yt_dlp/postprocessor/ffmpeg.py

   1 from __future__ import unicode_literals
   2
   3 import io
   4 import os
   5 import subprocess
   6 import time
   7 import re
   8 import json
   9
  10
  11 from .common import AudioConversionError, PostProcessor
  12
  13 from ..compat import compat_str, compat_numeric_types
  14 from ..utils import (
  15     encodeArgument,
  16     encodeFilename,
  17     get_exe_version,
  18     is_outdated_version,
  19     PostProcessingError,
  20     prepend_extension,
  21     shell_quote,
  22     dfxp2srt,
  23     ISO639Utils,
  24     process_communicate_or_kill,
  25     replace_extension,
  26     traverse_dict,
  27 )
  28
  29
  30 EXT_TO_OUT_FORMATS = {
  31     'aac': 'adts',
  32     'flac': 'flac',
  33     'm4a': 'ipod',
  34     'mka': 'matroska',
  35     'mkv': 'matroska',
  36     'mpg': 'mpeg',
  37     'ogv': 'ogg',
  38     'ts': 'mpegts',
  39     'wma': 'asf',
  40     'wmv': 'asf',
  41 }
  42 ACODECS = {
  43     'mp3': 'libmp3lame',
  44     'aac': 'aac',
  45     'flac': 'flac',
  46     'm4a': 'aac',
  47     'opus': 'libopus',
  48     'vorbis': 'libvorbis',
  49     'wav': None,
  50 }
  51
  52
  53 class FFmpegPostProcessorError(PostProcessingError):
  54     pass
  55
  56
  57 class FFmpegPostProcessor(PostProcessor):
  58     def __init__(self, downloader=None):
  59         PostProcessor.__init__(self, downloader)
  60         self._determine_executables()
  61
  62     def check_version(self):
  63         if not self.available:
  64             raise FFmpegPostProcessorError('ffmpeg not found. Please install or provide the path using --ffmpeg-location')
  65
  66         required_version = '10-0' if self.basename == 'avconv' else '1.0'
  67         if is_outdated_version(
  68                 self._versions[self.basename], required_version):
  69             warning = 'Your copy of %s is outdated, update %s to version %s or newer if you encounter any errors.' % (
  70                 self.basename, self.basename, required_version)
  71             self.report_warning(warning)
  72
  73     @staticmethod
  74     def get_versions(downloader=None):
  75         return FFmpegPostProcessor(downloader)._versions
  76
  77     def _determine_executables(self):
  78         programs = ['avprobe', 'avconv', 'ffmpeg', 'ffprobe']
  79         prefer_ffmpeg = True
  80
  81         def get_ffmpeg_version(path):
  82             ver = get_exe_version(path, args=['-version'])
  83             if ver:
  84                 regexs = [
  85                     r'(?:\d+:)?([0-9.]+)-[0-9]+ubuntu[0-9.]+$',  # Ubuntu, see [1]
  86                     r'n([0-9.]+)$',  # Arch Linux
  87                     # 1. http://www.ducea.com/2006/06/17/ubuntu-package-version-naming-explanation/
  88                 ]
  89                 for regex in regexs:
  90                     mobj = re.match(regex, ver)
  91                     if mobj:
  92                         ver = mobj.group(1)
  93             return ver
  94
  95         self.basename = None
  96         self.probe_basename = None
  97
  98         self._paths = None
  99         self._versions = None
 100         if self._downloader:
 101             prefer_ffmpeg = self.get_param('prefer_ffmpeg', True)
 102             location = self.get_param('ffmpeg_location')
 103             if location is not None:
 104                 if not os.path.exists(location):
 105                     self.report_warning(
 106                         'ffmpeg-location %s does not exist! '
 107                         'Continuing without ffmpeg.' % (location))
 108                     self._versions = {}
 109                     return
 110                 elif not os.path.isdir(location):
 111                     basename = os.path.splitext(os.path.basename(location))[0]
 112                     if basename not in programs:
 113                         self.report_warning(
 114                             'Cannot identify executable %s, its basename should be one of %s. '
 115                             'Continuing without ffmpeg.' %
 116                             (location, ', '.join(programs)))
 117                         self._versions = {}
 118                         return None
 119                     location = os.path.dirname(os.path.abspath(location))
 120                     if basename in ('ffmpeg', 'ffprobe'):
 121                         prefer_ffmpeg = True
 122
 123                 self._paths = dict(
 124                     (p, os.path.join(location, p)) for p in programs)
 125                 self._versions = dict(
 126                     (p, get_ffmpeg_version(self._paths[p])) for p in programs)
 127         if self._versions is None:
 128             self._versions = dict(
 129                 (p, get_ffmpeg_version(p)) for p in programs)
 130             self._paths = dict((p, p) for p in programs)
 131
 132         if prefer_ffmpeg is False:
