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
  14 from ..utils import (
  15     dfxp2srt,
  16     encodeArgument,
  17     encodeFilename,
  18     float_or_none,
  19     _get_exe_version_output,
  20     detect_exe_version,
  21     is_outdated_version,
  22     ISO639Utils,
  23     orderedSet,
  24     Popen,
  25     PostProcessingError,
  26     prepend_extension,
  27     replace_extension,
  28     shell_quote,
  29     traverse_obj,
  30     variadic,
  31     write_json_file,
  32 )
  33
  34
  35 EXT_TO_OUT_FORMATS = {
  36     'aac': 'adts',
  37     'flac': 'flac',
  38     'm4a': 'ipod',
  39     'mka': 'matroska',
  40     'mkv': 'matroska',
  41     'mpg': 'mpeg',
  42     'ogv': 'ogg',
  43     'ts': 'mpegts',
  44     'wma': 'asf',
  45     'wmv': 'asf',
  46     'vtt': 'webvtt',
  47 }
  48 ACODECS = {
  49     'mp3': 'libmp3lame',
  50     'aac': 'aac',
  51     'flac': 'flac',
  52     'm4a': 'aac',
  53     'opus': 'libopus',
  54     'vorbis': 'libvorbis',
  55     'wav': None,
  56     'alac': None,
  57 }
  58
  59
  60 class FFmpegPostProcessorError(PostProcessingError):
  61     pass
  62
  63
  64 class FFmpegPostProcessor(PostProcessor):
  65     def __init__(self, downloader=None):
  66         PostProcessor.__init__(self, downloader)
  67         self._determine_executables()
  68
  69     def check_version(self):
  70         if not self.available:
  71             raise FFmpegPostProcessorError('ffmpeg not found. Please install or provide the path using --ffmpeg-location')
  72
  73         required_version = '10-0' if self.basename == 'avconv' else '1.0'
  74         if is_outdated_version(
  75                 self._versions[self.basename], required_version):
  76             warning = 'Your copy of %s is outdated, update %s to version %s or newer if you encounter any errors.' % (
  77                 self.basename, self.basename, required_version)
  78             self.report_warning(warning)
  79
  80     @staticmethod
  81     def get_versions_and_features(downloader=None):
  82         pp = FFmpegPostProcessor(downloader)
  83         return pp._versions, pp._features
  84
  85     @staticmethod
  86     def get_versions(downloader=None):
  87         return FFmpegPostProcessor.get_version_and_features(downloader)[0]
  88
  89     def _determine_executables(self):
  90         programs = ['avprobe', 'avconv', 'ffmpeg', 'ffprobe']
  91
  92         def get_ffmpeg_version(path, prog):
  93             out = _get_exe_version_output(path, ['-bsfs'])
  94             ver = detect_exe_version(out) if out else False
  95             if ver:
  96                 regexs = [
  97                     r'(?:\d+:)?([0-9.]+)-[0-9]+ubuntu[0-9.]+$',  # Ubuntu, see [1]
  98                     r'n([0-9.]+)$',  # Arch Linux
  99                     # 1. http://www.ducea.com/2006/06/17/ubuntu-package-version-naming-explanation/
 100                 ]
 101                 for regex in regexs:
 102                     mobj = re.match(regex, ver)
 103                     if mobj:
 104                         ver = mobj.group(1)
 105             self._versions[prog] = ver
 106             if prog != 'ffmpeg' or not out:
 107                 return
 108
 109             mobj = re.search(r'(?m)^\s+libavformat\s+(?:[0-9. ]+)\s+/\s+(?P<runtime>[0-9. ]+)', out)
 110             lavf_runtime_version = mobj.group('runtime').replace(' ', '') if mobj else None
 111             self._features = {
 112                 'fdk': '--enable-libfdk-aac' in out,
 113                 'setts': 'setts' in out.splitlines(),
 114                 'needs_adtstoasc': is_outdated_version(lavf_runtime_version, '57.56.100', False),
 115             }
 116
 117         self.basename = None
 118         self.probe_basename = None
 119         self._paths = None
 120         self._versions = None
 121         self._features = {}
 122
 123         prefer_ffmpeg = self.get_param('prefer_ffmpeg', True)
 124         location = self.get_param('ffmpeg_location')
 125         if location is None:
 126             self._paths = {p: p for p in programs}
 127         else:
 128             if not os.path.exists(location):
 129                 self.report_warning(
 130                     'ffmpeg-location %s does not exist! '
 131                     'Continuing without ffmpeg.' % (location))
 132                 self._versions = {}
 133                 return
 134             elif os.path.isdir(location):
 135                 dirname, basename = location, None
 136             else:
 137                 basename = os.path.splitext(os.path.basename(location))[0]
 138                 basename = next((p for p in programs if basename.startswith(p)), 'ffmpeg')
 139                 dirname = os.path.dirname(os.path.abspath(location))
 140                 if basename in ('ffmpeg', 'ffprobe'):
 141                     prefer_ffmpeg = True
 142
 143             self._paths = dict(
 144                 (p, os.path.join(dirname, p)) for p in programs)
 145             if basename:
 146                 self._paths[basename] = location
 147
 148         self._versions = {}
 149         for p in programs:
 150             get_ffmpeg_version(self._paths[p], p)
 151
 152         if prefer_ffmpeg is False:
 153             prefs = ('avconv', 'ffmpeg')
 154         else:
 155             prefs = ('ffmpeg', 'avconv')
 156         for p in prefs:
 157             if self._versions[p]:
 158                 self.basename = p
