yt-dlp/yt_dlp/postprocessor/ffmpeg.py

1181 lines
48 KiB
Python
Raw Normal View History

import collections
import contextvars
import itertools
import json
import os
import re
import subprocess
import time
2022-06-06 18:19:57 +02:00
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,
is_outdated_version,
orderedSet,
prepend_extension,
replace_extension,
shell_quote,
traverse_obj,
2021-07-10 23:59:44 +02:00
variadic,
write_json_file,
)
EXT_TO_OUT_FORMATS = {
'aac': 'adts',
'flac': 'flac',
'm4a': 'ipod',
'mka': 'matroska',
'mkv': 'matroska',
'mpg': 'mpeg',
'ogv': 'ogg',
'ts': 'mpegts',
'wma': 'asf',
'wmv': 'asf',
'vtt': 'webvtt',
}
ACODECS = {
2022-06-06 18:19:57 +02:00
# 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):
2022-06-06 18:19:57 +02:00
return re.compile(r'{0}(?:/{0})*$'.format(r'(?:\s*\w+\s*>)?\s*(?:%s)\s*' % '|'.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._prefer_ffmpeg = self.get_param('prefer_ffmpeg', True)
self._paths = self._determine_executables()
@staticmethod
def get_versions_and_features(downloader=None):
pp = FFmpegPostProcessor(downloader)
return pp._versions, pp._features
@staticmethod
2015-02-13 11:14:01 +01:00
def get_versions(downloader=None):
return FFmpegPostProcessor.get_versions_and_features(downloader)[0]
_ffmpeg_to_avconv = {'ffmpeg': 'avconv', 'ffprobe': 'avprobe'}
2015-02-13 11:14:01 +01:00
def _determine_executables(self):
programs = [*self._ffmpeg_to_avconv.keys(), *self._ffmpeg_to_avconv.values()]
location = self.get_param('ffmpeg_location', self._ffmpeg_location.get())
if location is None:
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:
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.keys():
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:
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')
2021-11-29 18:46:06 +01:00
2015-02-17 17:26:41 +01:00
@property
2015-02-13 11:14:01 +01:00
def available(self):
return self.basename is not None
2015-02-13 11:14:01 +01:00
@property
def executable(self):
return self._paths.get(self.basename)
2015-02-13 11:14:01 +01:00
@property
def probe_available(self):
return self.probe_basename is not None
2015-02-13 11:14:01 +01:00
@property
def probe_executable(self):
return self._paths.get(self.probe_basename)
@staticmethod
def stream_copy_opts(copy=True, *, ext=None):
yield from ('-map', '0')
# Don't copy Apple TV chapters track, bin_data
# See https://github.com/yt-dlp/yt-dlp/issues/2, #19042, #19024, https://trac.ffmpeg.org/ticket/6016
yield from ('-dn', '-ignore_unknown')
if copy:
yield from ('-c', 'copy')
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')
try:
if self.probe_available:
cmd = [
encodeFilename(self.probe_executable, True),
encodeArgument('-show_streams')]
else:
cmd = [
encodeFilename(self.executable, True),
encodeArgument('-i')]
cmd.append(encodeFilename(self._ffmpeg_filename_argument(path), True))
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 OSError:
return None
output = stdout if self.probe_available else stderr
if self.probe_available:
audio_codec = None
for line in output.split('\n'):
if line.startswith('codec_name='):
audio_codec = line.split('=')[1].strip()
elif line.strip() == 'codec_type=audio' and audio_codec is not None:
return audio_codec
else:
# Stream #FILE_INDEX:STREAM_INDEX[STREAM_ID](LANGUAGE): CODEC_TYPE: CODEC_NAME
mobj = re.search(
r'Stream\s*#\d+:\d+(?:\[0x[0-9a-f]+\])?(?:\([a-z]{3}\))?:\s*Audio:\s*([0-9a-z]+)',
output)
if mobj:
return mobj.group(1)
return None
def get_metadata_object(self, path, opts=[]):
if self.probe_basename != 'ffprobe':
if self.probe_available:
self.report_warning('Only ffprobe is supported for metadata extraction')
raise PostProcessingError('ffprobe not found. Please install or provide the path using --ffmpeg-location')