 133             prefs = ('avconv', 'ffmpeg')
 134         else:
 135             prefs = ('ffmpeg', 'avconv')
 136         for p in prefs:
 137             if self._versions[p]:
 138                 self.basename = p
 139                 break
 140
 141         if prefer_ffmpeg is False:
 142             prefs = ('avprobe', 'ffprobe')
 143         else:
 144             prefs = ('ffprobe', 'avprobe')
 145         for p in prefs:
 146             if self._versions[p]:
 147                 self.probe_basename = p
 148                 break
 149
 150     @property
 151     def available(self):
 152         return self.basename is not None
 153
 154     @property
 155     def executable(self):
 156         return self._paths[self.basename]
 157
 158     @property
 159     def probe_available(self):
 160         return self.probe_basename is not None
 161
 162     @property
 163     def probe_executable(self):
 164         return self._paths[self.probe_basename]
 165
 166     def get_audio_codec(self, path):
 167         if not self.probe_available and not self.available:
 168             raise PostProcessingError('ffprobe and ffmpeg not found. Please install or provide the path using --ffmpeg-location')
 169         try:
 170             if self.probe_available:
 171                 cmd = [
 172                     encodeFilename(self.probe_executable, True),
 173                     encodeArgument('-show_streams')]
 174             else:
 175                 cmd = [
 176                     encodeFilename(self.executable, True),
 177                     encodeArgument('-i')]
 178             cmd.append(encodeFilename(self._ffmpeg_filename_argument(path), True))
 179             self.write_debug('%s command line: %s' % (self.basename, shell_quote(cmd)))
 180             handle = subprocess.Popen(
 181                 cmd, stderr=subprocess.PIPE,
 182                 stdout=subprocess.PIPE, stdin=subprocess.PIPE)
 183             stdout_data, stderr_data = process_communicate_or_kill(handle)
 184             expected_ret = 0 if self.probe_available else 1
 185             if handle.wait() != expected_ret:
 186                 return None
 187         except (IOError, OSError):
 188             return None
 189         output = (stdout_data if self.probe_available else stderr_data).decode('ascii', 'ignore')
 190         if self.probe_available:
 191             audio_codec = None
 192             for line in output.split('\n'):
 193                 if line.startswith('codec_name='):
 194                     audio_codec = line.split('=')[1].strip()
 195                 elif line.strip() == 'codec_type=audio' and audio_codec is not None:
 196                     return audio_codec
 197         else:
 198             # Stream #FILE_INDEX:STREAM_INDEX[STREAM_ID](LANGUAGE): CODEC_TYPE: CODEC_NAME
 199             mobj = re.search(
 200                 r'Stream\s*#\d+:\d+(?:\[0x[0-9a-f]+\])?(?:\([a-z]{3}\))?:\s*Audio:\s*([0-9a-z]+)',
 201                 output)
 202             if mobj:
 203                 return mobj.group(1)
 204         return None
 205
 206     def get_metadata_object(self, path, opts=[]):
 207         if self.probe_basename != 'ffprobe':
 208             if self.probe_available:
 209                 self.report_warning('Only ffprobe is supported for metadata extraction')
 210             raise PostProcessingError('ffprobe not found. Please install or provide the path using --ffmpeg-location')
 211         self.check_version()
 212
 213         cmd = [
 214             encodeFilename(self.probe_executable, True),
 215             encodeArgument('-hide_banner'),
 216             encodeArgument('-show_format'),
 217             encodeArgument('-show_streams'),
 218             encodeArgument('-print_format'),
 219             encodeArgument('json'),
 220         ]
 221
 222         cmd += opts
 223         cmd.append(encodeFilename(self._ffmpeg_filename_argument(path), True))
 224         self.write_debug('ffprobe command line: %s' % shell_quote(cmd))
 225         p = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE, stdin=subprocess.PIPE)
 226         stdout, stderr = p.communicate()
 227         return json.loads(stdout.decode('utf-8', 'replace'))
 228
 229     def get_stream_number(self, path, keys, value):
 230         streams = self.get_metadata_object(path)['streams']
 231         num = next(
 232             (i for i, stream in enumerate(streams) if traverse_dict(stream, keys, casesense=False) == value),
 233             None)
 234         return num, len(streams)
 235
 236     def run_ffmpeg_multiple_files(self, input_paths, out_path, opts):
 237         return self.real_run_ffmpeg(
 238             [(path, []) for path in input_paths],
 239             [(out_path, opts)])
 240
 241     def real_run_ffmpeg(self, input_path_opts, output_path_opts):
 242         self.check_version()
 243
 244         oldest_mtime = min(
