yt_dlp/postprocessor/ffmpeg.py

   1 from __future__ import unicode_literals
   2
   3 import io
   4 import itertools
   5 import os
   6 import subprocess
   7 import time
   8 import re
   9 import json
  10
  11 from .common import AudioConversionError, PostProcessor
  12
  13 from ..compat import compat_str, compat_numeric_types
  14 from ..utils import (
  15     dfxp2srt,
  16     encodeArgument,
  17     encodeFilename,
  18     float_or_none,
  19     get_exe_version,
  20     is_outdated_version,
  21     ISO639Utils,
  22     orderedSet,
  23     PostProcessingError,
  24     prepend_extension,
  25     process_communicate_or_kill,
  26     replace_extension,
  27     shell_quote,
  28     traverse_obj,
  29     variadic,
  30 )
  31
  32
  33 EXT_TO_OUT_FORMATS = {
  34     'aac': 'adts',
  35     'flac': 'flac',
  36     'm4a': 'ipod',
  37     'mka': 'matroska',
  38     'mkv': 'matroska',
  39     'mpg': 'mpeg',
  40     'ogv': 'ogg',
  41     'ts': 'mpegts',
  42     'wma': 'asf',
  43     'wmv': 'asf',
  44 }
  45 ACODECS = {
  46     'mp3': 'libmp3lame',
  47     'aac': 'aac',
  48     'flac': 'flac',
  49     'm4a': 'aac',
  50     'opus': 'libopus',
  51     'vorbis': 'libvorbis',
  52     'wav': None,
  53 }
  54
  55
  56 class FFmpegPostProcessorError(PostProcessingError):
  57     pass
  58
  59
  60 class FFmpegPostProcessor(PostProcessor):
  61     def __init__(self, downloader=None):
  62         PostProcessor.__init__(self, downloader)
  63         self._determine_executables()
  64
  65     def check_version(self):
  66         if not self.available:
  67             raise FFmpegPostProcessorError('ffmpeg not found. Please install or provide the path using --ffmpeg-location')
  68
  69         required_version = '10-0' if self.basename == 'avconv' else '1.0'
  70         if is_outdated_version(
  71                 self._versions[self.basename], required_version):
  72             warning = 'Your copy of %s is outdated, update %s to version %s or newer if you encounter any errors.' % (
  73                 self.basename, self.basename, required_version)
  74             self.report_warning(warning)
  75
  76     @staticmethod
  77     def get_versions(downloader=None):
  78         return FFmpegPostProcessor(downloader)._versions
  79
  80     def _determine_executables(self):
  81         programs = ['avprobe', 'avconv', 'ffmpeg', 'ffprobe']
  82         prefer_ffmpeg = True
  83
  84         def get_ffmpeg_version(path):
  85             ver = get_exe_version(path, args=['-version'])
  86             if ver:
  87                 regexs = [
  88                     r'(?:\d+:)?([0-9.]+)-[0-9]+ubuntu[0-9.]+$',  # Ubuntu, see [1]
  89                     r'n([0-9.]+)$',  # Arch Linux
  90                     # 1. http://www.ducea.com/2006/06/17/ubuntu-package-version-naming-explanation/
  91                 ]
  92                 for regex in regexs:
  93                     mobj = re.match(regex, ver)
  94                     if mobj:
  95                         ver = mobj.group(1)
  96             return ver
  97
  98         self.basename = None
  99         self.probe_basename = None
 100
 101         self._paths = None
 102         self._versions = None
 103         if self._downloader:
 104             prefer_ffmpeg = self.get_param('prefer_ffmpeg', True)
 105             location = self.get_param('ffmpeg_location')
 106             if location is not None:
 107                 if not os.path.exists(location):
 108                     self.report_warning(
 109                         'ffmpeg-location %s does not exist! '
 110                         'Continuing without ffmpeg.' % (location))
 111                     self._versions = {}
 112                     return
 113                 elif os.path.isdir(location):
 114                     dirname, basename = location, None
 115                 else:
 116                     basename = os.path.splitext(os.path.basename(location))[0]
 117                     basename = next((p for p in programs if basename.startswith(p)), 'ffmpeg')
 118                     dirname = os.path.dirname(os.path.abspath(location))
 119                     if basename in ('ffmpeg', 'ffprobe'):
 120                         prefer_ffmpeg = True
 121
 122                 self._paths = dict(
 123                     (p, os.path.join(dirname, p)) for p in programs)
 124                 if basename:
 125                     self._paths[basename] = location
 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 or provide the path using --ffmpeg-location')
 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 or provide the path using --ffmpeg-location')
 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_obj(stream, keys, casesense=False) == value),
 234             None)
 235         return num, len(streams)
 236
 237     def _get_real_video_duration(self, info, fatal=True):