self.check_version()
cmd = [
encodeFilename(self.probe_executable, True),
encodeArgument('-hide_banner'),
encodeArgument('-show_format'),
encodeArgument('-show_streams'),
encodeArgument('-print_format'),
encodeArgument('json'),
]
cmd += opts
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']
num = next(
(i for i, stream in enumerate(streams) if traverse_obj(stream, keys, casesense=False) == value),
None)
return num, len(streams)
def _get_real_video_duration(self, filepath, fatal=True):
try:
duration = float_or_none(
traverse_obj(self.get_metadata_object(filepath), ('format', 'duration')))
if not duration:
raise PostProcessingError('ffprobe returned empty duration')
return duration
except PostProcessingError as e:
if fatal:
raise PostProcessingError(f'Unable to determine video duration: {e.msg}')
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) > tolerance
def run_ffmpeg_multiple_files(self, input_paths, out_path, opts, **kwargs):
return self.real_run_ffmpeg(
[(path, []) for path in input_paths],
[(out_path, opts)], **kwargs)
def real_run_ffmpeg(self, input_path_opts, output_path_opts, *, expected_retcodes=(0,)):
self.check_version()
oldest_mtime = min(
os.stat(encodeFilename(path)).st_mtime for path, _ in input_path_opts if path)
cmd = [encodeFilename(self.executable, True), encodeArgument('-y')]
# avconv does not have repeat option
if self.basename == 'ffmpeg':
cmd += [encodeArgument('-loglevel'), encodeArgument('repeat+info')]
def make_args(file, args, name, number):
keys = ['_%s%d' % (name, number), '_%s' % name]
if name == 'o':
args += ['-movflags', '+faststart']
if number == 1:
keys.append('')
args += self._configuration_args(self.basename, keys)
if name == 'i':
args.append('-i')
return (
[encodeArgument(arg) for arg in args]
+ [encodeFilename(self._ffmpeg_filename_argument(file), True)])
for arg_type, path_opts in (('i', input_path_opts), ('o', output_path_opts)):
cmd += itertools.chain.from_iterable(
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))
_, 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.strip().splitlines()[-1])
for out_path, _ in output_path_opts:
if out_path:
self.try_utime(out_path, oldest_mtime, oldest_mtime)
return stderr
def run_ffmpeg(self, path, out_path, opts, **kwargs):
return self.run_ffmpeg_multiple_files([path], out_path, opts, **kwargs)
@staticmethod
def _ffmpeg_filename_argument(fn):
# Always use 'file:' because the filename may contain ':' (ffmpeg
# interprets that as a protocol) or can start with '-' (-- is broken in
# ffmpeg, see https://ffmpeg.org/trac/ffmpeg/ticket/2127 for details)
# Also leave '-' intact in order not to break streaming to stdout.
if fn.startswith(('http://', 'https://')):
return fn
return 'file:' + fn if fn != '-' else fn
@staticmethod
def _quote_for_ffmpeg(string):
# See https://ffmpeg.org/ffmpeg-utils.html#toc-Quoting-and-escaping
# A sequence of '' produces '\'''\'';
# final replace removes the empty '' between \' \'.
string = string.replace("'", r"'\''").replace("'''", "'")
# Handle potential ' at string boundaries.
string = string[1:] if string[0] == "'" else "'" + string
return string[:-1] if string[-1] == "'" else string + "'"
def force_keyframes(self, filename, timestamps):
timestamps = orderedSet(timestamps)
if timestamps[0] == 0:
timestamps = timestamps[1:]
keyframe_file = prepend_extension(filename, 'keyframes.temp')
self.to_screen(f'Re-encoding "{filename}" with appropriate keyframes')
self.run_ffmpeg(filename, keyframe_file, [
*self.stream_copy_opts(False, ext=determine_ext(filename)),
'-force_key_frames', ','.join(f'{t:.6f}' for t in timestamps)])
return keyframe_file
def concat_files(self, in_files, out_file, concat_opts=None):
"""
Use concat demuxer to concatenate multiple files having identical streams.
