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  70             warning = 'Your copy of %s is outdated, update %s to version %s or newer if you encounter any errors.'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