 159                 break
 160
 161         if prefer_ffmpeg is False:
 162             prefs = ('avprobe', 'ffprobe')
 163         else:
 164             prefs = ('ffprobe', 'avprobe')
 165         for p in prefs:
 166             if self._versions[p]:
 167                 self.probe_basename = p
 168                 break
 169
 170         if self.basename == 'avconv':
 171             self.deprecation_warning(
 172                 'Support for avconv is deprecated and may be removed in a future version. Use ffmpeg instead')
 173         if self.probe_basename == 'avprobe':
 174             self.deprecation_warning(
 175                 'Support for avprobe is deprecated and may be removed in a future version. Use ffprobe instead')
 176
 177     @property
 178     def available(self):
 179         return self.basename is not None
 180
 181     @property
 182     def executable(self):
 183         return self._paths[self.basename]
 184
 185     @property
 186     def probe_available(self):
 187         return self.probe_basename is not None
 188
 189     @property
 190     def probe_executable(self):
 191         return self._paths[self.probe_basename]
 192
 193     def get_audio_codec(self, path):
 194         if not self.probe_available and not self.available:
 195             raise PostProcessingError('ffprobe and ffmpeg not found. Please install or provide the path using --ffmpeg-location')
 196         try:
 197             if self.probe_available:
 198                 cmd = [
 199                     encodeFilename(self.probe_executable, True),
 200                     encodeArgument('-show_streams')]
 201             else:
 202                 cmd = [
 203                     encodeFilename(self.executable, True),
 204                     encodeArgument('-i')]
 205             cmd.append(encodeFilename(self._ffmpeg_filename_argument(path), True))
 206             self.write_debug('%s command line: %s' % (self.basename, shell_quote(cmd)))
 207             handle = Popen(cmd, stdin=subprocess.PIPE, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
 208             stdout_data, stderr_data = handle.communicate_or_kill()
 209             expected_ret = 0 if self.probe_available else 1
 210             if handle.wait() != expected_ret:
 211                 return None
 212         except (IOError, OSError):
 213             return None
 214         output = (stdout_data if self.probe_available else stderr_data).decode('ascii', 'ignore')
 215         if self.probe_available:
 216             audio_codec = None
 217             for line in output.split('\n'):
 218                 if line.startswith('codec_name='):
 219                     audio_codec = line.split('=')[1].strip()
 220                 elif line.strip() == 'codec_type=audio' and audio_codec is not None:
 221                     return audio_codec
 222         else:
 223             # Stream #FILE_INDEX:STREAM_INDEX[STREAM_ID](LANGUAGE): CODEC_TYPE: CODEC_NAME
 224             mobj = re.search(
 225                 r'Stream\s*#\d+:\d+(?:\[0x[0-9a-f]+\])?(?:\([a-z]{3}\))?:\s*Audio:\s*([0-9a-z]+)',
 226                 output)
 227             if mobj:
 228                 return mobj.group(1)
 229         return None
 230
 231     def get_metadata_object(self, path, opts=[]):
 232         if self.probe_basename != 'ffprobe':
 233             if self.probe_available:
 234                 self.report_warning('Only ffprobe is supported for metadata extraction')
 235             raise PostProcessingError('ffprobe not found. Please install or provide the path using --ffmpeg-location')
 236         self.check_version()
 237
 238         cmd = [
 239             encodeFilename(self.probe_executable, True),
 240             encodeArgument('-hide_banner'),
 241             encodeArgument('-show_format'),
 242             encodeArgument('-show_streams'),
 243             encodeArgument('-print_format'),
 244             encodeArgument('json'),
 245         ]
 246
 247         cmd += opts
 248         cmd.append(encodeFilename(self._ffmpeg_filename_argument(path), True))
 249         self.write_debug('ffprobe command line: %s' % shell_quote(cmd))
 250         p = Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE, stdin=subprocess.PIPE)
 251         stdout, stderr = p.communicate()
 252         return json.loads(stdout.decode('utf-8', 'replace'))
 253
 254     def get_stream_number(self, path, keys, value):
 255         streams = self.get_metadata_object(path)['streams']
 256         num = next(
 257             (i for i, stream in enumerate(streams) if traverse_obj(stream, keys, casesense=False) == value),
 258             None)
 259         return num, len(streams)
 260
 261     def _get_real_video_duration(self, filepath, fatal=True):
 262         try:
 263             duration = float_or_none(
 264                 traverse_obj(self.get_metadata_object(filepath), ('format', 'duration')))
 265             if not duration:
 266                 raise PostProcessingError('ffprobe returned empty duration')
 267             return duration
 268         except PostProcessingError as e:
 269             if fatal:
 270                 raise PostProcessingError(f'Unable to determine video duration: {e.msg}')
 271
 272     def _duration_mismatch(self, d1, d2):
 273         if not d1 or not d2:
 274             return None
 275         # The duration is often only known to nearest second. So there can be <1sec disparity natually.
