from __future__ import unicode_literals
import io
+import itertools
import os
import subprocess
import time
import re
import json
-
from .common import AudioConversionError, PostProcessor
-from ..compat import compat_str
+from ..compat import compat_str, compat_numeric_types
from ..utils import (
+ dfxp2srt,
encodeArgument,
encodeFilename,
+ float_or_none,
get_exe_version,
is_outdated_version,
+ ISO639Utils,
+ orderedSet,
PostProcessingError,
prepend_extension,
- shell_quote,
- subtitles_filename,
- dfxp2srt,
- ISO639Utils,
process_communicate_or_kill,
replace_extension,
- traverse_dict,
+ shell_quote,
+ traverse_obj,
+ variadic,
)
def check_version(self):
if not self.available:
- raise FFmpegPostProcessorError('ffmpeg not found. Please install')
+ 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(
'Continuing without ffmpeg.' % (location))
self._versions = {}
return
- elif not os.path.isdir(location):
+ elif os.path.isdir(location):
+ dirname, basename = location, None
+ else:
basename = os.path.splitext(os.path.basename(location))[0]
- if basename not in programs:
- self.report_warning(
- 'Cannot identify executable %s, its basename should be one of %s. '
- 'Continuing without ffmpeg.' %
- (location, ', '.join(programs)))
- self._versions = {}
- return None
- location = os.path.dirname(os.path.abspath(location))
+ 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(location, p)) for p in programs)
+ (p, os.path.join(dirname, p)) for p in programs)
+ if basename:
+ self._paths[basename] = location
self._versions = dict(
(p, get_ffmpeg_version(self._paths[p])) for p in programs)
if self._versions is None:
def get_audio_codec(self, path):
if not self.probe_available and not self.available:
- raise PostProcessingError('ffprobe and ffmpeg not found. Please install')
+ raise PostProcessingError('ffprobe and ffmpeg not found. Please install or provide the path using --ffmpeg-location')
try:
if self.probe_available:
cmd = [
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.')
+ raise PostProcessingError('ffprobe not found. Please install or provide the path using --ffmpeg-location')
self.check_version()
cmd = [
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_dict(stream, keys, casesense=False) == value),
+ (i for i, stream in enumerate(streams) if traverse_obj(stream, keys, casesense=False) == value),
None)
return num, len(streams)
- def run_ffmpeg_multiple_files(self, input_paths, out_path, opts):
+ def _get_real_video_duration(self, info, fatal=True):
+ try:
+ if '_real_duration' not in info:
+ info['_real_duration'] = float_or_none(
+ traverse_obj(self.get_metadata_object(info['filepath']), ('format', 'duration')))
+ if not info['_real_duration']:
+ raise PostProcessingError('ffprobe returned empty duration')
+ except PostProcessingError as e:
+ if fatal:
+ raise PostProcessingError(f'Unable to determine video duration; {e}')
+ return info.setdefault('_real_duration', None)
+
+ def _duration_mismatch(self, d1, d2):
+ if not d1 or not d2:
+ return None
+ return abs(d1 - d2) > 1
+
+ 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)])
+ [(out_path, opts)], **kwargs)
- def real_run_ffmpeg(self, input_path_opts, output_path_opts):
+ 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)
+ os.stat(encodeFilename(path)).st_mtime for path, _ in input_path_opts if path)
- cmd = [encodeFilename(self.executable, True), encodeArgument('-y')]
+ cmd = [encodeFilename(self.executable, True), encodeArgument('-y'), encodeArgument('-probesize'), encodeArgument('max')]
# avconv does not have repeat option
if self.basename == 'ffmpeg':
cmd += [encodeArgument('-loglevel'), encodeArgument('repeat+info')]
+ [encodeFilename(self._ffmpeg_filename_argument(file), True)])
for arg_type, path_opts in (('i', input_path_opts), ('o', output_path_opts)):
- cmd += [arg for i, o in enumerate(path_opts)
- for arg in make_args(o[0], o[1], arg_type, i + 1)]
+ 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))
p = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE, stdin=subprocess.PIPE)
stdout, stderr = process_communicate_or_kill(p)