Only inpoint, outpoint, and duration concat options are supported.
See https://ffmpeg.org/ffmpeg-formats.html#concat-1 for details
"""
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:
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)])
2022-05-01 01:28:26 +02:00
self._delete_downloaded_files(concat_file)
@classmethod
def _concat_spec(cls, in_files, concat_opts=None):
if concat_opts is None:
concat_opts = [{}] * len(in_files)
yield 'ffconcat version 1.0\n'
for file, opts in zip(in_files, concat_opts):
yield f'file {cls._quote_for_ffmpeg(cls._ffmpeg_filename_argument(file))}\n'
# Iterate explicitly to yield the following directives in order, ignoring the rest.
for directive in 'inpoint', 'outpoint', 'duration':
if directive in opts:
yield f'{directive} {opts[directive]}\n'
class FFmpegExtractAudioPP(FFmpegPostProcessor):
COMMON_AUDIO_EXTS = MEDIA_EXTENSIONS.common_audio + ('wma', )
2022-06-06 18:19:57 +02:00
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.mapping = preferredcodec or 'best'
self._preferredquality = float_or_none(preferredquality)
self._nopostoverwrites = nopostoverwrites
def _quality_args(self, codec):
if self._preferredquality is None:
return []
elif self._preferredquality > 10:
return ['-b:a', f'{self._preferredquality}k']
limits = {
'libmp3lame': (10, 0),
'libvorbis': (0, 10),
# FFmpeg's AAC encoder does not have an upper limit for the value of -q:a.
# Experimentally, with values over 4, bitrate changes were minimal or non-existent
'aac': (0.1, 4),
'libfdk_aac': (1, 5),
}.get(codec)
if not limits:
return []
q = limits[1] + (limits[0] - limits[1]) * (self._preferredquality / 10)
if codec == 'libfdk_aac':
return ['-vbr', f'{int(q)}']
return ['-q:a', f'{q}']
def run_ffmpeg(self, path, out_path, codec, more_opts):
if codec is None:
acodec_opts = []
else:
acodec_opts = ['-acodec', codec]
opts = ['-vn'] + acodec_opts + more_opts
try:
FFmpegPostProcessor.run_ffmpeg(self, path, out_path, opts)
except FFmpegPostProcessorError as err:
2022-06-06 18:19:57 +02:00
raise PostProcessingError(f'audio conversion failed: {err.msg}')
@PostProcessor._restrict_to(images=False)
def run(self, information):
orig_path = path = information['filepath']
target_format, _skip_msg = resolve_mapping(information['ext'], self.mapping)
2022-06-06 18:19:57 +02:00
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:
2014-11-26 13:05:11 +01:00
raise PostProcessingError('WARNING: unable to obtain file audio codec with ffprobe')
2022-06-06 18:19:57 +02:00
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)
2022-06-06 18:19:57 +02:00
extension, acodec, more_opts = ACODECS[target_format]
if acodec == 'aac' and self._features.get('fdk'):
2022-06-06 18:19:57 +02:00
acodec, more_opts = 'libfdk_aac', []
more_opts = list(more_opts)
if acodec != 'copy':
more_opts = self._quality_args(acodec)
2022-06-06 18:19:57 +02:00
# not os.path.splitext, since the latter does not work on unicode in all setups
temp_path = new_path = f'{path.rpartition(".")[0]}.{extension}'
if new_path == path:
if acodec == 'copy':
2022-06-06 18:19:57 +02:00
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)
return [], information
2022-06-06 18:19:57 +02:00
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)