 276         # Further excuse an additional <1sec difference.
 277         return abs(d1 - d2) > 2
 278
 279     def run_ffmpeg_multiple_files(self, input_paths, out_path, opts, **kwargs):
 280         return self.real_run_ffmpeg(
 281             [(path, []) for path in input_paths],
 282             [(out_path, opts)], **kwargs)
 283
 284     def real_run_ffmpeg(self, input_path_opts, output_path_opts, *, expected_retcodes=(0,)):
 285         self.check_version()
 286
 287         oldest_mtime = min(
 288             os.stat(encodeFilename(path)).st_mtime for path, _ in input_path_opts if path)
 289
 290         cmd = [encodeFilename(self.executable, True), encodeArgument('-y')]
 291         # avconv does not have repeat option
 292         if self.basename == 'ffmpeg':
 293             cmd += [encodeArgument('-loglevel'), encodeArgument('repeat+info')]
 294
 295         def make_args(file, args, name, number):
 296             keys = ['_%s%d' % (name, number), '_%s' % name]
 297             if name == 'o':
 298                 args += ['-movflags', '+faststart']
 299             elif number == 1:
 300                 keys.append('')
 301             args += self._configuration_args(self.basename, keys)
 302             if name == 'i':
 303                 args.append('-i')
 304             return (
 305                 [encodeArgument(arg) for arg in args]
 306                 + [encodeFilename(self._ffmpeg_filename_argument(file), True)])
 307
 308         for arg_type, path_opts in (('i', input_path_opts), ('o', output_path_opts)):
 309             cmd += itertools.chain.from_iterable(
 310                 make_args(path, list(opts), arg_type, i + 1)
 311                 for i, (path, opts) in enumerate(path_opts) if path)
 312
 313         self.write_debug('ffmpeg command line: %s' % shell_quote(cmd))
 314         p = Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE, stdin=subprocess.PIPE)
 315         stdout, stderr = p.communicate_or_kill()
 316         if p.returncode not in variadic(expected_retcodes):
 317             stderr = stderr.decode('utf-8', 'replace').strip()
 318             self.write_debug(stderr)
 319             raise FFmpegPostProcessorError(stderr.split('\n')[-1])
 320         for out_path, _ in output_path_opts:
 321             if out_path:
 322                 self.try_utime(out_path, oldest_mtime, oldest_mtime)
 323         return stderr.decode('utf-8', 'replace')
 324
 325     def run_ffmpeg(self, path, out_path, opts, **kwargs):
 326         return self.run_ffmpeg_multiple_files([path], out_path, opts, **kwargs)
 327
 328     @staticmethod
 329     def _ffmpeg_filename_argument(fn):
 330         # Always use 'file:' because the filename may contain ':' (ffmpeg
 331         # interprets that as a protocol) or can start with '-' (-- is broken in
 332         # ffmpeg, see https://ffmpeg.org/trac/ffmpeg/ticket/2127 for details)
 333         # Also leave '-' intact in order not to break streaming to stdout.
 334         if fn.startswith(('http://', 'https://')):
 335             return fn
 336         return 'file:' + fn if fn != '-' else fn
 337
 338     @staticmethod
 339     def _quote_for_ffmpeg(string):
 340         # See https://ffmpeg.org/ffmpeg-utils.html#toc-Quoting-and-escaping
 341         # A sequence of '' produces '\'''\'';
 342         # final replace removes the empty '' between \' \'.
 343         string = string.replace("'", r"'\''").replace("'''", "'")
 344         # Handle potential ' at string boundaries.
 345         string = string[1:] if string[0] == "'" else "'" + string
 346         return string[:-1] if string[-1] == "'" else string + "'"
 347
 348     def force_keyframes(self, filename, timestamps):
 349         timestamps = orderedSet(timestamps)
 350         if timestamps[0] == 0:
 351             timestamps = timestamps[1:]
 352         keyframe_file = prepend_extension(filename, 'keyframes.temp')
 353         self.to_screen(f'Re-encoding "{filename}" with appropriate keyframes')
 354         self.run_ffmpeg(filename, keyframe_file, ['-force_key_frames', ','.join(
 355             f'{t:.6f}' for t in timestamps)])
 356         return keyframe_file
 357
 358     def concat_files(self, in_files, out_file, concat_opts=None):
 359         """
 360         Use concat demuxer to concatenate multiple files having identical streams.
 361
 362         Only inpoint, outpoint, and duration concat options are supported.
 363         See https://ffmpeg.org/ffmpeg-formats.html#concat-1 for details
 364         """
 365         concat_file = f'{out_file}.concat'
 366         self.write_debug(f'Writing concat spec to {concat_file}')
 367         with open(concat_file, 'wt', encoding='utf-8') as f:
 368             f.writelines(self._concat_spec(in_files, concat_opts))
 369
 370         out_flags = ['-c', 'copy']
 371         if out_file.rpartition('.')[-1] in ('mp4', 'mov'):
 372             # For some reason, '-c copy' is not enough to copy subtitles
 373             out_flags.extend(['-c:s', 'mov_text'])
 374
 375         try:
 376             self.real_run_ffmpeg(
 377                 [(concat_file, ['-hide_banner', '-nostdin', '-f', 'concat', '-safe', '0'])],
 378                 [(out_file, out_flags)])
 379         finally:
 380             os.remove(concat_file)
 381
 382     @classmethod
 383     def _concat_spec(cls, in_files, concat_opts=None):
 384         if concat_opts is None:
 385             concat_opts = [{}] * len(in_files)
 386         yield 'ffconcat version 1.0\n'
 387         for file, opts in zip(in_files, concat_opts):
 388             yield f'file {cls._quote_for_ffmpeg(cls._ffmpeg_filename_argument(file))}\n'
 389             # Iterate explicitly to yield the following directives in order, ignoring the rest.