 238         try:
 239             if '_real_duration' not in info:
 240                 info['_real_duration'] = float_or_none(
 241                     traverse_obj(self.get_metadata_object(info['filepath']), ('format', 'duration')))
 242             if not info['_real_duration']:
 243                 raise PostProcessingError('ffprobe returned empty duration')
 244         except PostProcessingError as e:
 245             if fatal:
 246                 raise PostProcessingError(f'Unable to determine video duration; {e}')
 247         return info.setdefault('_real_duration', None)
 248
 249     def _duration_mismatch(self, d1, d2):
 250         if not d1 or not d2:
 251             return None
 252         return abs(d1 - d2) > 1
 253
 254     def run_ffmpeg_multiple_files(self, input_paths, out_path, opts, **kwargs):
 255         return self.real_run_ffmpeg(
 256             [(path, []) for path in input_paths],
 257             [(out_path, opts)], **kwargs)
 258
 259     def real_run_ffmpeg(self, input_path_opts, output_path_opts, *, expected_retcodes=(0,)):
 260         self.check_version()
 261
 262         oldest_mtime = min(
 263             os.stat(encodeFilename(path)).st_mtime for path, _ in input_path_opts if path)
 264
 265         cmd = [encodeFilename(self.executable, True), encodeArgument('-y'), encodeArgument('-probesize'), encodeArgument('max')]
 266         # avconv does not have repeat option
 267         if self.basename == 'ffmpeg':
 268             cmd += [encodeArgument('-loglevel'), encodeArgument('repeat+info')]
 269
 270         def make_args(file, args, name, number):
 271             keys = ['_%s%d' % (name, number), '_%s' % name]
 272             if name == 'o' and number == 1:
 273                 keys.append('')
 274             args += self._configuration_args(self.basename, keys)
 275             if name == 'i':
 276                 args.append('-i')
 277             return (
 278                 [encodeArgument(arg) for arg in args]
 279                 + [encodeFilename(self._ffmpeg_filename_argument(file), True)])
 280
 281         for arg_type, path_opts in (('i', input_path_opts), ('o', output_path_opts)):
 282             cmd += itertools.chain.from_iterable(
 283                 make_args(path, list(opts), arg_type, i + 1)
 284                 for i, (path, opts) in enumerate(path_opts) if path)
 285
 286         self.write_debug('ffmpeg command line: %s' % shell_quote(cmd))
 287         p = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE, stdin=subprocess.PIPE)
 288         stdout, stderr = process_communicate_or_kill(p)
 289         if p.returncode not in variadic(expected_retcodes):
 290             stderr = stderr.decode('utf-8', 'replace').strip()
 291             self.write_debug(stderr)
 292             raise FFmpegPostProcessorError(stderr.split('\n')[-1])
 293         for out_path, _ in output_path_opts:
 294             if out_path:
 295                 self.try_utime(out_path, oldest_mtime, oldest_mtime)
 296         return stderr.decode('utf-8', 'replace')
 297
 298     def run_ffmpeg(self, path, out_path, opts, **kwargs):
 299         return self.run_ffmpeg_multiple_files([path], out_path, opts, **kwargs)
 300
 301     @staticmethod
 302     def _ffmpeg_filename_argument(fn):
 303         # Always use 'file:' because the filename may contain ':' (ffmpeg
 304         # interprets that as a protocol) or can start with '-' (-- is broken in
 305         # ffmpeg, see https://ffmpeg.org/trac/ffmpeg/ticket/2127 for details)
 306         # Also leave '-' intact in order not to break streaming to stdout.
 307         if fn.startswith(('http://', 'https://')):
 308             return fn
 309         return 'file:' + fn if fn != '-' else fn
 310
 311     @staticmethod
 312     def _quote_for_ffmpeg(string):
 313         # See https://ffmpeg.org/ffmpeg-utils.html#toc-Quoting-and-escaping
 314         # A sequence of '' produces '\'''\'';
 315         # final replace removes the empty '' between \' \'.
 316         string = string.replace("'", r"'\''").replace("'''", "'")
 317         # Handle potential ' at string boundaries.
 318         string = string[1:] if string[0] == "'" else "'" + string
 319         return string[:-1] if string[-1] == "'" else string + "'"
 320
 321     def force_keyframes(self, filename, timestamps):
 322         timestamps = orderedSet(timestamps)
 323         if timestamps[0] == 0:
 324             timestamps = timestamps[1:]
 325         keyframe_file = prepend_extension(filename, 'keyframes.temp')
 326         self.to_screen(f'Re-encoding "{filename}" with appropriate keyframes')
 327         self.run_ffmpeg(filename, keyframe_file, ['-force_key_frames', ','.join(
 328             f'{t:.6f}' for t in timestamps)])
 329         return keyframe_file
 330
 331     def concat_files(self, in_files, out_file, concat_opts=None):
 332         """
