *self.stream_copy_opts(False), '-ss', self.trim])
return [], info
class FFmpegSubtitlesConvertorPP(FFmpegPostProcessor):
- SUPPORTED_EXTS = ('srt', 'vtt', 'ass', 'lrc')
+ SUPPORTED_EXTS = MEDIA_EXTENSIONS.subtitles
def __init__(self, downloader=None, format=None):
- super(FFmpegSubtitlesConvertorPP, self).__init__(downloader)
+ super().__init__(downloader)
self.format = format
def run(self, info):
continue
ext = sub['ext']
if ext == new_ext:
- self.to_screen('Subtitle file for %s is already in the requested format' % new_ext)
+ self.to_screen(f'Subtitle file for {new_ext} is already in the requested format')
continue
elif ext == 'json':
self.to_screen(
with open(dfxp_file, 'rb') as f:
srt_data = dfxp2srt(f.read())
- with io.open(srt_file, 'wt', encoding='utf-8') as f:
+ with open(srt_file, 'w', encoding='utf-8') as f:
f.write(srt_data)
old_file = srt_file
self.run_ffmpeg(old_file, new_file, ['-f', new_format])
- with io.open(new_file, 'rt', encoding='utf-8') as f:
+ with open(new_file, encoding='utf-8') as f:
subs[lang] = {
'ext': new_ext,
'data': f.read(),
self.to_screen('Chapter %03d; Destination: %s' % (number, destination))
return (
destination,
- ['-ss', compat_str(chapter['start_time']),
- '-t', compat_str(chapter['end_time'] - chapter['start_time'])])
+ ['-ss', str(chapter['start_time']),
+ '-t', str(chapter['end_time'] - chapter['start_time'])])
@PostProcessor._restrict_to(images=False)
def run(self, info):
+ self._fixup_chapters(info)
chapters = info.get('chapters') or []
if not chapters:
self.to_screen('Chapter information is unavailable')
in_file = info['filepath']
if self._force_keyframes and len(chapters) > 1:
in_file = self.force_keyframes(in_file, (c['start_time'] for c in chapters))
- self.to_screen('Splitting video by chapters; %d chapters found' % len(chapters))
+ self.to_screen(f'Splitting video by chapters; {len(chapters)} chapters found')
for idx, chapter in enumerate(chapters):
destination, opts = self._ffmpeg_args_for_chapter(idx + 1, chapter, info)
self.real_run_ffmpeg([(in_file, opts)], [(destination, self.stream_copy_opts())])
if in_file != info['filepath']:
- os.remove(in_file)
+ self._delete_downloaded_files(in_file, msg=None)
return [], info
class FFmpegThumbnailsConvertorPP(FFmpegPostProcessor):
- SUPPORTED_EXTS = ('jpg', 'png', 'webp')
+ SUPPORTED_EXTS = MEDIA_EXTENSIONS.thumbnails
+ FORMAT_RE = create_mapping_re(SUPPORTED_EXTS)
def __init__(self, downloader=None, format=None):
- super(FFmpegThumbnailsConvertorPP, self).__init__(downloader)
- self.format = format
+ super().__init__(downloader)
+ self.mapping = format
- @staticmethod
- def is_webp(path):
- with open(encodeFilename(path), 'rb') as f:
- b = f.read(12)
- return b[0:4] == b'RIFF' and b[8:] == b'WEBP'
+ @classmethod
+ def is_webp(cls, path):
+ deprecation_warning(f'{cls.__module__}.{cls.__name__}.is_webp is deprecated')
+ return imghdr.what(path) == 'webp'
def fixup_webp(self, info, idx=-1):
thumbnail_filename = info['thumbnails'][idx]['filepath']
_, thumbnail_ext = os.path.splitext(thumbnail_filename)
if thumbnail_ext:
- thumbnail_ext = thumbnail_ext[1:].lower()
- if thumbnail_ext != 'webp' and self.is_webp(thumbnail_filename):
- self.to_screen('Correcting thumbnail "%s" extension to webp' % thumbnail_filename)
+ if thumbnail_ext.lower() != '.webp' and imghdr.what(thumbnail_filename) == 'webp':
+ self.to_screen(f'Correcting thumbnail "{thumbnail_filename}" extension to webp')
webp_filename = replace_extension(thumbnail_filename, 'webp')
os.replace(thumbnail_filename, webp_filename)
info['thumbnails'][idx]['filepath'] = webp_filename
@staticmethod
def _options(target_ext):
+ yield from ('-update', '1')
if target_ext == 'jpg':
- return ['-bsf:v', 'mjpeg2jpeg']
- return []
+ yield from ('-bsf:v', 'mjpeg2jpeg')