 390             for directive in 'inpoint', 'outpoint', 'duration':
 391                 if directive in opts:
 392                     yield f'{directive} {opts[directive]}\n'
 393
 394
 395 class FFmpegExtractAudioPP(FFmpegPostProcessor):
 396     COMMON_AUDIO_EXTS = ('wav', 'flac', 'm4a', 'aiff', 'mp3', 'ogg', 'mka', 'opus', 'wma')
 397     SUPPORTED_EXTS = ('best', 'aac', 'flac', 'mp3', 'm4a', 'opus', 'vorbis', 'wav', 'alac')
 398
 399     def __init__(self, downloader=None, preferredcodec=None, preferredquality=None, nopostoverwrites=False):
 400         FFmpegPostProcessor.__init__(self, downloader)
 401         self._preferredcodec = preferredcodec or 'best'
 402         self._preferredquality = float_or_none(preferredquality)
 403         self._nopostoverwrites = nopostoverwrites
 404
 405     def _quality_args(self, codec):
 406         if self._preferredquality is None:
 407             return []
 408         elif self._preferredquality > 10:
 409             return ['-b:a', f'{self._preferredquality}k']
 410
 411         limits = {
 412             'libmp3lame': (10, 0),
 413             'libvorbis': (0, 10),
 414             # FFmpeg's AAC encoder does not have an upper limit for the value of -q:a.
 415             # Experimentally, with values over 4, bitrate changes were minimal or non-existent
 416             'aac': (0.1, 4),
 417             'libfdk_aac': (1, 5),
 418         }.get(codec)
 419         if not limits:
 420             return []
 421
 422         q = limits[1] + (limits[0] - limits[1]) * (self._preferredquality / 10)
 423         if codec == 'libfdk_aac':
 424             return ['-vbr', f'{int(q)}']
 425         return ['-q:a', f'{q}']
 426
 427     def run_ffmpeg(self, path, out_path, codec, more_opts):
 428         if codec is None:
 429             acodec_opts = []
 430         else:
 431             acodec_opts = ['-acodec', codec]
 432         opts = ['-vn'] + acodec_opts + more_opts
 433         try:
 434             FFmpegPostProcessor.run_ffmpeg(self, path, out_path, opts)
 435         except FFmpegPostProcessorError as err:
 436             raise AudioConversionError(err.msg)
 437
 438     @PostProcessor._restrict_to(images=False)
 439     def run(self, information):
 440         orig_path = path = information['filepath']
 441         orig_ext = information['ext']
 442
 443         if self._preferredcodec == 'best' and orig_ext in self.COMMON_AUDIO_EXTS:
 444             self.to_screen('Skipping audio extraction since the file is already in a common audio format')
 445             return [], information
 446
 447         filecodec = self.get_audio_codec(path)
 448         if filecodec is None:
 449             raise PostProcessingError('WARNING: unable to obtain file audio codec with ffprobe')
 450
 451         more_opts = []
 452         if self._preferredcodec == 'best' or self._preferredcodec == filecodec or (self._preferredcodec == 'm4a' and filecodec == 'aac'):
 453             if filecodec == 'aac' and self._preferredcodec in ['m4a', 'best']:
 454                 # Lossless, but in another container
 455                 acodec = 'copy'
 456                 extension = 'm4a'
 457                 more_opts = ['-bsf:a', 'aac_adtstoasc']
 458             elif filecodec in ['aac', 'flac', 'mp3', 'vorbis', 'opus']:
 459                 # Lossless if possible
 460                 acodec = 'copy'
 461                 extension = filecodec
 462                 if filecodec == 'aac':
 463                     more_opts = ['-f', 'adts']
 464                 if filecodec == 'vorbis':
 465                     extension = 'ogg'
 466             elif filecodec == 'alac':
 467                 acodec = None
 468                 extension = 'm4a'
 469                 more_opts += ['-acodec', 'alac']
 470             else:
 471                 # MP3 otherwise.