 245             os.stat(encodeFilename(path)).st_mtime for path, _ in input_path_opts)
 246
 247         cmd = [encodeFilename(self.executable, True), encodeArgument('-y')]
 248         # avconv does not have repeat option
 249         if self.basename == 'ffmpeg':
 250             cmd += [encodeArgument('-loglevel'), encodeArgument('repeat+info')]
 251
 252         def make_args(file, args, name, number):
 253             keys = ['_%s%d' % (name, number), '_%s' % name]
 254             if name == 'o' and number == 1:
 255                 keys.append('')
 256             args += self._configuration_args(self.basename, keys)
 257             if name == 'i':
 258                 args.append('-i')
 259             return (
 260                 [encodeArgument(arg) for arg in args]
 261                 + [encodeFilename(self._ffmpeg_filename_argument(file), True)])
 262
 263         for arg_type, path_opts in (('i', input_path_opts), ('o', output_path_opts)):
 264             cmd += [arg for i, o in enumerate(path_opts)
 265                     for arg in make_args(o[0], o[1], arg_type, i + 1)]
 266
 267         self.write_debug('ffmpeg command line: %s' % shell_quote(cmd))
 268         p = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE, stdin=subprocess.PIPE)
 269         stdout, stderr = process_communicate_or_kill(p)
 270         if p.returncode != 0:
 271             stderr = stderr.decode('utf-8', 'replace').strip()
 272             if self.get_param('verbose', False):
 273                 self.report_error(stderr)
 274             raise FFmpegPostProcessorError(stderr.split('\n')[-1])
 275         for out_path, _ in output_path_opts:
 276             self.try_utime(out_path, oldest_mtime, oldest_mtime)
 277         return stderr.decode('utf-8', 'replace')
 278
 279     def run_ffmpeg(self, path, out_path, opts):
 280         return self.run_ffmpeg_multiple_files([path], out_path, opts)
 281
 282     def _ffmpeg_filename_argument(self, fn):
 283         # Always use 'file:' because the filename may contain ':' (ffmpeg
 284         # interprets that as a protocol) or can start with '-' (-- is broken in
 285         # ffmpeg, see https://ffmpeg.org/trac/ffmpeg/ticket/2127 for details)
 286         # Also leave '-' intact in order not to break streaming to stdout.
 287         if fn.startswith(('http://', 'https://')):
 288             return fn
 289         return 'file:' + fn if fn != '-' else fn
 290
 291
 292 class FFmpegExtractAudioPP(FFmpegPostProcessor):
 293     COMMON_AUDIO_EXTS = ('wav', 'flac', 'm4a', 'aiff', 'mp3', 'ogg', 'mka', 'opus', 'wma')
 294     SUPPORTED_EXTS = ('best', 'aac', 'flac', 'mp3', 'm4a', 'opus', 'vorbis', 'wav')
 295
 296     def __init__(self, downloader=None, preferredcodec=None, preferredquality=None, nopostoverwrites=False):
 297         FFmpegPostProcessor.__init__(self, downloader)
 298         self._preferredcodec = preferredcodec or 'best'
 299         self._preferredquality = preferredquality
 300         self._nopostoverwrites = nopostoverwrites
 301
 302     def run_ffmpeg(self, path, out_path, codec, more_opts):
 303         if codec is None:
 304             acodec_opts = []
 305         else:
 306             acodec_opts = ['-acodec', codec]
 307         opts = ['-vn'] + acodec_opts + more_opts
 308         try:
 309             FFmpegPostProcessor.run_ffmpeg(self, path, out_path, opts)
 310         except FFmpegPostProcessorError as err:
 311             raise AudioConversionError(err.msg)
 312
 313     def run(self, information):
 314         path = information['filepath']
 315         orig_ext = information['ext']
 316
 317         if self._preferredcodec == 'best' and orig_ext in self.COMMON_AUDIO_EXTS:
 318             self.to_screen('Skipping audio extraction since the file is already in a common audio format')
 319             return [], information
 320
 321         filecodec = self.get_audio_codec(path)
 322         if filecodec is None:
 323             raise PostProcessingError('WARNING: unable to obtain file audio codec with ffprobe')
 324
 325         more_opts = []
 326         if self._preferredcodec == 'best' or self._preferredcodec == filecodec or (self._preferredcodec == 'm4a' and filecodec == 'aac'):
 327             if filecodec == 'aac' and self._preferredcodec in ['m4a', 'best']:
 328                 # Lossless, but in another container
 329                 acodec = 'copy'
 330                 extension = 'm4a'
 331                 more_opts = ['-bsf:a', 'aac_adtstoasc']
 332             elif filecodec in ['aac', 'flac', 'mp3', 'vorbis', 'opus']:
 333                 # Lossless if possible
 334                 acodec = 'copy'
 335                 extension = filecodec
 336                 if filecodec == 'aac':
 337                     more_opts = ['-f', 'adts']
 338                 if filecodec == 'vorbis':
 339                     extension = 'ogg'
 340             else:
 341                 # MP3 otherwise.