 333         Use concat demuxer to concatenate multiple files having identical streams.
 334
 335         Only inpoint, outpoint, and duration concat options are supported.
 336         See https://ffmpeg.org/ffmpeg-formats.html#concat-1 for details
 337         """
 338         concat_file = f'{out_file}.concat'
 339         self.write_debug(f'Writing concat spec to {concat_file}')
 340         with open(concat_file, 'wt', encoding='utf-8') as f:
 341             f.writelines(self._concat_spec(in_files, concat_opts))
 342
 343         out_flags = ['-c', 'copy']
 344         if out_file.rpartition('.')[-1] in ('mp4', 'mov'):
 345             # For some reason, '-c copy' is not enough to copy subtitles
 346             out_flags.extend(['-c:s', 'mov_text', '-movflags', '+faststart'])
 347
 348         try:
 349             self.real_run_ffmpeg(
 350                 [(concat_file, ['-hide_banner', '-nostdin', '-f', 'concat', '-safe', '0'])],
 351                 [(out_file, out_flags)])
 352         finally:
 353             os.remove(concat_file)
 354
 355     @classmethod
 356     def _concat_spec(cls, in_files, concat_opts=None):
 357         if concat_opts is None:
 358             concat_opts = [{}] * len(in_files)
 359         yield 'ffconcat version 1.0\n'
 360         for file, opts in zip(in_files, concat_opts):
 361             yield f'file {cls._quote_for_ffmpeg(cls._ffmpeg_filename_argument(file))}\n'
 362             # Iterate explicitly to yield the following directives in order, ignoring the rest.
 363             for directive in 'inpoint', 'outpoint', 'duration':
 364                 if directive in opts:
 365                     yield f'{directive} {opts[directive]}\n'
 366
 367
 368 class FFmpegExtractAudioPP(FFmpegPostProcessor):
 369     COMMON_AUDIO_EXTS = ('wav', 'flac', 'm4a', 'aiff', 'mp3', 'ogg', 'mka', 'opus', 'wma')
 370     SUPPORTED_EXTS = ('best', 'aac', 'flac', 'mp3', 'm4a', 'opus', 'vorbis', 'wav')
 371
 372     def __init__(self, downloader=None, preferredcodec=None, preferredquality=None, nopostoverwrites=False):
 373         FFmpegPostProcessor.__init__(self, downloader)
 374         self._preferredcodec = preferredcodec or 'best'
 375         self._preferredquality = preferredquality
 376         self._nopostoverwrites = nopostoverwrites
 377
 378     def run_ffmpeg(self, path, out_path, codec, more_opts):
 379         if codec is None:
 380             acodec_opts = []
 381         else:
 382             acodec_opts = ['-acodec', codec]
 383         opts = ['-vn'] + acodec_opts + more_opts
 384         try:
 385             FFmpegPostProcessor.run_ffmpeg(self, path, out_path, opts)
 386         except FFmpegPostProcessorError as err:
 387             raise AudioConversionError(err.msg)
 388
 389     @PostProcessor._restrict_to(images=False)
 390     def run(self, information):
 391         path = information['filepath']
 392         orig_ext = information['ext']
 393
 394         if self._preferredcodec == 'best' and orig_ext in self.COMMON_AUDIO_EXTS:
 395             self.to_screen('Skipping audio extraction since the file is already in a common audio format')
 396             return [], information
 397
 398         filecodec = self.get_audio_codec(path)
 399         if filecodec is None:
 400             raise PostProcessingError('WARNING: unable to obtain file audio codec with ffprobe')
 401
 402         more_opts = []
 403         if self._preferredcodec == 'best' or self._preferredcodec == filecodec or (self._preferredcodec == 'm4a' and filecodec == 'aac'):
 404             if filecodec == 'aac' and self._preferredcodec in ['m4a', 'best']:
 405                 # Lossless, but in another container
 406                 acodec = 'copy'
 407                 extension = 'm4a'
 408                 more_opts = ['-bsf:a', 'aac_adtstoasc']
 409             elif filecodec in ['aac', 'flac', 'mp3', 'vorbis', 'opus']:
 410                 # Lossless if possible
 411                 acodec = 'copy'
 412                 extension = filecodec
 413                 if filecodec == 'aac':
 414                     more_opts = ['-f', 'adts']
 415                 if filecodec == 'vorbis':
 416                     extension = 'ogg'
 417             else:
 418                 # MP3 otherwise.
 419                 acodec = 'libmp3lame'
 420                 extension = 'mp3'
 421                 more_opts = []
 422                 if self._preferredquality is not None:
 423                     if int(self._preferredquality) < 10:
 424                         more_opts += ['-q:a', self._preferredquality]
 425                     else:
 426                         more_opts += ['-b:a', self._preferredquality + 'k']
 427         else:
 428             # We convert the audio (lossy if codec is lossy)
 429             acodec = ACODECS[self._preferredcodec]
 430             extension = self._preferredcodec
 431             more_opts = []
 432             if self._preferredquality is not None:
 433                 # The opus codec doesn't support the -aq option
 434                 if int(self._preferredquality) < 10 and extension != 'opus':
 435                     more_opts += ['-q:a', self._preferredquality]
 436                 else:
 437                     more_opts += ['-b:a', self._preferredquality + 'k']
 438             if self._preferredcodec == 'aac':
 439                 more_opts += ['-f', 'adts']
 440             if self._preferredcodec == 'm4a':
 441                 more_opts += ['-bsf:a', 'aac_adtstoasc']
 442             if self._preferredcodec == 'vorbis':
 443                 extension = 'ogg'
 444             if self._preferredcodec == 'wav':
 445                 extension = 'wav'
 446                 more_opts += ['-f', 'wav']
 447
 448         prefix, sep, ext = path.rpartition('.')  # not os.path.splitext, since the latter does not work on unicode in all setups
 449         new_path = prefix + sep + extension
 450
 451         information['filepath'] = new_path
 452         information['ext'] = extension
 453
 454         # If we download foo.mp3 and convert it to... foo.mp3, then don't delete foo.mp3, silly.