 472                 acodec = 'libmp3lame'
 473                 extension = 'mp3'
 474                 more_opts = self._quality_args(acodec)
 475         else:
 476             # We convert the audio (lossy if codec is lossy)
 477             acodec = ACODECS[self._preferredcodec]
 478             if acodec == 'aac' and self._features.get('fdk'):
 479                 acodec = 'libfdk_aac'
 480             extension = self._preferredcodec
 481             more_opts = self._quality_args(acodec)
 482             if self._preferredcodec == 'aac':
 483                 more_opts += ['-f', 'adts']
 484             elif self._preferredcodec == 'm4a':
 485                 more_opts += ['-bsf:a', 'aac_adtstoasc']
 486             elif self._preferredcodec == 'vorbis':
 487                 extension = 'ogg'
 488             elif self._preferredcodec == 'wav':
 489                 extension = 'wav'
 490                 more_opts += ['-f', 'wav']
 491             elif self._preferredcodec == 'alac':
 492                 extension = 'm4a'
 493                 more_opts += ['-acodec', 'alac']
 494
 495         prefix, sep, ext = path.rpartition('.')  # not os.path.splitext, since the latter does not work on unicode in all setups
 496         temp_path = new_path = prefix + sep + extension
 497
 498         if new_path == path:
 499             orig_path = prepend_extension(path, 'orig')
 500             temp_path = prepend_extension(path, 'temp')
 501         if (self._nopostoverwrites and os.path.exists(encodeFilename(new_path))
 502                 and os.path.exists(encodeFilename(orig_path))):
 503             self.to_screen('Post-process file %s exists, skipping' % new_path)
 504             return [], information
 505
 506         try:
 507             self.to_screen(f'Destination: {new_path}')
 508             self.run_ffmpeg(path, temp_path, acodec, more_opts)
 509         except AudioConversionError as e:
 510             raise PostProcessingError(
 511                 'audio conversion failed: ' + e.msg)
 512         except Exception:
 513             raise PostProcessingError('error running ' + self.basename)
 514
 515         os.replace(path, orig_path)
 516         os.replace(temp_path, new_path)
 517         information['filepath'] = new_path
 518         information['ext'] = extension
 519
 520         # Try to update the date time for extracted audio file.
 521         if information.get('filetime') is not None:
 522             self.try_utime(
 523                 new_path, time.time(), information['filetime'],
 524                 errnote='Cannot update utime of audio file')
 525
 526         return [orig_path], information
 527
 528
 529 class FFmpegVideoConvertorPP(FFmpegPostProcessor):
 530     SUPPORTED_EXTS = ('mp4', 'mkv', 'flv', 'webm', 'mov', 'avi', 'mp3', 'mka', 'm4a', 'ogg', 'opus')
 531     FORMAT_RE = re.compile(r'{0}(?:/{0})*$'.format(r'(?:\w+>)?(?:%s)' % '|'.join(SUPPORTED_EXTS)))
 532     _ACTION = 'converting'
 533
 534     def __init__(self, downloader=None, preferedformat=None):
 535         super(FFmpegVideoConvertorPP, self).__init__(downloader)
 536         self._preferedformats = preferedformat.lower().split('/')
 537
 538     def _target_ext(self, source_ext):
 539         for pair in self._preferedformats:
 540             kv = pair.split('>')
 541             if len(kv) == 1 or kv[0].strip() == source_ext:
 542                 return kv[-1].strip()
 543
 544     @staticmethod
 545     def _options(target_ext):
 546         if target_ext == 'avi':
 547             return ['-c:v', 'libxvid', '-vtag', 'XVID']
 548         return []
 549
 550     @PostProcessor._restrict_to(images=False)
 551     def run(self, info):
 552         filename, source_ext = info['filepath'], info['ext'].lower()
 553         target_ext = self._target_ext(source_ext)
 554         _skip_msg = (
 555             f'could not find a mapping for {source_ext}' if not target_ext
 556             else f'already is in target format {source_ext}' if source_ext == target_ext
 557             else None)
 558         if _skip_msg:
 559             self.to_screen(f'Not {self._ACTION} media file {filename!r}; {_skip_msg}')
 560             return [], info
 561
 562         outpath = replace_extension(filename, target_ext, source_ext)
 563         self.to_screen(f'{self._ACTION.title()} video from {source_ext} to {target_ext}; Destination: {outpath}')
 564         self.run_ffmpeg(filename, outpath, self._options(target_ext))
 565
 566         info['filepath'] = outpath
 567         info['format'] = info['ext'] = target_ext
 568         return [filename], info
 569
 570
 571 class FFmpegVideoRemuxerPP(FFmpegVideoConvertorPP):
 572     _ACTION = 'remuxing'
 573
 574     @staticmethod
 575     def _options(target_ext):
 576         return ['-c', 'copy', '-map', '0', '-dn']
 577
 578
 579 class FFmpegEmbedSubtitlePP(FFmpegPostProcessor):
 580     def __init__(self, downloader=None, already_have_subtitle=False):
 581         super(FFmpegEmbedSubtitlePP, self).__init__(downloader)
 582         self._already_have_subtitle = already_have_subtitle
 583
 584     @PostProcessor._restrict_to(images=False)
 585     def run(self, info):
 586         if info['ext'] not in ('mp4', 'webm', 'mkv'):
 587             self.to_screen('Subtitles can only be embedded in mp4, webm or mkv files')
 588             return [], info