information['filepath'] = new_path
information['ext'] = extension
# Try to update the date time for extracted audio file.
if information.get('filetime') is not None:
self.try_utime(
2022-06-06 18:19:57 +02:00
new_path, time.time(), information['filetime'], errnote='Cannot update utime of audio file')
return [orig_path], information
class FFmpegVideoConvertorPP(FFmpegPostProcessor):
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().__init__(downloader)
self.mapping = preferedformat
@staticmethod
def _options(target_ext):
yield from FFmpegPostProcessor.stream_copy_opts(False)
if target_ext == 'avi':
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, _skip_msg = resolve_mapping(source_ext, self.mapping)
if _skip_msg:
2022-01-19 23:57:36 +01:00
self.to_screen(f'Not {self._ACTION} media file "{filename}"; {_skip_msg}')
return [], info
outpath = replace_extension(filename, target_ext, source_ext)
self.to_screen(f'{self._ACTION.title()} video from {source_ext} to {target_ext}; Destination: {outpath}')
self.run_ffmpeg(filename, outpath, self._options(target_ext))
info['filepath'] = outpath
info['format'] = info['ext'] = target_ext
return [filename], info
class FFmpegVideoRemuxerPP(FFmpegVideoConvertorPP):
_ACTION = 'remuxing'
@staticmethod
def _options(target_ext):
return FFmpegPostProcessor.stream_copy_opts()
class FFmpegEmbedSubtitlePP(FFmpegPostProcessor):
SUPPORTED_EXTS = ('mp4', 'mov', 'm4a', 'webm', 'mkv', 'mka')
def __init__(self, downloader=None, already_have_subtitle=False):
super().__init__(downloader)
self._already_have_subtitle = already_have_subtitle
@PostProcessor._restrict_to(images=False)
def run(self, info):
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:
self.to_screen('There aren\'t any subtitles to embed')
return [], info
filename = info['filepath']
# 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 info.get('duration') and not info.get('__real_download') and self._duration_mismatch(
self._get_real_video_duration(filename, False), info['duration']):
self.to_screen(f'Skipping {self.pp_key()} since the real and expected durations mismatch')
return [], info
'''
ext = info['ext']
sub_langs, sub_names, sub_filenames = [], [], []
webm_vtt_warn = False
mp4_ass_warn = False
for lang, sub_info in subtitles.items():
if not os.path.exists(sub_info.get('filepath', '')):
self.report_warning(f'Skipping embedding {lang} subtitle because the file is missing')
continue
sub_ext = sub_info['ext']
if sub_ext == 'json':
self.report_warning('JSON subtitles cannot be embedded')
elif ext != 'webm' or ext == 'webm' and sub_ext == 'vtt':
sub_langs.append(lang)
sub_names.append(sub_info.get('name'))
sub_filenames.append(sub_info['filepath'])
else:
if not webm_vtt_warn and ext == 'webm' and sub_ext != 'vtt':
webm_vtt_warn = True
self.report_warning('Only WebVTT subtitles can be embedded in webm files')
if not mp4_ass_warn and ext == 'mp4' and sub_ext == 'ass':
mp4_ass_warn = True
self.report_warning('ASS subtitles cannot be properly embedded in mp4 files; expect issues')
if not sub_langs:
return [], info
input_files = [filename] + sub_filenames
opts = [
*self.stream_copy_opts(ext=info['ext']),
# Don't copy the existing subtitles, we may be running the
# postprocessor a second time
'-map', '-0:s',
]
for i, (lang, name) in enumerate(zip(sub_langs, sub_names)):
opts.extend(['-map', '%d:0' % (i + 1)])
lang_code = ISO639Utils.short2long(lang) or lang
opts.extend(['-metadata:s:s:%d' % i, 'language=%s' % lang_code])
if name:
opts.extend(['-metadata:s:s:%d' % i, 'handler_name=%s' % name,
'-metadata:s:s:%d' % i, 'title=%s' % name])
temp_filename = prepend_extension(filename, 'temp')