- if p.returncode != 0:
+ if p.returncode not in variadic(expected_retcodes):
stderr = stderr.decode('utf-8', 'replace').strip()
- if self.get_param('verbose', False):
- self.report_error(stderr)
+ self.write_debug(stderr)
raise FFmpegPostProcessorError(stderr.split('\n')[-1])
for out_path, _ in output_path_opts:
- self.try_utime(out_path, oldest_mtime, oldest_mtime)
+ if out_path:
+ self.try_utime(out_path, oldest_mtime, oldest_mtime)
return stderr.decode('utf-8', 'replace')
- def run_ffmpeg(self, path, out_path, opts):
- return self.run_ffmpeg_multiple_files([path], out_path, opts)
+ def run_ffmpeg(self, path, out_path, opts, **kwargs):
+ return self.run_ffmpeg_multiple_files([path], out_path, opts, **kwargs)
- def _ffmpeg_filename_argument(self, fn):
+ @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)
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, ['-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 = ['-c', 'copy']
+ if out_file.rpartition('.')[-1] in ('mp4', 'mov'):
+ # For some reason, '-c copy' is not enough to copy subtitles
+ out_flags.extend(['-c:s', 'mov_text', '-movflags', '+faststart'])
+
+ try:
+ self.real_run_ffmpeg(
+ [(concat_file, ['-hide_banner', '-nostdin', '-f', 'concat', '-safe', '0'])],
+ [(out_file, out_flags)])
+ finally:
+ os.remove(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_EXTENSIONS = ('wav', 'flac', 'm4a', 'aiff', 'mp3', 'ogg', 'mka', 'opus', 'wma')
+ COMMON_AUDIO_EXTS = ('wav', 'flac', 'm4a', 'aiff', 'mp3', 'ogg', 'mka', 'opus', 'wma')
+ SUPPORTED_EXTS = ('best', 'aac', 'flac', 'mp3', 'm4a', 'opus', 'vorbis', 'wav')
def __init__(self, downloader=None, preferredcodec=None, preferredquality=None, nopostoverwrites=False):
FFmpegPostProcessor.__init__(self, downloader)
- if preferredcodec is None:
- preferredcodec = 'best'
- self._preferredcodec = preferredcodec
+ self._preferredcodec = preferredcodec or 'best'
self._preferredquality = preferredquality
self._nopostoverwrites = nopostoverwrites
except FFmpegPostProcessorError as err:
raise AudioConversionError(err.msg)
+ @PostProcessor._restrict_to(images=False)
def run(self, information):
path = information['filepath']
orig_ext = information['ext']
- if self._preferredcodec == 'best' and orig_ext in self.COMMON_AUDIO_EXTENSIONS:
+ 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')
return [], information
return [path], information
-class FFmpegVideoRemuxerPP(FFmpegPostProcessor):
+class FFmpegVideoConvertorPP(FFmpegPostProcessor):
+ SUPPORTED_EXTS = ('mp4', 'mkv', 'flv', 'webm', 'mov', 'avi', 'mp3', 'mka', 'm4a', 'ogg', 'opus')
+ FORMAT_RE = re.compile(r'{0}(?:/{0})*$'.format(r'(?:\w+>)?(?:%s)' % '|'.join(SUPPORTED_EXTS)))
+ _ACTION = 'converting'
+
def __init__(self, downloader=None, preferedformat=None):
- super(FFmpegVideoRemuxerPP, self).__init__(downloader)
+ super(FFmpegVideoConvertorPP, self).__init__(downloader)
self._preferedformats = preferedformat.lower().split('/')
- def run(self, information):
- path = information['filepath']
- sourceext, targetext = information['ext'].lower(), None
+ def _target_ext(self, source_ext):
for pair in self._preferedformats:
kv = pair.split('>')
- if len(kv) == 1 or kv[0].strip() == sourceext:
- targetext = kv[-1].strip()
- break
+ if len(kv) == 1 or kv[0].strip() == source_ext:
+ return kv[-1].strip()
+ @staticmethod
+ def _options(target_ext):
+ if target_ext == 'avi':
+ return ['-c:v', 'libxvid', '-vtag', 'XVID']
+ return []
+
+ @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 = (
- 'could not find a mapping for %s' if not targetext
- else 'already is in target format %s' if sourceext == targetext
+ 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)
if _skip_msg:
- self.to_screen('Not remuxing media file %s; %s' % (path, _skip_msg % sourceext))
- return [], information
+ self.to_screen(f'Not {self._ACTION} media file {filename!r}; {_skip_msg}')
+ return [], info
- options = ['-c', 'copy', '-map', '0', '-dn']
- if targetext in ['mp4', 'm4a', 'mov']:
- options.extend(['-movflags', '+faststart'])
- prefix, sep, oldext = path.rpartition('.')