def convert_thumbnail(self, thumbnail_filename, target_ext):
thumbnail_conv_filename = replace_extension(thumbnail_filename, target_ext)
- self.to_screen('Converting thumbnail "%s" to %s' % (thumbnail_filename, target_ext))
+ self.to_screen(f'Converting thumbnail "{thumbnail_filename}" to {target_ext}')
+ _, source_ext = os.path.splitext(thumbnail_filename)
self.real_run_ffmpeg(
- [(thumbnail_filename, ['-f', 'image2', '-pattern_type', 'none'])],
- [(thumbnail_conv_filename.replace('%', '%%'), self._options(target_ext))])
+ [(thumbnail_filename, [] if source_ext == '.gif' else ['-f', 'image2', '-pattern_type', 'none'])],
+ [(thumbnail_conv_filename, self._options(target_ext))])
return thumbnail_conv_filename
def run(self, info):
continue
has_thumbnail = True
self.fixup_webp(info, idx)
- _, thumbnail_ext = os.path.splitext(original_thumbnail)
- if thumbnail_ext:
- thumbnail_ext = thumbnail_ext[1:].lower()
+ original_thumbnail = thumbnail_dict['filepath'] # Path can change during fixup
+ thumbnail_ext = os.path.splitext(original_thumbnail)[1][1:].lower()
if thumbnail_ext == 'jpeg':
thumbnail_ext = 'jpg'
- if thumbnail_ext == self.format:
- self.to_screen('Thumbnail "%s" is already in the requested format' % original_thumbnail)
+ target_ext, _skip_msg = resolve_mapping(thumbnail_ext, self.mapping)
+ if _skip_msg:
+ self.to_screen(f'Not converting thumbnail "{original_thumbnail}"; {_skip_msg}')
continue
- thumbnail_dict['filepath'] = self.convert_thumbnail(original_thumbnail, self.format)
+ thumbnail_dict['filepath'] = self.convert_thumbnail(original_thumbnail, target_ext)
files_to_delete.append(original_thumbnail)
info['__files_to_move'][thumbnail_dict['filepath']] = replace_extension(
- info['__files_to_move'][original_thumbnail], self.format)
+ info['__files_to_move'][original_thumbnail], target_ext)
if not has_thumbnail:
self.to_screen('There aren\'t any thumbnails to convert')
self._only_multi_video = only_multi_video
super().__init__(downloader)
+ def _get_codecs(self, file):
+ codecs = traverse_obj(self.get_metadata_object(file), ('streams', ..., 'codec_name'))
+ self.write_debug(f'Codecs = {", ".join(codecs)}')
+ return tuple(codecs)
+
def concat_files(self, in_files, out_file):
if not self._downloader._ensure_dir_exists(out_file):
return
os.replace(in_files[0], out_file)
return []
- codecs = [traverse_obj(self.get_metadata_object(file), ('streams', ..., 'codec_name')) for file in in_files]
- if len(set(map(tuple, codecs))) > 1:
+ if len(set(map(self._get_codecs, in_files))) > 1:
raise PostProcessingError(
'The files have different streams/codecs and cannot be concatenated. '
'Either select different formats or --recode-video them to a common format')
entries = info.get('entries') or []
if not any(entries) or (self._only_multi_video and info['_type'] != 'multi_video'):
return [], info
- elif any(len(entry) > 1 for entry in traverse_obj(entries, (..., 'requested_downloads')) or []):
+ elif traverse_obj(entries, (..., lambda k, v: k == 'requested_downloads' and len(v) > 1)):
raise PostProcessingError('Concatenation is not supported when downloading multiple separate formats')
in_files = traverse_obj(entries, (..., 'requested_downloads', 0, 'filepath')) or []
if len(in_files) < len(entries):
raise PostProcessingError('Aborting concatenation because some downloads failed')
- ie_copy = self._downloader._playlist_infodict(info)
exts = traverse_obj(entries, (..., 'requested_downloads', 0, 'ext'), (..., 'ext'))
- ie_copy['ext'] = exts[0] if len(set(exts)) == 1 else 'mkv'
+ ie_copy = collections.ChainMap({'ext': exts[0] if len(set(exts)) == 1 else 'mkv'},
+ info, self._downloader._playlist_infodict(info))
out_file = self._downloader.prepare_filename(ie_copy, 'pl_video')
files_to_delete = self.concat_files(in_files, out_file)