self.to_screen('Embedding subtitles in "%s"' % filename)
self.run_ffmpeg_multiple_files(input_files, temp_filename, opts)
os.replace(temp_filename, filename)
files_to_delete = [] if self._already_have_subtitle else sub_filenames
return files_to_delete, info
class FFmpegMetadataPP(FFmpegPostProcessor):
def __init__(self, downloader, add_metadata=True, add_chapters=True, add_infojson='if_exists'):
FFmpegPostProcessor.__init__(self, downloader)
self._add_metadata = add_metadata
self._add_chapters = add_chapters
self._add_infojson = add_infojson
@staticmethod
def _options(target_ext):
audio_only = target_ext == 'm4a'
yield from FFmpegPostProcessor.stream_copy_opts(not audio_only)
if audio_only:
yield from ('-vn', '-acodec', 'copy')
@PostProcessor._restrict_to(images=False)
def run(self, info):
filename, metadata_filename = info['filepath'], None
files_to_delete, options = [], []
if self._add_chapters and info.get('chapters'):
metadata_filename = replace_extension(filename, 'meta')
options.extend(self._get_chapter_opts(info['chapters'], metadata_filename))
files_to_delete.append(metadata_filename)
if self._add_metadata:
options.extend(self._get_metadata_opts(info))
if self._add_infojson:
if info['ext'] in ('mkv', 'mka'):
infojson_filename = info.get('infojson_filename')
options.extend(self._get_infojson_opts(info, infojson_filename))
if not infojson_filename:
files_to_delete.append(info.get('infojson_filename'))
elif self._add_infojson is True:
self.to_screen('The info-json can only be attached to mkv/mka files')
if not options:
self.to_screen('There isn\'t any metadata to add')
return [], info
temp_filename = prepend_extension(filename, 'temp')
self.to_screen('Adding metadata to "%s"' % filename)
self.run_ffmpeg_multiple_files(
(filename, metadata_filename), temp_filename,
itertools.chain(self._options(info['ext']), *options))
2022-05-01 01:28:26 +02:00
self._delete_downloaded_files(*files_to_delete)
os.replace(temp_filename, filename)
return [], info
@staticmethod
def _get_chapter_opts(chapters, metadata_filename):
with open(metadata_filename, 'wt', encoding='utf-8') as f:
def ffmpeg_escape(text):
return re.sub(r'([\\=;#\n])', r'\\\1', text)
metadata_file_content = ';FFMETADATA1\n'
for chapter in chapters:
metadata_file_content += '[CHAPTER]\nTIMEBASE=1/1000\n'
metadata_file_content += 'START=%d\n' % (chapter['start_time'] * 1000)
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)
f.write(metadata_file_content)
yield ('-map_metadata', '1')
def _get_metadata_opts(self, info):
meta_prefix = 'meta'
metadata = collections.defaultdict(dict)
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))
if info.get(key) is not None), None)
if value not in ('', None):
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)})
2022-07-08 21:37:47 +02:00
# 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('album')
add('album_artist')
add('disc', 'disc_number')
add('show', 'series')
add('season_number')
add('episode_id', ('episode', 'episode_id'))
add('episode_sort', 'episode_number')
2021-11-14 20:55:47 +01:00
if 'embed-metadata' in self.get_param('compat_opts', []):
add('comment', 'description')
metadata['common'].pop('synopsis', None)
meta_regex = rf'{re.escape(meta_prefix)}(?P<i>\d+)?_(?P<key>.+)'
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.replace('\0', '')
2022-04-07 12:16:53 +02:00
# 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}')
2014-02-22 18:23:30 +01:00
stream_idx = 0
for fmt in info.get('requested_formats') or []:
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):
if lang:
metadata[str(i)].setdefault('language', lang)
for name, value in metadata[str(i)].items():