 589         subtitles = info.get('requested_subtitles')
 590         if not subtitles:
 591             self.to_screen('There aren\'t any subtitles to embed')
 592             return [], info
 593
 594         filename = info['filepath']
 595
 596         # Disabled temporarily. There needs to be a way to overide this
 597         # in case of duration actually mismatching in extractor
 598         # See: https://github.com/yt-dlp/yt-dlp/issues/1870, https://github.com/yt-dlp/yt-dlp/issues/1385
 599         '''
 600         if info.get('duration') and not info.get('__real_download') and self._duration_mismatch(
 601                 self._get_real_video_duration(filename, False), info['duration']):
 602             self.to_screen(f'Skipping {self.pp_key()} since the real and expected durations mismatch')
 603             return [], info
 604         '''
 605
 606         ext = info['ext']
 607         sub_langs, sub_names, sub_filenames = [], [], []
 608         webm_vtt_warn = False
 609         mp4_ass_warn = False
 610
 611         for lang, sub_info in subtitles.items():
 612             if not os.path.exists(sub_info.get('filepath', '')):
 613                 self.report_warning(f'Skipping embedding {lang} subtitle because the file is missing')
 614                 continue
 615             sub_ext = sub_info['ext']
 616             if sub_ext == 'json':
 617                 self.report_warning('JSON subtitles cannot be embedded')
 618             elif ext != 'webm' or ext == 'webm' and sub_ext == 'vtt':
 619                 sub_langs.append(lang)
 620                 sub_names.append(sub_info.get('name'))
 621                 sub_filenames.append(sub_info['filepath'])
 622             else:
 623                 if not webm_vtt_warn and ext == 'webm' and sub_ext != 'vtt':
 624                     webm_vtt_warn = True
 625                     self.report_warning('Only WebVTT subtitles can be embedded in webm files')
 626             if not mp4_ass_warn and ext == 'mp4' and sub_ext == 'ass':
 627                 mp4_ass_warn = True
 628                 self.report_warning('ASS subtitles cannot be properly embedded in mp4 files; expect issues')
 629
 630         if not sub_langs:
 631             return [], info
 632
 633         input_files = [filename] + sub_filenames
 634
 635         opts = [
 636             '-c', 'copy', '-map', '0', '-dn',
 637             # Don't copy the existing subtitles, we may be running the
 638             # postprocessor a second time
 639             '-map', '-0:s',
 640             # Don't copy Apple TV chapters track, bin_data (see #19042, #19024,
 641             # https://trac.ffmpeg.org/ticket/6016)
 642             '-map', '-0:d',
 643         ]
 644         if info['ext'] == 'mp4':
 645             opts += ['-c:s', 'mov_text']
 646         for i, (lang, name) in enumerate(zip(sub_langs, sub_names)):
 647             opts.extend(['-map', '%d:0' % (i + 1)])
 648             lang_code = ISO639Utils.short2long(lang) or lang
 649             opts.extend(['-metadata:s:s:%d' % i, 'language=%s' % lang_code])
 650             if name:
 651                 opts.extend(['-metadata:s:s:%d' % i, 'handler_name=%s' % name,
 652                              '-metadata:s:s:%d' % i, 'title=%s' % name])
 653
 654         temp_filename = prepend_extension(filename, 'temp')
 655         self.to_screen('Embedding subtitles in "%s"' % filename)
 656         self.run_ffmpeg_multiple_files(input_files, temp_filename, opts)
 657         os.replace(temp_filename, filename)
 658
 659         files_to_delete = [] if self._already_have_subtitle else sub_filenames
 660         return files_to_delete, info
 661
 662
 663 class FFmpegMetadataPP(FFmpegPostProcessor):
 664
 665     def __init__(self, downloader, add_metadata=True, add_chapters=True, add_infojson='if_exists'):
 666         FFmpegPostProcessor.__init__(self, downloader)
 667         self._add_metadata = add_metadata
 668         self._add_chapters = add_chapters
 669         self._add_infojson = add_infojson
 670
 671     @staticmethod
 672     def _options(target_ext):
 673         yield from ('-map', '0', '-dn')
 674         if target_ext == 'm4a':
 675             yield from ('-vn', '-acodec', 'copy')
 676         else:
 677             yield from ('-c', 'copy')
 678
 679     @PostProcessor._restrict_to(images=False)
 680     def run(self, info):
 681         filename, metadata_filename = info['filepath'], None
 682         files_to_delete, options = [], []
 683         if self._add_chapters and info.get('chapters'):
 684             metadata_filename = replace_extension(filename, 'meta')
 685             options.extend(self._get_chapter_opts(info['chapters'], metadata_filename))
 686             files_to_delete.append(metadata_filename)
 687         if self._add_metadata:
 688             options.extend(self._get_metadata_opts(info))
 689
 690         if self._add_infojson:
 691             if info['ext'] in ('mkv', 'mka'):
 692                 infojson_filename = info.get('infojson_filename')
 693                 options.extend(self._get_infojson_opts(info, infojson_filename))
 694                 if not infojson_filename:
 695                     files_to_delete.append(info.get('infojson_filename'))
 696             elif self._add_infojson is True:
 697                 self.to_screen('The info-json can only be attached to mkv/mka files')