 342                 acodec = 'libmp3lame'
 343                 extension = 'mp3'
 344                 more_opts = []
 345                 if self._preferredquality is not None:
 346                     if int(self._preferredquality) < 10:
 347                         more_opts += ['-q:a', self._preferredquality]
 348                     else:
 349                         more_opts += ['-b:a', self._preferredquality + 'k']
 350         else:
 351             # We convert the audio (lossy if codec is lossy)
 352             acodec = ACODECS[self._preferredcodec]
 353             extension = self._preferredcodec
 354             more_opts = []
 355             if self._preferredquality is not None:
 356                 # The opus codec doesn't support the -aq option
 357                 if int(self._preferredquality) < 10 and extension != 'opus':
 358                     more_opts += ['-q:a', self._preferredquality]
 359                 else:
 360                     more_opts += ['-b:a', self._preferredquality + 'k']
 361             if self._preferredcodec == 'aac':
 362                 more_opts += ['-f', 'adts']
 363             if self._preferredcodec == 'm4a':
 364                 more_opts += ['-bsf:a', 'aac_adtstoasc']
 365             if self._preferredcodec == 'vorbis':
 366                 extension = 'ogg'
 367             if self._preferredcodec == 'wav':
 368                 extension = 'wav'
 369                 more_opts += ['-f', 'wav']
 370
 371         prefix, sep, ext = path.rpartition('.')  # not os.path.splitext, since the latter does not work on unicode in all setups
 372         new_path = prefix + sep + extension
 373
 374         information['filepath'] = new_path
 375         information['ext'] = extension
 376
 377         # If we download foo.mp3 and convert it to... foo.mp3, then don't delete foo.mp3, silly.
 378         if (new_path == path
 379                 or (self._nopostoverwrites and os.path.exists(encodeFilename(new_path)))):
 380             self.to_screen('Post-process file %s exists, skipping' % new_path)
 381             return [], information
 382
 383         try:
 384             self.to_screen('Destination: ' + new_path)
 385             self.run_ffmpeg(path, new_path, acodec, more_opts)
 386         except AudioConversionError as e:
 387             raise PostProcessingError(
 388                 'audio conversion failed: ' + e.msg)
 389         except Exception:
 390             raise PostProcessingError('error running ' + self.basename)
 391
 392         # Try to update the date time for extracted audio file.
 393         if information.get('filetime') is not None:
 394             self.try_utime(
 395                 new_path, time.time(), information['filetime'],
 396                 errnote='Cannot update utime of audio file')
 397
 398         return [path], information
 399
 400
 401 class FFmpegVideoConvertorPP(FFmpegPostProcessor):
 402     SUPPORTED_EXTS = ('mp4', 'mkv', 'flv', 'webm', 'mov', 'avi', 'mp3', 'mka', 'm4a', 'ogg', 'opus')
 403     FORMAT_RE = re.compile(r'{0}(?:/{0})*$'.format(r'(?:\w+>)?(?:%s)' % '|'.join(SUPPORTED_EXTS)))
 404     _action = 'converting'
 405
 406     def __init__(self, downloader=None, preferedformat=None):
 407         super(FFmpegVideoConvertorPP, self).__init__(downloader)
 408         self._preferedformats = preferedformat.lower().split('/')
 409
 410     def _target_ext(self, source_ext):
 411         for pair in self._preferedformats:
 412             kv = pair.split('>')
 413             if len(kv) == 1 or kv[0].strip() == source_ext:
 414                 return kv[-1].strip()
 415
 416     @staticmethod
 417     def _options(target_ext):
 418         if target_ext == 'avi':
 419             return ['-c:v', 'libxvid', '-vtag', 'XVID']
 420         return []
 421
 422     def run(self, information):
 423         path, source_ext = information['filepath'], information['ext'].lower()
 424         target_ext = self._target_ext(source_ext)
 425         _skip_msg = (
 426             'could not find a mapping for %s' if not target_ext
 427             else 'already is in target format %s' if source_ext == target_ext
 428             else None)
 429         if _skip_msg:
 430             self.to_screen('Not %s media file "%s"; %s' % (self._action, path, _skip_msg % source_ext))
 431             return [], information
 432
 433         prefix, sep, oldext = path.rpartition('.')