yield (f'-metadata:s:{i}', f'{name}={value}')
stream_idx += stream_count
def _get_infojson_opts(self, info, infofn):
if not infofn or not os.path.exists(infofn):
if self._add_infojson is not True:
return
infofn = infofn or '%s.temp' % (
self._downloader.prepare_filename(info, 'infojson')
or replace_extension(self._downloader.prepare_filename(info), 'info.json', info['ext']))
if not self._downloader._ensure_dir_exists(infofn):
return
self.write_debug(f'Writing info-json to: {infofn}')
write_json_file(self._downloader.sanitize_info(info, self.get_param('clean_infojson', True)), infofn)
info['infojson_filename'] = infofn
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)
new_stream -= 1
yield (
'-attach', 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']
temp_filename = prepend_extension(filename, 'temp')
args = ['-c', 'copy']
audio_streams = 0
for (i, fmt) in enumerate(info['requested_formats']):
if fmt.get('acodec') != 'none':
args.extend(['-map', f'{i}:a:0'])
aac_fixup = fmt['protocol'].startswith('m3u8') and self.get_audio_codec(fmt['filepath']) == 'aac'
if aac_fixup:
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)
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
def can_merge(self):
# TODO: figure out merge-capable ffmpeg version
if self.basename != 'avconv':
return True
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, '
'yt-dlp will download single file media. '
'Update %s to version %s or newer to fix this.') % (
self.basename, self.basename, required_version)
self.report_warning(warning)
return False
return True
2014-02-22 13:55:51 +01:00
2021-06-20 00:19:23 +02:00
class FFmpegFixupPostProcessor(FFmpegPostProcessor):
def _fixup(self, msg, filename, options):
temp_filename = prepend_extension(filename, 'temp')
self.to_screen(f'{msg} of "{filename}"')
self.run_ffmpeg(filename, temp_filename, options)
os.replace(temp_filename, filename)
2021-06-20 00:19:23 +02:00
class FFmpegFixupStretchedPP(FFmpegFixupPostProcessor):
@PostProcessor._restrict_to(images=False, audio=False)
def run(self, info):
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])
return [], info
2021-06-20 00:19:23 +02:00
class FFmpegFixupM4aPP(FFmpegFixupPostProcessor):
@PostProcessor._restrict_to(images=False, video=False)
def run(self, info):
2021-06-20 00:19:23 +02:00
if info.get('container') == 'm4a_dash':
self._fixup('Correcting container', info['filepath'], [*self.stream_copy_opts(), '-f', 'mp4'])
return [], info
2021-06-20 00:19:23 +02:00
class FFmpegFixupM3u8PP(FFmpegFixupPostProcessor):
def _needs_fixup(self, info):
yield info['ext'] in ('mp4', 'm4a')
yield info['protocol'].startswith('m3u8')
try:
metadata = self.get_metadata_object(info['filepath'])
except PostProcessingError as e:
self.report_warning(f'Unable to extract metadata: {e.msg}')
yield True
else:
yield traverse_obj(metadata, ('format', 'format_name'), casesense=False) == 'mpegts'
@PostProcessor._restrict_to(images=False)
def run(self, info):
if all(self._needs_fixup(info)):
self._fixup('Fixing MPEG-TS in MP4 container', info['filepath'], [
*self.stream_copy_opts(), '-f', 'mp4', '-bsf:a', 'aac_adtstoasc'])
return [], info
class FFmpegFixupTimestampPP(FFmpegFixupPostProcessor):
def __init__(self, downloader=None, trim=0.001):
# "trim" should be used when the video contains unintended packets
super().__init__(downloader)
assert isinstance(trim, (int, float))
self.trim = str(trim)
@PostProcessor._restrict_to(images=False)
def run(self, info):
if not self._features.get('setts'):
self.report_warning(
'A re-encode is needed to fix timestamps in older versions of ffmpeg. '