 455         if (new_path == path
 456                 or (self._nopostoverwrites and os.path.exists(encodeFilename(new_path)))):
 457             self.to_screen('Post-process file %s exists, skipping' % new_path)
 458             return [], information
 459
 460         try:
 461             self.to_screen('Destination: ' + new_path)
 462             self.run_ffmpeg(path, new_path, acodec, more_opts)
 463         except AudioConversionError as e:
 464             raise PostProcessingError(
 465                 'audio conversion failed: ' + e.msg)
 466         except Exception:
 467             raise PostProcessingError('error running ' + self.basename)
 468
 469         # Try to update the date time for extracted audio file.
 470         if information.get('filetime') is not None:
 471             self.try_utime(
 472                 new_path, time.time(), information['filetime'],
 473                 errnote='Cannot update utime of audio file')
 474
 475         return [path], information
 476
 477
 478 class FFmpegVideoConvertorPP(FFmpegPostProcessor):
 479     SUPPORTED_EXTS = ('mp4', 'mkv', 'flv', 'webm', 'mov', 'avi', 'mp3', 'mka', 'm4a', 'ogg', 'opus')
 480     FORMAT_RE = re.compile(r'{0}(?:/{0})*$'.format(r'(?:\w+>)?(?:%s)' % '|'.join(SUPPORTED_EXTS)))
 481     _ACTION = 'converting'
 482
 483     def __init__(self, downloader=None, preferedformat=None):
 484         super(FFmpegVideoConvertorPP, self).__init__(downloader)
 485         self._preferedformats = preferedformat.lower().split('/')
 486
 487     def _target_ext(self, source_ext):
 488         for pair in self._preferedformats:
 489             kv = pair.split('>')
 490             if len(kv) == 1 or kv[0].strip() == source_ext:
 491                 return kv[-1].strip()
 492
 493     @staticmethod
 494     def _options(target_ext):
 495         if target_ext == 'avi':
 496             return ['-c:v', 'libxvid', '-vtag', 'XVID']
 497         return []
 498
 499     @PostProcessor._restrict_to(images=False)
 500     def run(self, info):
 501         filename, source_ext = info['filepath'], info['ext'].lower()
 502         target_ext = self._target_ext(source_ext)
 503         _skip_msg = (
 504             f'could not find a mapping for {source_ext}' if not target_ext
 505             else f'already is in target format {source_ext}' if source_ext == target_ext
 506             else None)
 507         if _skip_msg:
 508             self.to_screen(f'Not {self._ACTION} media file {filename!r}; {_skip_msg}')
 509             return [], info
 510
 511         outpath = replace_extension(filename, target_ext, source_ext)
 512         self.to_screen(f'{self._ACTION.title()} video from {source_ext} to {target_ext}; Destination: {outpath}')
 513         self.run_ffmpeg(filename, outpath, self._options(target_ext))
 514
 515         info['filepath'] = outpath
 516         info['format'] = info['ext'] = target_ext
 517         return [filename], info
 518
 519
 520 class FFmpegVideoRemuxerPP(FFmpegVideoConvertorPP):
 521     _ACTION = 'remuxing'
 522
 523     @staticmethod
 524     def _options(target_ext):
 525         options = ['-c', 'copy', '-map', '0', '-dn']
 526         if target_ext in ['mp4', 'm4a', 'mov']:
 527             options.extend(['-movflags', '+faststart'])
 528         return options
 529
 530
 531 class FFmpegEmbedSubtitlePP(FFmpegPostProcessor):
 532     def __init__(self, downloader=None, already_have_subtitle=False):
 533         super(FFmpegEmbedSubtitlePP, self).__init__(downloader)
 534         self._already_have_subtitle = already_have_subtitle
 535
 536     @PostProcessor._restrict_to(images=False)
 537     def run(self, information):
 538         if information['ext'] not in ('mp4', 'webm', 'mkv'):
 539             self.to_screen('Subtitles can only be embedded in mp4, webm or mkv files')
 540             return [], information
 541         subtitles = information.get('requested_subtitles')
 542         if not subtitles:
 543             self.to_screen('There aren\'t any subtitles to embed')
 544             return [], information
 545
 546         filename = information['filepath']
 547         if information.get('duration') and self._duration_mismatch(
 548                 self._get_real_video_duration(information, False), information['duration']):
 549             self.to_screen(f'Skipping {self.pp_key()} since the real and expected durations mismatch')
 550             return [], information
 551
 552         ext = information['ext']
 553         sub_langs, sub_names, sub_filenames = [], [], []
 554         webm_vtt_warn = False
 555         mp4_ass_warn = False
 556
 557         for lang, sub_info in subtitles.items():
 558             if not os.path.exists(information.get('filepath', '')):
 559                 self.report_warning(f'Skipping embedding {lang} subtitle because the file is missing')
 560                 continue
 561             sub_ext = sub_info['ext']
 562             if sub_ext == 'json':
 563                 self.report_warning('JSON subtitles cannot be embedded')
 564             elif ext != 'webm' or ext == 'webm' and sub_ext == 'vtt':
 565                 sub_langs.append(lang)
 566                 sub_names.append(sub_info.get('name'))