- outpath = prefix + sep + targetext
- self.to_screen('Remuxing video from %s to %s; Destination: %s' % (sourceext, targetext, outpath))
- self.run_ffmpeg(path, outpath, options)
- information['filepath'] = outpath
- information['format'] = targetext
- information['ext'] = targetext
- return [path], information
+ 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 FFmpegVideoConvertorPP(FFmpegPostProcessor):
- def __init__(self, downloader=None, preferedformat=None):
- super(FFmpegVideoConvertorPP, self).__init__(downloader)
- self._preferedformat = preferedformat
- def run(self, information):
- path = information['filepath']
- if information['ext'] == self._preferedformat:
- self.to_screen('Not converting video file %s - already is in target format %s' % (path, self._preferedformat))
- return [], information
- options = []
- if self._preferedformat == 'avi':
- options.extend(['-c:v', 'libxvid', '-vtag', 'XVID'])
- prefix, sep, ext = path.rpartition('.')
- outpath = prefix + sep + self._preferedformat
- self.to_screen('Converting video from %s to %s, Destination: ' % (information['ext'], self._preferedformat) + outpath)
- self.run_ffmpeg(path, outpath, options)
- information['filepath'] = outpath
- information['format'] = self._preferedformat
- information['ext'] = self._preferedformat
- return [path], information
+class FFmpegVideoRemuxerPP(FFmpegVideoConvertorPP):
+ _ACTION = 'remuxing'
+
+ @staticmethod
+ def _options(target_ext):
+ options = ['-c', 'copy', '-map', '0', '-dn']
+ if target_ext in ['mp4', 'm4a', 'mov']:
+ options.extend(['-movflags', '+faststart'])
+ return options
class FFmpegEmbedSubtitlePP(FFmpegPostProcessor):
super(FFmpegEmbedSubtitlePP, self).__init__(downloader)
self._already_have_subtitle = already_have_subtitle
+ @PostProcessor._restrict_to(images=False)
def run(self, information):
if information['ext'] not in ('mp4', 'webm', 'mkv'):
self.to_screen('Subtitles can only be embedded in mp4, webm or mkv files')
return [], information
filename = information['filepath']
+ if information.get('duration') and self._duration_mismatch(
+ self._get_real_video_duration(information, False), information['duration']):
+ self.to_screen(f'Skipping {self.pp_key()} since the real and expected durations mismatch')
+ return [], information
ext = information['ext']
- sub_langs = []
- sub_filenames = []
+ 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(information.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_filenames.append(subtitles_filename(filename, lang, sub_ext, ext))
+ 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
]
if information['ext'] == 'mp4':
opts += ['-c:s', 'mov_text']
- for (i, lang) in enumerate(sub_langs):
+ 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.remove(encodeFilename(filename))
- os.rename(encodeFilename(temp_filename), encodeFilename(filename))
+ os.replace(temp_filename, filename)
files_to_delete = [] if self._already_have_subtitle else sub_filenames
return files_to_delete, information
class FFmpegMetadataPP(FFmpegPostProcessor):
+
+ def __init__(self, downloader, add_metadata=True, add_chapters=True):
+ FFmpegPostProcessor.__init__(self, downloader)
+ self._add_metadata = add_metadata
+ self._add_chapters = add_chapters
+
+ @staticmethod
+ def _options(target_ext):
+ yield from ('-map', '0', '-dn')
+ if target_ext == 'm4a':
+ yield from ('-vn', '-acodec', 'copy')
+ else:
+ yield from ('-c', 'copy')
+
+ @PostProcessor._restrict_to(images=False)
def run(self, info):
+ filename, metadata_filename = info['filepath'], None
+ 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))
+ if self._add_metadata:
+ options.extend(self._get_metadata_opts(info))
+
+ 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))
+ if metadata_filename:
+ os.remove(metadata_filename)
+ 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:
+ 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):
metadata = {}
def add(meta_list, info_list=None):
- if not info_list:
- info_list = meta_list
- if not isinstance(meta_list, (list, tuple)):
- meta_list = (meta_list,)
- if not isinstance(info_list, (list, tuple)):
- info_list = (info_list,)
- for info_f in info_list:
- if info.get(info_f) is not None:
- for meta_f in meta_list:
+ if not meta_list:
+ return
+ for info_f in variadic(info_list or meta_list):
+ if isinstance(info.get(info_f), (compat_str, compat_numeric_types)):
+ for meta_f in variadic(meta_list):
metadata[meta_f] = info[info_f]
break
add('title', ('track', 'title'))
add('date', 'upload_date')
- add(('description', 'comment'), 'description')
- add('purl', 'webpage_url')
+ add(('description', 'synopsis'), 'description')
+ add(('purl', 'comment'), 'webpage_url')
add('track', 'track_number')
add('artist', ('artist', 'creator', 'uploader', 'uploader_id'))
add('genre')
add('episode_id', ('episode', 'episode_id'))
add('episode_sort', 'episode_number')
- if not metadata:
- self.to_screen('There isn\'t any metadata to add')
- return [], info
-
- filename = info['filepath']
- temp_filename = prepend_extension(filename, 'temp')
- in_filenames = [filename]
- options = ['-map', '0', '-dn']
-
- if info['ext'] == 'm4a':
- options.extend(['-vn', '-acodec', 'copy'])
- else:
- options.extend(['-c', 'copy'])
-
- for (name, value) in metadata.items():
- options.extend(['-metadata', '%s=%s' % (name, value)])
-
- chapters = info.get('chapters', [])
- if chapters:
- metadata_filename = replace_extension(filename, 'meta')
- with io.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)
- in_filenames.append(metadata_filename)
- options.extend(['-map_metadata', '1'])
-
- if '__infojson_filename' in info and info['ext'] in ('mkv', 'mka'):
- old_stream, new_stream = self.get_stream_number(
- filename, ('tags', 'mimetype'), 'application/json')
+ prefix = 'meta_'
+ for key in filter(lambda k: k.startswith(prefix), info.keys()):
+ add(key[len(prefix):], key)
+
+ for name, value in metadata.items():
+ yield ('-metadata', f'{name}={value}')
+
+ 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
+ if fmt.get('language'):
+ lang = ISO639Utils.short2long(fmt['language']) or fmt['language']
+ for i in range(stream_count):
+ yield ('-metadata:s:%d' % (stream_idx + i), 'language=%s' % lang)
+ stream_idx += stream_count
+
+ if ('no-attach-info-json' not in self.get_param('compat_opts', [])
+ and '__infojson_filename' in info and info['ext'] in ('mkv', 'mka')):
+ old_stream, new_stream = self.get_stream_number(info['filepath'], ('tags', 'mimetype'), 'application/json')
if old_stream is not None:
- options.extend(['-map', '-0:%d' % old_stream])
+ yield ('-map', '-0:%d' % old_stream)
new_stream -= 1
- options.extend([
- '-attach', info['__infojson_filename'],
- '-metadata:s:%d' % new_stream, 'mimetype=application/json'
- ])
-
- self.to_screen('Adding metadata to \'%s\'' % filename)
- self.run_ffmpeg_multiple_files(in_filenames, temp_filename, options)
- if chapters:
- os.remove(metadata_filename)
- os.remove(encodeFilename(filename))
- os.rename(encodeFilename(temp_filename), encodeFilename(filename))
- return [], info
+ yield ('-attach', info['__infojson_filename'],
+ '-metadata:s:%d' % new_stream, 'mimetype=application/json')
class FFmpegMergerPP(FFmpegPostProcessor):
+ @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', '%u:a:0' % (i)])