'Please install ffmpeg 4.4 or later to fixup without re-encoding')
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])
return [], info
2022-01-19 23:57:36 +01:00
class FFmpegCopyStreamPP(FFmpegFixupPostProcessor):
MESSAGE = 'Copying stream'
@PostProcessor._restrict_to(images=False)
def run(self, info):
self._fixup(self.MESSAGE, info['filepath'], self.stream_copy_opts())
return [], info
2022-01-19 23:57:36 +01:00
class FFmpegFixupDurationPP(FFmpegCopyStreamPP):
MESSAGE = 'Fixing video duration'
2022-01-19 23:57:36 +01:00
class FFmpegFixupDuplicateMoovPP(FFmpegCopyStreamPP):
MESSAGE = 'Fixing duplicate MOOV atoms'
class FFmpegSubtitlesConvertorPP(FFmpegPostProcessor):
SUPPORTED_EXTS = MEDIA_EXTENSIONS.subtitles
2021-05-22 10:24:12 +02:00
def __init__(self, downloader=None, format=None):
super().__init__(downloader)
self.format = format
def run(self, info):
subs = info.get('requested_subtitles')
new_ext = self.format
new_format = new_ext
if new_format == 'vtt':
new_format = 'webvtt'
if subs is None:
self.to_screen('There aren\'t any subtitles to convert')
return [], info
self.to_screen('Converting subtitles')
sub_filenames = []
for lang, sub in subs.items():
if not os.path.exists(sub.get('filepath', '')):
self.report_warning(f'Skipping embedding {lang} subtitle because the file is missing')
continue
ext = sub['ext']
if ext == new_ext:
self.to_screen('Subtitle file for %s is already in the requested format' % new_ext)
continue
elif ext == 'json':
self.to_screen(
'You have requested to convert json subtitles into another format, '
'which is currently not possible')
continue
old_file = sub['filepath']
sub_filenames.append(old_file)
new_file = replace_extension(old_file, new_ext)
2017-04-12 21:38:43 +02:00
if ext in ('dfxp', 'ttml', 'tt'):
self.report_warning(
'You have requested to convert dfxp (TTML) subtitles into another format, '
'which results in style information loss')
dfxp_file = old_file
srt_file = replace_extension(old_file, 'srt')
with open(dfxp_file, 'rb') as f:
srt_data = dfxp2srt(f.read())
with open(srt_file, 'wt', encoding='utf-8') as f:
f.write(srt_data)
old_file = srt_file
subs[lang] = {
'ext': 'srt',
'data': srt_data,
'filepath': srt_file,
}
if new_ext == 'srt':
continue
else:
sub_filenames.append(srt_file)
self.run_ffmpeg(old_file, new_file, ['-f', new_format])
with open(new_file, encoding='utf-8') as f:
subs[lang] = {
'ext': new_ext,
'data': f.read(),
'filepath': new_file,
}
info['__files_to_move'][new_file] = replace_extension(
2021-08-13 17:10:13 +02:00
info['__files_to_move'][sub['filepath']], new_ext)
return sub_filenames, info
class FFmpegSplitChaptersPP(FFmpegPostProcessor):
def __init__(self, downloader, force_keyframes=False):
FFmpegPostProcessor.__init__(self, downloader)
self._force_keyframes = force_keyframes
def _prepare_filename(self, number, chapter, info):
info = info.copy()
info.update({
'section_number': number,
'section_title': chapter.get('title'),
'section_start': chapter.get('start_time'),
'section_end': chapter.get('end_time'),
})
return self._downloader.prepare_filename(info, 'chapter')
def _ffmpeg_args_for_chapter(self, number, chapter, info):
destination = self._prepare_filename(number, chapter, info)
if not self._downloader._ensure_dir_exists(encodeFilename(destination)):
return
chapter['filepath'] = destination
self.to_screen('Chapter %03d; Destination: %s' % (number, destination))
return (
destination,
['-ss', str(chapter['start_time']),
'-t', str(chapter['end_time'] - chapter['start_time'])])
@PostProcessor._restrict_to(images=False)
def run(self, info):
chapters = info.get('chapters') or []
if not chapters:
self.to_screen('Chapter information is unavailable')