 567                 sub_filenames.append(sub_info['filepath'])
 568             else:
 569                 if not webm_vtt_warn and ext == 'webm' and sub_ext != 'vtt':
 570                     webm_vtt_warn = True
 571                     self.report_warning('Only WebVTT subtitles can be embedded in webm files')
 572             if not mp4_ass_warn and ext == 'mp4' and sub_ext == 'ass':
 573                 mp4_ass_warn = True
 574                 self.report_warning('ASS subtitles cannot be properly embedded in mp4 files; expect issues')
 575
 576         if not sub_langs:
 577             return [], information
 578
 579         input_files = [filename] + sub_filenames
 580
 581         opts = [
 582             '-c', 'copy', '-map', '0', '-dn',
 583             # Don't copy the existing subtitles, we may be running the
 584             # postprocessor a second time
 585             '-map', '-0:s',
 586             # Don't copy Apple TV chapters track, bin_data (see #19042, #19024,
 587             # https://trac.ffmpeg.org/ticket/6016)
 588             '-map', '-0:d',
 589         ]
 590         if information['ext'] == 'mp4':
 591             opts += ['-c:s', 'mov_text']
 592         for i, (lang, name) in enumerate(zip(sub_langs, sub_names)):
 593             opts.extend(['-map', '%d:0' % (i + 1)])
 594             lang_code = ISO639Utils.short2long(lang) or lang
 595             opts.extend(['-metadata:s:s:%d' % i, 'language=%s' % lang_code])
 596             if name:
 597                 opts.extend(['-metadata:s:s:%d' % i, 'handler_name=%s' % name,
 598                              '-metadata:s:s:%d' % i, 'title=%s' % name])
 599
 600         temp_filename = prepend_extension(filename, 'temp')
 601         self.to_screen('Embedding subtitles in "%s"' % filename)
 602         self.run_ffmpeg_multiple_files(input_files, temp_filename, opts)
 603         os.replace(temp_filename, filename)
 604
 605         files_to_delete = [] if self._already_have_subtitle else sub_filenames
 606         return files_to_delete, information
 607
 608
 609 class FFmpegMetadataPP(FFmpegPostProcessor):
 610
 611     def __init__(self, downloader, add_metadata=True, add_chapters=True):
 612         FFmpegPostProcessor.__init__(self, downloader)
 613         self._add_metadata = add_metadata
 614         self._add_chapters = add_chapters
 615
 616     @staticmethod
 617     def _options(target_ext):
 618         yield from ('-map', '0', '-dn')
 619         if target_ext == 'm4a':
 620             yield from ('-vn', '-acodec', 'copy')
 621         else:
 622             yield from ('-c', 'copy')
 623
 624     @PostProcessor._restrict_to(images=False)
 625     def run(self, info):
 626         filename, metadata_filename = info['filepath'], None
 627         options = []
 628         if self._add_chapters and info.get('chapters'):
 629             metadata_filename = replace_extension(filename, 'meta')
 630             options.extend(self._get_chapter_opts(info['chapters'], metadata_filename))
 631         if self._add_metadata:
 632             options.extend(self._get_metadata_opts(info))
 633
 634         if not options:
 635             self.to_screen('There isn\'t any metadata to add')
 636             return [], info
 637
 638         temp_filename = prepend_extension(filename, 'temp')
 639         self.to_screen('Adding metadata to "%s"' % filename)
 640         self.run_ffmpeg_multiple_files(
 641             (filename, metadata_filename), temp_filename,
 642             itertools.chain(self._options(info['ext']), *options))
 643         if metadata_filename:
 644             os.remove(metadata_filename)
 645         os.replace(temp_filename, filename)
 646         return [], info
 647
 648     @staticmethod
 649     def _get_chapter_opts(chapters, metadata_filename):
 650         with io.open(metadata_filename, 'wt', encoding='utf-8') as f:
 651             def ffmpeg_escape(text):
 652                 return re.sub(r'([\\=;#\n])', r'\\\1', text)
 653
 654             metadata_file_content = ';FFMETADATA1\n'
 655             for chapter in chapters:
 656                 metadata_file_content += '[CHAPTER]\nTIMEBASE=1/1000\n'
 657                 metadata_file_content += 'START=%d\n' % (chapter['start_time'] * 1000)
 658                 metadata_file_content += 'END=%d\n' % (chapter['end_time'] * 1000)
 659                 chapter_title = chapter.get('title')
 660                 if chapter_title:
 661                     metadata_file_content += 'title=%s\n' % ffmpeg_escape(chapter_title)
 662             f.write(metadata_file_content)
 663         yield ('-map_metadata', '1')
 664
 665     def _get_metadata_opts(self, info):
 666         metadata = {}
 667
 668         def add(meta_list, info_list=None):
 669             if not meta_list:
 670                 return
 671             for info_f in variadic(info_list or meta_list):
 672                 if isinstance(info.get(info_f), (compat_str, compat_numeric_types)):
 673                     for meta_f in variadic(meta_list):
 674                         metadata[meta_f] = info[info_f]
 675                     break
 676
 677         # See [1-4] for some info on media metadata/metadata supported
 678         # by ffmpeg.