 698
 699         if not options:
 700             self.to_screen('There isn\'t any metadata to add')
 701             return [], info
 702
 703         temp_filename = prepend_extension(filename, 'temp')
 704         self.to_screen('Adding metadata to "%s"' % filename)
 705         self.run_ffmpeg_multiple_files(
 706             (filename, metadata_filename), temp_filename,
 707             itertools.chain(self._options(info['ext']), *options))
 708         for file in filter(None, files_to_delete):
 709             os.remove(file)  # Don't obey --keep-files
 710         os.replace(temp_filename, filename)
 711         return [], info
 712
 713     @staticmethod
 714     def _get_chapter_opts(chapters, metadata_filename):
 715         with io.open(metadata_filename, 'wt', encoding='utf-8') as f:
 716             def ffmpeg_escape(text):
 717                 return re.sub(r'([\\=;#\n])', r'\\\1', text)
 718
 719             metadata_file_content = ';FFMETADATA1\n'
 720             for chapter in chapters:
 721                 metadata_file_content += '[CHAPTER]\nTIMEBASE=1/1000\n'
 722                 metadata_file_content += 'START=%d\n' % (chapter['start_time'] * 1000)
 723                 metadata_file_content += 'END=%d\n' % (chapter['end_time'] * 1000)
 724                 chapter_title = chapter.get('title')
 725                 if chapter_title:
 726                     metadata_file_content += 'title=%s\n' % ffmpeg_escape(chapter_title)
 727             f.write(metadata_file_content)
 728         yield ('-map_metadata', '1')
 729
 730     def _get_metadata_opts(self, info):
 731         metadata = {}
 732         meta_prefix = 'meta_'
 733
 734         def add(meta_list, info_list=None):
 735             value = next((
 736                 str(info[key]) for key in [meta_prefix] + list(variadic(info_list or meta_list))
 737                 if info.get(key) is not None), None)
 738             if value not in ('', None):
 739                 metadata.update({meta_f: value for meta_f in variadic(meta_list)})
 740
 741         # See [1-4] for some info on media metadata/metadata supported
 742         # by ffmpeg.
 743         # 1. https://kdenlive.org/en/project/adding-meta-data-to-mp4-video/
 744         # 2. https://wiki.multimedia.cx/index.php/FFmpeg_Metadata
 745         # 3. https://kodi.wiki/view/Video_file_tagging
 746
 747         add('title', ('track', 'title'))
 748         add('date', 'upload_date')
 749         add(('description', 'synopsis'), 'description')
 750         add(('purl', 'comment'), 'webpage_url')
 751         add('track', 'track_number')
 752         add('artist', ('artist', 'creator', 'uploader', 'uploader_id'))
 753         add('genre')
 754         add('album')
 755         add('album_artist')
 756         add('disc', 'disc_number')
 757         add('show', 'series')
 758         add('season_number')
 759         add('episode_id', ('episode', 'episode_id'))
 760         add('episode_sort', 'episode_number')
 761         if 'embed-metadata' in self.get_param('compat_opts', []):
 762             add('comment', 'description')
 763             metadata.pop('synopsis', None)
 764
 765         for key, value in info.items():
 766             if value is not None and key != meta_prefix and key.startswith(meta_prefix):
 767                 metadata[key[len(meta_prefix):]] = value
 768
 769         for name, value in metadata.items():
 770             yield ('-metadata', f'{name}={value}')
 771
 772         stream_idx = 0
 773         for fmt in info.get('requested_formats') or []:
 774             stream_count = 2 if 'none' not in (fmt.get('vcodec'), fmt.get('acodec')) else 1
 775             if fmt.get('language'):
 776                 lang = ISO639Utils.short2long(fmt['language']) or fmt['language']
 777                 for i in range(stream_count):
 778                     yield ('-metadata:s:%d' % (stream_idx + i), 'language=%s' % lang)
 779             stream_idx += stream_count
 780
 781     def _get_infojson_opts(self, info, infofn):
 782         if not infofn or not os.path.exists(infofn):
 783             if self._add_infojson is not True:
 784                 return
 785             infofn = infofn or '%s.temp' % (
 786                 self._downloader.prepare_filename(info, 'infojson')
 787                 or replace_extension(self._downloader.prepare_filename(info), 'info.json', info['ext']))
 788             if not self._downloader._ensure_dir_exists(infofn):
 789                 return
 790             self.write_debug(f'Writing info-json to: {infofn}')
 791             write_json_file(self._downloader.sanitize_info(info, self.get_param('clean_infojson', True)), infofn)
 792             info['infojson_filename'] = infofn
 793
 794         old_stream, new_stream = self.get_stream_number(info['filepath'], ('tags', 'mimetype'), 'application/json')
 795         if old_stream is not None:
 796             yield ('-map', '-0:%d' % old_stream)
 797             new_stream -= 1
 798
 799         yield ('-attach', infofn,
 800                '-metadata:s:%d' % new_stream, 'mimetype=application/json')
 801
 802
 803 class FFmpegMergerPP(FFmpegPostProcessor):
 804     @PostProcessor._restrict_to(images=False)
 805     def run(self, info):
 806         filename = info['filepath']
 807         temp_filename = prepend_extension(filename, 'temp')