return [], info
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))
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']:
2022-05-01 01:28:26 +02:00
self._delete_downloaded_files(in_file, msg=None)
return [], info
class FFmpegThumbnailsConvertorPP(FFmpegPostProcessor):
SUPPORTED_EXTS = MEDIA_EXTENSIONS.thumbnails
FORMAT_RE = create_mapping_re(SUPPORTED_EXTS)
2021-05-22 10:24:12 +02:00
def __init__(self, downloader=None, format=None):
super().__init__(downloader)
self.mapping = format
@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:
if thumbnail_ext.lower() != '.webp' and imghdr.what(thumbnail_filename) == 'webp':
self.to_screen('Correcting thumbnail "%s" extension to webp' % thumbnail_filename)
webp_filename = replace_extension(thumbnail_filename, 'webp')
os.replace(thumbnail_filename, webp_filename)
info['thumbnails'][idx]['filepath'] = webp_filename
info['__files_to_move'][webp_filename] = replace_extension(
info['__files_to_move'].pop(thumbnail_filename), 'webp')
2021-05-22 10:24:12 +02:00
@staticmethod
def _options(target_ext):
yield from ('-update', '1')
2021-05-22 10:24:12 +02:00
if target_ext == 'jpg':
yield from ('-bsf:v', 'mjpeg2jpeg')
2021-05-22 10:24:12 +02:00
def convert_thumbnail(self, thumbnail_filename, target_ext):
thumbnail_conv_filename = replace_extension(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, [] 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):
files_to_delete = []
has_thumbnail = False
for idx, thumbnail_dict in enumerate(info.get('thumbnails') or []):
original_thumbnail = thumbnail_dict.get('filepath')
if not original_thumbnail:
continue
has_thumbnail = True
self.fixup_webp(info, idx)
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'
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, target_ext)
files_to_delete.append(original_thumbnail)
info['__files_to_move'][thumbnail_dict['filepath']] = replace_extension(
info['__files_to_move'][original_thumbnail], target_ext)
if not has_thumbnail:
self.to_screen('There aren\'t any thumbnails to convert')
return files_to_delete, info
class FFmpegConcatPP(FFmpegPostProcessor):
def __init__(self, downloader, only_multi_video=False):
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
if len(in_files) == 1:
2022-01-19 23:57:36 +01:00
if os.path.realpath(in_files[0]) != os.path.realpath(out_file):
self.to_screen(f'Moving "{in_files[0]}" to "{out_file}"')
os.replace(in_files[0], out_file)
2022-01-19 23:57:36 +01:00
return []
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')
2022-01-19 23:57:36 +01:00
self.to_screen(f'Concatenating {len(in_files)} files; Destination: {out_file}')
super().concat_files(in_files, out_file)
2022-01-19 23:57:36 +01:00
return in_files
2022-02-18 18:46:16 +01:00
@PostProcessor._restrict_to(images=False, simulated=False)
def run(self, info):
entries = info.get('entries') or []
2022-02-18 18:46:16 +01:00
if not any(entries) or (self._only_multi_video and info['_type'] != 'multi_video'):
return [], info
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')
2022-02-18 18:46:16 +01:00
in_files = traverse_obj(entries, (..., 'requested_downloads', 0, 'filepath')) or []
if len(in_files) < len(entries):
raise PostProcessingError('Aborting concatenation because some downloads failed')
exts = traverse_obj(entries, (..., 'requested_downloads', 0, 'ext'), (..., 'ext'))
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')
2022-01-19 23:57:36 +01:00
files_to_delete = self.concat_files(in_files, out_file)
info['requested_downloads'] = [{
'filepath': out_file,
'ext': ie_copy['ext'],
}]
2022-01-19 23:57:36 +01:00
return files_to_delete, info