 679         # 1. https://kdenlive.org/en/project/adding-meta-data-to-mp4-video/
 680         # 2. https://wiki.multimedia.cx/index.php/FFmpeg_Metadata
 681         # 3. https://kodi.wiki/view/Video_file_tagging
 682
 683         add('title', ('track', 'title'))
 684         add('date', 'upload_date')
 685         add(('description', 'synopsis'), 'description')
 686         add(('purl', 'comment'), 'webpage_url')
 687         add('track', 'track_number')
 688         add('artist', ('artist', 'creator', 'uploader', 'uploader_id'))
 689         add('genre')
 690         add('album')
 691         add('album_artist')
 692         add('disc', 'disc_number')
 693         add('show', 'series')
 694         add('season_number')
 695         add('episode_id', ('episode', 'episode_id'))
 696         add('episode_sort', 'episode_number')
 697
 698         prefix = 'meta_'
 699         for key in filter(lambda k: k.startswith(prefix), info.keys()):
 700             add(key[len(prefix):], key)
 701
 702         for name, value in metadata.items():
 703             yield ('-metadata', f'{name}={value}')
 704
 705         stream_idx = 0
 706         for fmt in info.get('requested_formats') or []:
 707             stream_count = 2 if 'none' not in (fmt.get('vcodec'), fmt.get('acodec')) else 1
 708             if fmt.get('language'):
 709                 lang = ISO639Utils.short2long(fmt['language']) or fmt['language']
 710                 for i in range(stream_count):
 711                     yield ('-metadata:s:%d' % (stream_idx + i), 'language=%s' % lang)
 712             stream_idx += stream_count
 713
 714         if ('no-attach-info-json' not in self.get_param('compat_opts', [])
 715                 and '__infojson_filename' in info and info['ext'] in ('mkv', 'mka')):
 716             old_stream, new_stream = self.get_stream_number(info['filepath'], ('tags', 'mimetype'), 'application/json')
 717             if old_stream is not None:
 718                 yield ('-map', '-0:%d' % old_stream)
 719                 new_stream -= 1
 720
 721             yield ('-attach', info['__infojson_filename'],
 722                    '-metadata:s:%d' % new_stream, 'mimetype=application/json')
 723
 724
 725 class FFmpegMergerPP(FFmpegPostProcessor):
 726     @PostProcessor._restrict_to(images=False)
 727     def run(self, info):
 728         filename = info['filepath']
 729         temp_filename = prepend_extension(filename, 'temp')
 730         args = ['-c', 'copy']
 731         audio_streams = 0
 732         for (i, fmt) in enumerate(info['requested_formats']):
 733             if fmt.get('acodec') != 'none':
 734                 args.extend(['-map', f'{i}:a:0'])
 735                 if self.get_audio_codec(fmt['filepath']) == 'aac':
 736                     args.extend([f'-bsf:a:{audio_streams}', 'aac_adtstoasc'])
 737                 audio_streams += 1
 738             if fmt.get('vcodec') != 'none':
 739                 args.extend(['-map', '%u:v:0' % (i)])
 740         self.to_screen('Merging formats into "%s"' % filename)
 741         self.run_ffmpeg_multiple_files(info['__files_to_merge'], temp_filename, args)
 742         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 743         return info['__files_to_merge'], info
 744
 745     def can_merge(self):
 746         # TODO: figure out merge-capable ffmpeg version
 747         if self.basename != 'avconv':
 748             return True
 749
 750         required_version = '10-0'
 751         if is_outdated_version(
 752                 self._versions[self.basename], required_version):
 753             warning = ('Your copy of %s is outdated and unable to properly mux separate video and audio files, '
 754                        'yt-dlp will download single file media. '
 755                        'Update %s to version %s or newer to fix this.') % (
 756                            self.basename, self.basename, required_version)
 757             self.report_warning(warning)
 758             return False
 759         return True
 760
 761
 762 class FFmpegFixupPostProcessor(FFmpegPostProcessor):
 763     def _fixup(self, msg, filename, options):
 764         temp_filename = prepend_extension(filename, 'temp')
 765
 766         self.to_screen(f'{msg} of "{filename}"')
 767         self.run_ffmpeg(filename, temp_filename, options)
 768
 769         os.replace(temp_filename, filename)
 770
 771
 772 class FFmpegFixupStretchedPP(FFmpegFixupPostProcessor):
 773     @PostProcessor._restrict_to(images=False, audio=False)
 774     def run(self, info):
 775         stretched_ratio = info.get('stretched_ratio')
 776         if stretched_ratio not in (None, 1):
 777             self._fixup('Fixing aspect ratio', info['filepath'], [
 778                 '-c', 'copy', '-map', '0', '-dn', '-aspect', '%f' % stretched_ratio])
 779         return [], info
 780
 781
 782 class FFmpegFixupM4aPP(FFmpegFixupPostProcessor):
 783     @PostProcessor._restrict_to(images=False, video=False)
 784     def run(self, info):
 785         if info.get('container') == 'm4a_dash':
 786             self._fixup('Correcting container', info['filepath'], [
 787                 '-c', 'copy', '-map', '0', '-dn', '-f', 'mp4'])