 434         outpath = prefix + sep + target_ext
 435         self.to_screen('%s video from %s to %s; Destination: %s' % (self._action.title(), source_ext, target_ext, outpath))
 436         self.run_ffmpeg(path, outpath, self._options(target_ext))
 437
 438         information['filepath'] = outpath
 439         information['format'] = information['ext'] = target_ext
 440         return [path], information
 441
 442
 443 class FFmpegVideoRemuxerPP(FFmpegVideoConvertorPP):
 444     _action = 'remuxing'
 445
 446     @staticmethod
 447     def _options(target_ext):
 448         options = ['-c', 'copy', '-map', '0', '-dn']
 449         if target_ext in ['mp4', 'm4a', 'mov']:
 450             options.extend(['-movflags', '+faststart'])
 451         return options
 452
 453
 454 class FFmpegEmbedSubtitlePP(FFmpegPostProcessor):
 455     def __init__(self, downloader=None, already_have_subtitle=False):
 456         super(FFmpegEmbedSubtitlePP, self).__init__(downloader)
 457         self._already_have_subtitle = already_have_subtitle
 458
 459     def run(self, information):
 460         if information['ext'] not in ('mp4', 'webm', 'mkv'):
 461             self.to_screen('Subtitles can only be embedded in mp4, webm or mkv files')
 462             return [], information
 463         subtitles = information.get('requested_subtitles')
 464         if not subtitles:
 465             self.to_screen('There aren\'t any subtitles to embed')
 466             return [], information
 467
 468         filename = information['filepath']
 469
 470         ext = information['ext']
 471         sub_langs, sub_names, sub_filenames = [], [], []
 472         webm_vtt_warn = False
 473         mp4_ass_warn = False
 474
 475         for lang, sub_info in subtitles.items():
 476             sub_ext = sub_info['ext']
 477             if sub_ext == 'json':
 478                 self.report_warning('JSON subtitles cannot be embedded')
 479             elif ext != 'webm' or ext == 'webm' and sub_ext == 'vtt':
 480                 sub_langs.append(lang)
 481                 sub_names.append(sub_info.get('name'))
 482                 sub_filenames.append(sub_info['filepath'])
 483             else:
 484                 if not webm_vtt_warn and ext == 'webm' and sub_ext != 'vtt':
 485                     webm_vtt_warn = True
 486                     self.report_warning('Only WebVTT subtitles can be embedded in webm files')
 487             if not mp4_ass_warn and ext == 'mp4' and sub_ext == 'ass':
 488                 mp4_ass_warn = True
 489                 self.report_warning('ASS subtitles cannot be properly embedded in mp4 files; expect issues')
 490
 491         if not sub_langs:
 492             return [], information
 493
 494         input_files = [filename] + sub_filenames
 495
 496         opts = [
 497             '-c', 'copy', '-map', '0', '-dn',
 498             # Don't copy the existing subtitles, we may be running the
 499             # postprocessor a second time
 500             '-map', '-0:s',
 501             # Don't copy Apple TV chapters track, bin_data (see #19042, #19024,
 502             # https://trac.ffmpeg.org/ticket/6016)
 503             '-map', '-0:d',
 504         ]
 505         if information['ext'] == 'mp4':
 506             opts += ['-c:s', 'mov_text']
 507         for i, (lang, name) in enumerate(zip(sub_langs, sub_names)):
 508             opts.extend(['-map', '%d:0' % (i + 1)])
 509             lang_code = ISO639Utils.short2long(lang) or lang
 510             opts.extend(['-metadata:s:s:%d' % i, 'language=%s' % lang_code])
 511             if name:
 512                 opts.extend(['-metadata:s:s:%d' % i, 'handler_name=%s' % name,
 513                              '-metadata:s:s:%d' % i, 'title=%s' % name])
 514
 515         temp_filename = prepend_extension(filename, 'temp')
 516         self.to_screen('Embedding subtitles in "%s"' % filename)
 517         self.run_ffmpeg_multiple_files(input_files, temp_filename, opts)
 518         os.remove(encodeFilename(filename))
 519         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 520
 521         files_to_delete = [] if self._already_have_subtitle else sub_filenames
 522         return files_to_delete, information
 523
 524
 525 class FFmpegMetadataPP(FFmpegPostProcessor):
 526     def run(self, info):
 527         metadata = {}
 528
 529         def add(meta_list, info_list=None):
 530             if not meta_list:
 531                 return
 532             if not info_list:
 533                 info_list = meta_list
 534             if not isinstance(meta_list, (list, tuple)):
 535                 meta_list = (meta_list,)
 536             if not isinstance(info_list, (list, tuple)):
 537                 info_list = (info_list,)
 538             for info_f in info_list:
 539                 if isinstance(info.get(info_f), (compat_str, compat_numeric_types)):
 540                     for meta_f in meta_list:
 541                         metadata[meta_f] = info[info_f]
 542                     break
 543
 544         # See [1-4] for some info on media metadata/metadata supported
 545         # by ffmpeg.