+ args.extend(['-map', f'{i}:a:0'])
+ if self.get_audio_codec(fmt['filepath']) == 'aac':
+ 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)
return True
-class FFmpegFixupStretchedPP(FFmpegPostProcessor):
- def run(self, info):
- stretched_ratio = info.get('stretched_ratio')
- if stretched_ratio is None or stretched_ratio == 1:
- return [], info
-
- filename = info['filepath']
+class FFmpegFixupPostProcessor(FFmpegPostProcessor):
+ def _fixup(self, msg, filename, options):
temp_filename = prepend_extension(filename, 'temp')
- options = ['-c', 'copy', '-map', '0', '-dn', '-aspect', '%f' % stretched_ratio]
- self.to_screen('Fixing aspect ratio in "%s"' % filename)
+ self.to_screen(f'{msg} of "{filename}"')
self.run_ffmpeg(filename, temp_filename, options)
- os.remove(encodeFilename(filename))
- os.rename(encodeFilename(temp_filename), encodeFilename(filename))
+ os.replace(temp_filename, filename)
+
+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'], [
+ '-c', 'copy', '-map', '0', '-dn', '-aspect', '%f' % stretched_ratio])
return [], info
-class FFmpegFixupM4aPP(FFmpegPostProcessor):
+class FFmpegFixupM4aPP(FFmpegFixupPostProcessor):
+ @PostProcessor._restrict_to(images=False, video=False)
def run(self, info):
- if info.get('container') != 'm4a_dash':
- return [], info
+ if info.get('container') == 'm4a_dash':
+ self._fixup('Correcting container', info['filepath'], [
+ '-c', 'copy', '-map', '0', '-dn', '-f', 'mp4'])
+ return [], info
- filename = info['filepath']
- temp_filename = prepend_extension(filename, 'temp')
- options = ['-c', 'copy', '-map', '0', '-dn', '-f', 'mp4']
- self.to_screen('Correcting container in "%s"' % filename)
- self.run_ffmpeg(filename, temp_filename, options)
+class FFmpegFixupM3u8PP(FFmpegFixupPostProcessor):
+ @PostProcessor._restrict_to(images=False)
+ def run(self, info):
+ if self.get_audio_codec(info['filepath']) == 'aac':
+ self._fixup('Fixing malformed AAC bitstream', info['filepath'], [
+ '-c', 'copy', '-map', '0', '-dn', '-f', 'mp4', '-bsf:a', 'aac_adtstoasc'])
+ return [], info
- os.remove(encodeFilename(filename))
- os.rename(encodeFilename(temp_filename), encodeFilename(filename))
- return [], info
+class FFmpegFixupTimestampPP(FFmpegFixupPostProcessor):
+ def __init__(self, downloader=None, trim=0.001):
+ # "trim" should be used when the video contains unintended packets
+ super(FFmpegFixupTimestampPP, self).__init__(downloader)
+ assert isinstance(trim, (int, float))
+ self.trim = str(trim)
-class FFmpegFixupM3u8PP(FFmpegPostProcessor):
+ @PostProcessor._restrict_to(images=False)
def run(self, info):
- filename = info['filepath']
- if self.get_audio_codec(filename) == 'aac':
- temp_filename = prepend_extension(filename, 'temp')
+ required_version = '4.4'
+ if is_outdated_version(self._versions[self.basename], required_version):
+ self.report_warning(
+ 'A re-encode is needed to fix timestamps in older versions of ffmpeg. '
+ f'Please install ffmpeg {required_version} 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 + ['-map', '0', '-dn', '-ss', self.trim])
+ return [], info
- options = ['-c', 'copy', '-map', '0', '-dn', '-f', 'mp4', '-bsf:a', 'aac_adtstoasc']
- self.to_screen('Fixing malformed AAC bitstream in "%s"' % filename)
- self.run_ffmpeg(filename, temp_filename, options)
- os.remove(encodeFilename(filename))
- os.rename(encodeFilename(temp_filename), encodeFilename(filename))
+class FFmpegFixupDurationPP(FFmpegFixupPostProcessor):
+ @PostProcessor._restrict_to(images=False)
+ def run(self, info):
+ self._fixup('Fixing video duration', info['filepath'], ['-c', 'copy', '-map', '0', '-dn'])