 788         return [], info
 789
 790
 791 class FFmpegFixupM3u8PP(FFmpegFixupPostProcessor):
 792     @PostProcessor._restrict_to(images=False)
 793     def run(self, info):
 794         if self.get_audio_codec(info['filepath']) == 'aac':
 795             self._fixup('Fixing malformed AAC bitstream', info['filepath'], [
 796                 '-c', 'copy', '-map', '0', '-dn', '-f', 'mp4', '-bsf:a', 'aac_adtstoasc'])
 797         return [], info
 798
 799
 800 class FFmpegFixupTimestampPP(FFmpegFixupPostProcessor):
 801
 802     def __init__(self, downloader=None, trim=0.001):
 803         # "trim" should be used when the video contains unintended packets
 804         super(FFmpegFixupTimestampPP, self).__init__(downloader)
 805         assert isinstance(trim, (int, float))
 806         self.trim = str(trim)
 807
 808     @PostProcessor._restrict_to(images=False)
 809     def run(self, info):
 810         required_version = '4.4'
 811         if is_outdated_version(self._versions[self.basename], required_version):
 812             self.report_warning(
 813                 'A re-encode is needed to fix timestamps in older versions of ffmpeg. '
 814                 f'Please install ffmpeg {required_version} or later to fixup without re-encoding')
 815             opts = ['-vf', 'setpts=PTS-STARTPTS']
 816         else:
 817             opts = ['-c', 'copy', '-bsf', 'setts=ts=TS-STARTPTS']
 818         self._fixup('Fixing frame timestamp', info['filepath'], opts + ['-map', '0', '-dn', '-ss', self.trim])
 819         return [], info
 820
 821
 822 class FFmpegFixupDurationPP(FFmpegFixupPostProcessor):
 823     @PostProcessor._restrict_to(images=False)
 824     def run(self, info):
 825         self._fixup('Fixing video duration', info['filepath'], ['-c', 'copy', '-map', '0', '-dn'])
 826         return [], info
 827
 828
 829 class FFmpegSubtitlesConvertorPP(FFmpegPostProcessor):
 830     SUPPORTED_EXTS = ('srt', 'vtt', 'ass', 'lrc')
 831
 832     def __init__(self, downloader=None, format=None):
 833         super(FFmpegSubtitlesConvertorPP, self).__init__(downloader)
 834         self.format = format
 835
 836     def run(self, info):
 837         subs = info.get('requested_subtitles')
 838         new_ext = self.format
 839         new_format = new_ext
 840         if new_format == 'vtt':
 841             new_format = 'webvtt'
 842         if subs is None:
 843             self.to_screen('There aren\'t any subtitles to convert')
 844             return [], info
 845         self.to_screen('Converting subtitles')
 846         sub_filenames = []
 847         for lang, sub in subs.items():
 848             ext = sub['ext']
 849             if ext == new_ext:
 850                 self.to_screen('Subtitle file for %s is already in the requested format' % new_ext)
 851                 continue
 852             elif ext == 'json':
 853                 self.to_screen(
 854                     'You have requested to convert json subtitles into another format, '
 855                     'which is currently not possible')
 856                 continue
 857             old_file = sub['filepath']
 858             sub_filenames.append(old_file)
 859             new_file = replace_extension(old_file, new_ext)
 860
 861             if ext in ('dfxp', 'ttml', 'tt'):
 862                 self.report_warning(
 863                     'You have requested to convert dfxp (TTML) subtitles into another format, '
 864                     'which results in style information loss')
 865
 866                 dfxp_file = old_file
 867                 srt_file = replace_extension(old_file, 'srt')
 868
 869                 with open(dfxp_file, 'rb') as f:
 870                     srt_data = dfxp2srt(f.read())
 871
 872                 with io.open(srt_file, 'wt', encoding='utf-8') as f:
 873                     f.write(srt_data)
 874                 old_file = srt_file
 875
 876                 subs[lang] = {
 877                     'ext': 'srt',
 878                     'data': srt_data,
 879                     'filepath': srt_file,
 880                 }
 881
 882                 if new_ext == 'srt':
 883                     continue
 884                 else:
 885                     sub_filenames.append(srt_file)
 886
 887             self.run_ffmpeg(old_file, new_file, ['-f', new_format])
 888
 889             with io.open(new_file, 'rt', encoding='utf-8') as f:
 890                 subs[lang] = {
 891                     'ext': new_ext,
 892                     'data': f.read(),
 893                     'filepath': new_file,
 894                 }
 895
 896             info['__files_to_move'][new_file] = replace_extension(
 897                 info['__files_to_move'][sub['filepath']], new_ext)
 898
 899         return sub_filenames, info
 900
 901
 902 class FFmpegSplitChaptersPP(FFmpegPostProcessor):
 903     def __init__(self, downloader, force_keyframes=False):
 904         FFmpegPostProcessor.__init__(self, downloader)
 905         self._force_keyframes = force_keyframes
 906
 907     def _prepare_filename(self, number, chapter, info):
 908         info = info.copy()
 909         info.update({
