import collections
+import contextvars
import itertools
import json
import os
import subprocess
import time
-from .common import AudioConversionError, PostProcessor
+from .common import PostProcessor
from ..compat import functools, imghdr
from ..utils import (
+ MEDIA_EXTENSIONS,
ISO639Utils,
Popen,
PostProcessingError,
'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)\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)
def _determine_executables(self):
programs = [*self._ffmpeg_to_avconv.keys(), *self._ffmpeg_to_avconv.values()]
- location = self.get_param('ffmpeg_location')
+ location = self.get_param('ffmpeg_location', self._ffmpeg_location.get())
if location is None:
return {p: p for p in programs}
encodeArgument('-i')]
cmd.append(encodeFilename(self._ffmpeg_filename_argument(path), True))
self.write_debug(f'{self.basename} command line: {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:
+ 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_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']
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()
+ _, 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)
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
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
+
+ # 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':
- self.to_screen(f'File is already in target format {self._preferredcodec}, skipping')
+ 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')
self.to_screen('Post-process file %s exists, skipping' % new_path)
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 = (*MEDIA_EXTENSIONS.common_video, *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._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()
+ self.mapping = preferedformat
@staticmethod
def _options(target_ext):
@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
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
'''
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')
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')
class FFmpegSubtitlesConvertorPP(FFmpegPostProcessor):
- SUPPORTED_EXTS = ('srt', 'vtt', 'ass', 'lrc')
+ SUPPORTED_EXTS = MEDIA_EXTENSIONS.subtitles
def __init__(self, downloader=None, format=None):
super().__init__(downloader)
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().__init__(downloader)
- self.format = format
+ self.mapping = format
@classmethod
def is_webp(cls, path):
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()
+ 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')
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)