 546         # 1. https://kdenlive.org/en/project/adding-meta-data-to-mp4-video/
 547         # 2. https://wiki.multimedia.cx/index.php/FFmpeg_Metadata
 548         # 3. https://kodi.wiki/view/Video_file_tagging
 549
 550         add('title', ('track', 'title'))
 551         add('date', 'upload_date')
 552         add(('description', 'synopsis'), 'description')
 553         add(('purl', 'comment'), 'webpage_url')
 554         add('track', 'track_number')
 555         add('artist', ('artist', 'creator', 'uploader', 'uploader_id'))
 556         add('genre')
 557         add('album')
 558         add('album_artist')
 559         add('disc', 'disc_number')
 560         add('show', 'series')
 561         add('season_number')
 562         add('episode_id', ('episode', 'episode_id'))
 563         add('episode_sort', 'episode_number')
 564
 565         prefix = 'meta_'
 566         for key in filter(lambda k: k.startswith(prefix), info.keys()):
 567             add(key[len(prefix):], key)
 568
 569         if not metadata:
 570             self.to_screen('There isn\'t any metadata to add')
 571             return [], info
 572
 573         filename = info['filepath']
 574         temp_filename = prepend_extension(filename, 'temp')
 575         in_filenames = [filename]
 576         options = ['-map', '0', '-dn']
 577
 578         if info['ext'] == 'm4a':
 579             options.extend(['-vn', '-acodec', 'copy'])
 580         else:
 581             options.extend(['-c', 'copy'])
 582
 583         for name, value in metadata.items():
 584             options.extend(['-metadata', '%s=%s' % (name, value)])
 585
 586         chapters = info.get('chapters', [])
 587         if chapters:
 588             metadata_filename = replace_extension(filename, 'meta')
 589             with io.open(metadata_filename, 'wt', encoding='utf-8') as f:
 590                 def ffmpeg_escape(text):
 591                     return re.sub(r'(=|;|#|\\|\n)', r'\\\1', text)
 592
 593                 metadata_file_content = ';FFMETADATA1\n'
 594                 for chapter in chapters:
 595                     metadata_file_content += '[CHAPTER]\nTIMEBASE=1/1000\n'
 596                     metadata_file_content += 'START=%d\n' % (chapter['start_time'] * 1000)
 597                     metadata_file_content += 'END=%d\n' % (chapter['end_time'] * 1000)
 598                     chapter_title = chapter.get('title')
 599                     if chapter_title:
 600                         metadata_file_content += 'title=%s\n' % ffmpeg_escape(chapter_title)
 601                 f.write(metadata_file_content)
 602                 in_filenames.append(metadata_filename)
 603                 options.extend(['-map_metadata', '1'])
 604
 605         if ('no-attach-info-json' not in self.get_param('compat_opts', [])
 606                 and '__infojson_filename' in info and info['ext'] in ('mkv', 'mka')):
 607             old_stream, new_stream = self.get_stream_number(
 608                 filename, ('tags', 'mimetype'), 'application/json')
 609             if old_stream is not None:
 610                 options.extend(['-map', '-0:%d' % old_stream])
 611                 new_stream -= 1
 612
 613             options.extend([
 614                 '-attach', info['__infojson_filename'],
 615                 '-metadata:s:%d' % new_stream, 'mimetype=application/json'
 616             ])
 617
 618         self.to_screen('Adding metadata to \'%s\'' % filename)
 619         self.run_ffmpeg_multiple_files(in_filenames, temp_filename, options)
 620         if chapters:
 621             os.remove(metadata_filename)
 622         os.remove(encodeFilename(filename))
 623         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 624         return [], info
 625
 626
 627 class FFmpegMergerPP(FFmpegPostProcessor):
 628     def run(self, info):
 629         filename = info['filepath']
 630         temp_filename = prepend_extension(filename, 'temp')
 631         args = ['-c', 'copy']
 632         for (i, fmt) in enumerate(info['requested_formats']):
 633             if fmt.get('acodec') != 'none':
 634                 args.extend(['-map', '%u:a:0' % (i)])
 635             if fmt.get('vcodec') != 'none':
 636                 args.extend(['-map', '%u:v:0' % (i)])
 637         self.to_screen('Merging formats into "%s"' % filename)
 638         self.run_ffmpeg_multiple_files(info['__files_to_merge'], temp_filename, args)
 639         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 640         return info['__files_to_merge'], info
 641
 642     def can_merge(self):
 643         # TODO: figure out merge-capable ffmpeg version
 644         if self.basename != 'avconv':
 645             return True
 646
 647         required_version = '10-0'
 648         if is_outdated_version(
 649                 self._versions[self.basename], required_version):
 650             warning = ('Your copy of %s is outdated and unable to properly mux separate video and audio files, '
 651                        'yt-dlp will download single file media. '
 652                        'Update %s to version %s or newer to fix this.') % (