 910             'section_number': number,
 911             'section_title': chapter.get('title'),
 912             'section_start': chapter.get('start_time'),
 913             'section_end': chapter.get('end_time'),
 914         })
 915         return self._downloader.prepare_filename(info, 'chapter')
 916
 917     def _ffmpeg_args_for_chapter(self, number, chapter, info):
 918         destination = self._prepare_filename(number, chapter, info)
 919         if not self._downloader._ensure_dir_exists(encodeFilename(destination)):
 920             return
 921
 922         chapter['filepath'] = destination
 923         self.to_screen('Chapter %03d; Destination: %s' % (number, destination))
 924         return (
 925             destination,
 926             ['-ss', compat_str(chapter['start_time']),
 927              '-t', compat_str(chapter['end_time'] - chapter['start_time'])])
 928
 929     @PostProcessor._restrict_to(images=False)
 930     def run(self, info):
 931         chapters = info.get('chapters') or []
 932         if not chapters:
 933             self.to_screen('Chapter information is unavailable')
 934             return [], info
 935
 936         in_file = info['filepath']
 937         if self._force_keyframes and len(chapters) > 1:
 938             in_file = self.force_keyframes(in_file, (c['start_time'] for c in chapters))
 939         self.to_screen('Splitting video by chapters; %d chapters found' % len(chapters))
 940         for idx, chapter in enumerate(chapters):
 941             destination, opts = self._ffmpeg_args_for_chapter(idx + 1, chapter, info)
 942             self.real_run_ffmpeg([(in_file, opts)], [(destination, ['-c', 'copy'])])
 943         if in_file != info['filepath']:
 944             os.remove(in_file)
 945         return [], info
 946
 947
 948 class FFmpegThumbnailsConvertorPP(FFmpegPostProcessor):
 949     SUPPORTED_EXTS = ('jpg', 'png')
 950
 951     def __init__(self, downloader=None, format=None):
 952         super(FFmpegThumbnailsConvertorPP, self).__init__(downloader)
 953         self.format = format
 954
 955     @staticmethod
 956     def is_webp(path):
 957         with open(encodeFilename(path), 'rb') as f:
 958             b = f.read(12)
 959         return b[0:4] == b'RIFF' and b[8:] == b'WEBP'
 960
 961     def fixup_webp(self, info, idx=-1):
 962         thumbnail_filename = info['thumbnails'][idx]['filepath']
 963         _, thumbnail_ext = os.path.splitext(thumbnail_filename)
 964         if thumbnail_ext:
 965             thumbnail_ext = thumbnail_ext[1:].lower()
 966             if thumbnail_ext != 'webp' and self.is_webp(thumbnail_filename):
 967                 self.to_screen('Correcting thumbnail "%s" extension to webp' % thumbnail_filename)
 968                 webp_filename = replace_extension(thumbnail_filename, 'webp')
 969                 os.replace(thumbnail_filename, webp_filename)
 970                 info['thumbnails'][idx]['filepath'] = webp_filename
 971                 info['__files_to_move'][webp_filename] = replace_extension(
 972                     info['__files_to_move'].pop(thumbnail_filename), 'webp')
 973
 974     @staticmethod
 975     def _options(target_ext):
 976         if target_ext == 'jpg':
 977             return ['-bsf:v', 'mjpeg2jpeg']
 978         return []
 979
 980     def convert_thumbnail(self, thumbnail_filename, target_ext):
 981         thumbnail_conv_filename = replace_extension(thumbnail_filename, target_ext)
 982
 983         self.to_screen('Converting thumbnail "%s" to %s' % (thumbnail_filename, target_ext))
 984         self.real_run_ffmpeg(
 985             [(thumbnail_filename, ['-f', 'image2', '-pattern_type', 'none'])],
 986             [(thumbnail_conv_filename.replace('%', '%%'), self._options(target_ext))])
 987         return thumbnail_conv_filename
 988
 989     def run(self, info):
 990         files_to_delete = []
 991         has_thumbnail = False
 992
 993         for idx, thumbnail_dict in enumerate(info['thumbnails']):
 994             if 'filepath' not in thumbnail_dict:
 995                 continue
 996             has_thumbnail = True
 997             self.fixup_webp(info, idx)
 998             original_thumbnail = thumbnail_dict['filepath']
 999             _, thumbnail_ext = os.path.splitext(original_thumbnail)
1000             if thumbnail_ext:
1001                 thumbnail_ext = thumbnail_ext[1:].lower()
1002             if thumbnail_ext == 'jpeg':
1003                 thumbnail_ext = 'jpg'
1004             if thumbnail_ext == self.format:
1005                 self.to_screen('Thumbnail "%s" is already in the requested format' % original_thumbnail)
1006                 continue
1007             thumbnail_dict['filepath'] = self.convert_thumbnail(original_thumbnail, self.format)
1008             files_to_delete.append(original_thumbnail)
1009             info['__files_to_move'][thumbnail_dict['filepath']] = replace_extension(
1010                 info['__files_to_move'][original_thumbnail], self.format)
1011
1012         if not has_thumbnail:
1013             self.to_screen('There aren\'t any thumbnails to convert')
1014         return files_to_delete, info