return [], info
class FFmpegSubtitlesConvertorPP(FFmpegPostProcessor):
+ SUPPORTED_EXTS = ('srt', 'vtt', 'ass', 'lrc')
+
def __init__(self, downloader=None, format=None):
super(FFmpegSubtitlesConvertorPP, self).__init__(downloader)
self.format = format
def run(self, info):
subs = info.get('requested_subtitles')
- filename = info['filepath']
new_ext = self.format
new_format = new_ext
if new_format == 'vtt':
'You have requested to convert json subtitles into another format, '
'which is currently not possible')
continue
- old_file = subtitles_filename(filename, lang, ext, info.get('ext'))
+ old_file = sub['filepath']
sub_filenames.append(old_file)
- new_file = subtitles_filename(filename, lang, new_ext, info.get('ext'))
+ new_file = replace_extension(old_file, new_ext)
if ext in ('dfxp', 'ttml', 'tt'):
self.report_warning(
'which results in style information loss')
dfxp_file = old_file
- srt_file = subtitles_filename(filename, lang, 'srt', info.get('ext'))
+ srt_file = replace_extension(old_file, 'srt')
with open(dfxp_file, 'rb') as f:
srt_data = dfxp2srt(f.read())
subs[lang] = {
'ext': 'srt',
- 'data': srt_data
+ 'data': srt_data,
+ 'filepath': srt_file,
}
if new_ext == 'srt':
subs[lang] = {
'ext': new_ext,
'data': f.read(),
+ 'filepath': new_file,
}
+ info['__files_to_move'][new_file] = replace_extension(
+ 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()
if not self._downloader._ensure_dir_exists(encodeFilename(destination)):
return
- chapter['_filename'] = destination
+ chapter['filepath'] = destination
self.to_screen('Chapter %03d; Destination: %s' % (number, destination))
return (
destination,
['-ss', compat_str(chapter['start_time']),
- '-to', compat_str(chapter['end_time'])])
+ '-t', compat_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.report_warning('There are no tracks to extract')
+ 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([(info['filepath'], opts)], [(destination, ['-c', 'copy'])])
+ self.real_run_ffmpeg([(in_file, opts)], [(destination, ['-c', 'copy'])])
+ if in_file != info['filepath']:
+ os.remove(in_file)
return [], info
+
+
+class FFmpegThumbnailsConvertorPP(FFmpegPostProcessor):
+ SUPPORTED_EXTS = ('jpg', 'png')
+
+ def __init__(self, downloader=None, format=None):
+ super(FFmpegThumbnailsConvertorPP, self).__init__(downloader)
+ self.format = 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'
+
+ 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)
+ 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')
+
+ @staticmethod
+ def _options(target_ext):
+ if target_ext == 'jpg':
+ return ['-bsf:v', 'mjpeg2jpeg']
+ return []
+
+ 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.real_run_ffmpeg(
+ [(thumbnail_filename, ['-f', 'image2', '-pattern_type', 'none'])],
+ [(thumbnail_conv_filename.replace('%', '%%'), 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['thumbnails']):
+ if 'filepath' not in thumbnail_dict:
+ continue
+ has_thumbnail = True
+ self.fixup_webp(info, idx)
+ original_thumbnail = thumbnail_dict['filepath']
+ _, thumbnail_ext = os.path.splitext(original_thumbnail)
+ if thumbnail_ext:
+ thumbnail_ext = thumbnail_ext[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)
+ continue
+ thumbnail_dict['filepath'] = self.convert_thumbnail(original_thumbnail, self.format)
+ files_to_delete.append(original_thumbnail)
+ info['__files_to_move'][thumbnail_dict['filepath']] = replace_extension(
+ info['__files_to_move'][original_thumbnail], self.format)
+
+ if not has_thumbnail:
+ self.to_screen('There aren\'t any thumbnails to convert')
+ return files_to_delete, info
projects / yt-dlp.git / blobdiff