 653                            self.basename, self.basename, required_version)
 654             self.report_warning(warning)
 655             return False
 656         return True
 657
 658
 659 class FFmpegFixupStretchedPP(FFmpegPostProcessor):
 660     def run(self, info):
 661         stretched_ratio = info.get('stretched_ratio')
 662         if stretched_ratio is None or stretched_ratio == 1:
 663             return [], info
 664
 665         filename = info['filepath']
 666         temp_filename = prepend_extension(filename, 'temp')
 667
 668         options = ['-c', 'copy', '-map', '0', '-dn', '-aspect', '%f' % stretched_ratio]
 669         self.to_screen('Fixing aspect ratio in "%s"' % filename)
 670         self.run_ffmpeg(filename, temp_filename, options)
 671
 672         os.remove(encodeFilename(filename))
 673         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 674
 675         return [], info
 676
 677
 678 class FFmpegFixupM4aPP(FFmpegPostProcessor):
 679     def run(self, info):
 680         if info.get('container') != 'm4a_dash':
 681             return [], info
 682
 683         filename = info['filepath']
 684         temp_filename = prepend_extension(filename, 'temp')
 685
 686         options = ['-c', 'copy', '-map', '0', '-dn', '-f', 'mp4']
 687         self.to_screen('Correcting container in "%s"' % filename)
 688         self.run_ffmpeg(filename, temp_filename, options)
 689
 690         os.remove(encodeFilename(filename))
 691         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 692
 693         return [], info
 694
 695
 696 class FFmpegFixupM3u8PP(FFmpegPostProcessor):
 697     def run(self, info):
 698         filename = info['filepath']
 699         if self.get_audio_codec(filename) == 'aac':
 700             temp_filename = prepend_extension(filename, 'temp')
 701
 702             options = ['-c', 'copy', '-map', '0', '-dn', '-f', 'mp4', '-bsf:a', 'aac_adtstoasc']
 703             self.to_screen('Fixing malformed AAC bitstream in "%s"' % filename)
 704             self.run_ffmpeg(filename, temp_filename, options)
 705
 706             os.remove(encodeFilename(filename))
 707             os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 708         return [], info
 709
 710
 711 class FFmpegSubtitlesConvertorPP(FFmpegPostProcessor):
 712     SUPPORTED_EXTS = ('srt', 'vtt', 'ass', 'lrc')
 713
 714     def __init__(self, downloader=None, format=None):
 715         super(FFmpegSubtitlesConvertorPP, self).__init__(downloader)
 716         self.format = format
 717
 718     def run(self, info):
 719         subs = info.get('requested_subtitles')
 720         new_ext = self.format
 721         new_format = new_ext
 722         if new_format == 'vtt':
 723             new_format = 'webvtt'
 724         if subs is None:
 725             self.to_screen('There aren\'t any subtitles to convert')
 726             return [], info
 727         self.to_screen('Converting subtitles')
 728         sub_filenames = []
 729         for lang, sub in subs.items():
 730             ext = sub['ext']
 731             if ext == new_ext:
 732                 self.to_screen('Subtitle file for %s is already in the requested format' % new_ext)
 733                 continue
 734             elif ext == 'json':
 735                 self.to_screen(
 736                     'You have requested to convert json subtitles into another format, '
 737                     'which is currently not possible')
 738                 continue
 739             old_file = sub['filepath']
 740             sub_filenames.append(old_file)
 741             new_file = replace_extension(old_file, new_ext)
 742
 743             if ext in ('dfxp', 'ttml', 'tt'):
 744                 self.report_warning(
 745                     'You have requested to convert dfxp (TTML) subtitles into another format, '
 746                     'which results in style information loss')
 747
 748                 dfxp_file = old_file
 749                 srt_file = replace_extension(old_file, 'srt')
 750
 751                 with open(dfxp_file, 'rb') as f:
 752                     srt_data = dfxp2srt(f.read())
 753
 754                 with io.open(srt_file, 'wt', encoding='utf-8') as f:
 755                     f.write(srt_data)
 756                 old_file = srt_file
 757
 758                 subs[lang] = {
 759                     'ext': 'srt',
 760                     'data': srt_data,
 761                     'filepath': srt_file,
 762                 }
 763
 764                 if new_ext == 'srt':
 765                     continue
 766                 else:
 767                     sub_filenames.append(srt_file)
 768
 769             self.run_ffmpeg(old_file, new_file, ['-f', new_format])
 770
 771             with io.open(new_file, 'rt', encoding='utf-8') as f:
 772                 subs[lang] = {
 773                     'ext': new_ext,
 774                     'data': f.read(),
 775                     'filepath': new_file,
 776                 }
 777
 778             info['__files_to_move'][new_file] = replace_extension(
 779                 info['__files_to_move'][old_file], new_ext)
 780
 781         return sub_filenames, info
 782
 783
 784 class FFmpegSplitChaptersPP(FFmpegPostProcessor):
 785
 786     def _prepare_filename(self, number, chapter, info):
 787         info = info.copy()
 788         info.update({
 789             'section_number': number,
 790             'section_title': chapter.get('title'),
 791             'section_start': chapter.get('start_time'),
 792             'section_end': chapter.get('end_time'),
 793         })
 794         return self._downloader.prepare_filename(info, 'chapter')
 795
 796     def _ffmpeg_args_for_chapter(self, number, chapter, info):
 797         destination = self._prepare_filename(number, chapter, info)
 798         if not self._downloader._ensure_dir_exists(encodeFilename(destination)):
 799             return
 800
 801         chapter['filepath'] = destination
 802         self.to_screen('Chapter %03d; Destination: %s' % (number, destination))
 803         return (
 804             destination,
 805             ['-ss', compat_str(chapter['start_time']),
 806              '-t', compat_str(chapter['end_time'] - chapter['start_time'])])
 807
 808     def run(self, info):
 809         chapters = info.get('chapters') or []
 810         if not chapters:
 811             self.report_warning('Chapter information is unavailable')
 812             return [], info
 813
 814         self.to_screen('Splitting video by chapters; %d chapters found' % len(chapters))
 815         for idx, chapter in enumerate(chapters):
 816             destination, opts = self._ffmpeg_args_for_chapter(idx + 1, chapter, info)
 817             self.real_run_ffmpeg([(info['filepath'], opts)], [(destination, ['-c', 'copy'])])
 818         return [], info
 819
 820
 821 class FFmpegThumbnailsConvertorPP(FFmpegPostProcessor):
 822     SUPPORTED_EXTS = ('jpg', 'png')
 823
 824     def __init__(self, downloader=None, format=None):
 825         super(FFmpegThumbnailsConvertorPP, self).__init__(downloader)
 826         self.format = format
 827
 828     @staticmethod
 829     def is_webp(path):
 830         with open(encodeFilename(path), 'rb') as f:
 831             b = f.read(12)
 832         return b[0:4] == b'RIFF' and b[8:] == b'WEBP'
 833
 834     def fixup_webp(self, info, idx=-1):
 835         thumbnail_filename = info['thumbnails'][idx]['filepath']
 836         _, thumbnail_ext = os.path.splitext(thumbnail_filename)
 837         if thumbnail_ext:
 838             thumbnail_ext = thumbnail_ext[1:].lower()
 839             if thumbnail_ext != 'webp' and self.is_webp(thumbnail_filename):
 840                 self.to_screen('Correcting thumbnail "%s" extension to webp' % thumbnail_filename)
 841                 webp_filename = replace_extension(thumbnail_filename, 'webp')
 842                 if os.path.exists(webp_filename):
 843                     os.remove(webp_filename)
 844                 os.rename(encodeFilename(thumbnail_filename), encodeFilename(webp_filename))
 845                 info['thumbnails'][idx]['filepath'] = webp_filename
 846                 info['__files_to_move'][webp_filename] = replace_extension(
 847                     info['__files_to_move'].pop(thumbnail_filename), 'webp')
 848
 849     @staticmethod
 850     def _options(target_ext):
 851         if target_ext == 'jpg':
 852             return ['-bsf:v', 'mjpeg2jpeg']
 853         return []
 854
 855     def convert_thumbnail(self, thumbnail_filename, target_ext):
 856         thumbnail_conv_filename = replace_extension(thumbnail_filename, target_ext)
 857
 858         self.to_screen('Converting thumbnail "%s" to %s' % (thumbnail_filename, target_ext))
 859         self.real_run_ffmpeg(
 860             [(thumbnail_filename, ['-f', 'image2', '-pattern_type', 'none'])],
 861             [(thumbnail_conv_filename.replace('%', '%%'), self._options(target_ext))])
 862         return thumbnail_conv_filename
 863
 864     def run(self, info):
 865         files_to_delete = []
 866         has_thumbnail = False
 867
 868         for idx, thumbnail_dict in enumerate(info['thumbnails']):
 869             if 'filepath' not in thumbnail_dict:
 870                 continue
 871             has_thumbnail = True
 872             self.fixup_webp(info, idx)
 873             original_thumbnail = thumbnail_dict['filepath']
 874             _, thumbnail_ext = os.path.splitext(original_thumbnail)
 875             if thumbnail_ext:
 876                 thumbnail_ext = thumbnail_ext[1:].lower()
 877             if thumbnail_ext == self.format:
 878                 self.to_screen('Thumbnail "%s" is already in the requested format' % original_thumbnail)
 879                 continue
 880             thumbnail_dict['filepath'] = self.convert_thumbnail(original_thumbnail, self.format)
 881             files_to_delete.append(original_thumbnail)
 882             info['__files_to_move'][thumbnail_dict['filepath']] = replace_extension(
 883                 info['__files_to_move'][original_thumbnail], self.format)
 884
 885         if not has_thumbnail:
 886             self.to_screen('There aren\'t any thumbnails to convert')
 887         return files_to_delete, info