 808         args = ['-c', 'copy']
 809         audio_streams = 0
 810         for (i, fmt) in enumerate(info['requested_formats']):
 811             if fmt.get('acodec') != 'none':
 812                 args.extend(['-map', f'{i}:a:0'])
 813                 aac_fixup = fmt['protocol'].startswith('m3u8') and self.get_audio_codec(fmt['filepath']) == 'aac'
 814                 if aac_fixup:
 815                     args.extend([f'-bsf:a:{audio_streams}', 'aac_adtstoasc'])
 816                 audio_streams += 1
 817             if fmt.get('vcodec') != 'none':
 818                 args.extend(['-map', '%u:v:0' % (i)])
 819         self.to_screen('Merging formats into "%s"' % filename)
 820         self.run_ffmpeg_multiple_files(info['__files_to_merge'], temp_filename, args)
 821         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 822         return info['__files_to_merge'], info
 823
 824     def can_merge(self):
 825         # TODO: figure out merge-capable ffmpeg version
 826         if self.basename != 'avconv':
 827             return True
 828
 829         required_version = '10-0'
 830         if is_outdated_version(
 831                 self._versions[self.basename], required_version):
 832             warning = ('Your copy of %s is outdated and unable to properly mux separate video and audio files, '
 833                        'yt-dlp will download single file media. '
 834                        'Update %s to version %s or newer to fix this.') % (
 835                            self.basename, self.basename, required_version)
 836             self.report_warning(warning)
 837             return False
 838         return True
 839
 840
 841 class FFmpegFixupPostProcessor(FFmpegPostProcessor):
 842     def _fixup(self, msg, filename, options):
 843         temp_filename = prepend_extension(filename, 'temp')
 844
 845         self.to_screen(f'{msg} of "{filename}"')
 846         self.run_ffmpeg(filename, temp_filename, options)
 847
 848         os.replace(temp_filename, filename)
 849
 850
 851 class FFmpegFixupStretchedPP(FFmpegFixupPostProcessor):
 852     @PostProcessor._restrict_to(images=False, audio=False)
 853     def run(self, info):
 854         stretched_ratio = info.get('stretched_ratio')
 855         if stretched_ratio not in (None, 1):
 856             self._fixup('Fixing aspect ratio', info['filepath'], [
 857                 '-c', 'copy', '-map', '0', '-dn', '-aspect', '%f' % stretched_ratio])
 858         return [], info
 859
 860
 861 class FFmpegFixupM4aPP(FFmpegFixupPostProcessor):
 862     @PostProcessor._restrict_to(images=False, video=False)
 863     def run(self, info):
 864         if info.get('container') == 'm4a_dash':
 865             self._fixup('Correcting container', info['filepath'], [
 866                 '-c', 'copy', '-map', '0', '-dn', '-f', 'mp4'])
 867         return [], info
 868
 869
 870 class FFmpegFixupM3u8PP(FFmpegFixupPostProcessor):
 871     def _needs_fixup(self, info):
 872         yield info['ext'] in ('mp4', 'm4a')
 873         yield info['protocol'].startswith('m3u8')
 874         try:
 875             metadata = self.get_metadata_object(info['filepath'])
 876         except PostProcessingError as e:
 877             self.report_warning(f'Unable to extract metadata: {e.msg}')
 878             yield True
 879         else:
 880             yield traverse_obj(metadata, ('format', 'format_name'), casesense=False) == 'mpegts'
 881
 882     @PostProcessor._restrict_to(images=False)
 883     def run(self, info):
 884         if all(self._needs_fixup(info)):
 885             self._fixup('Fixing MPEG-TS in MP4 container', info['filepath'], [
 886                 '-c', 'copy', '-map', '0', '-dn', '-f', 'mp4', '-bsf:a', 'aac_adtstoasc'])
 887         return [], info
 888
 889
 890 class FFmpegFixupTimestampPP(FFmpegFixupPostProcessor):
 891
 892     def __init__(self, downloader=None, trim=0.001):
 893         # "trim" should be used when the video contains unintended packets
 894         super(FFmpegFixupTimestampPP, self).__init__(downloader)
 895         assert isinstance(trim, (int, float))
 896         self.trim = str(trim)
 897
 898     @PostProcessor._restrict_to(images=False)
 899     def run(self, info):
 900         if not self._features.get('setts'):
 901             self.report_warning(
 902                 'A re-encode is needed to fix timestamps in older versions of ffmpeg. '
 903                 'Please install ffmpeg 4.4 or later to fixup without re-encoding')
 904             opts = ['-vf', 'setpts=PTS-STARTPTS']
 905         else:
 906             opts = ['-c', 'copy', '-bsf', 'setts=ts=TS-STARTPTS']
 907         self._fixup('Fixing frame timestamp', info['filepath'], opts + ['-map', '0', '-dn', '-ss', self.trim])
 908         return [], info
 909
 910
 911 class FFmpegCopyStreamPostProcessor(FFmpegFixupPostProcessor):
 912     MESSAGE = 'Copying stream'
 913
 914     @PostProcessor._restrict_to(images=False)
 915     def run(self, info):
 916         self._fixup(self.MESSAGE, info['filepath'], ['-c', 'copy', '-map', '0', '-dn'])
 917         return [], info
 918
 919
 920 class FFmpegFixupDurationPP(FFmpegCopyStreamPostProcessor):
 921     MESSAGE = 'Fixing video duration'
 922
 923
 924 class FFmpegFixupDuplicateMoovPP(FFmpegCopyStreamPostProcessor):
 925     MESSAGE = 'Fixing duplicate MOOV atoms'
 926
 927
 928 class FFmpegSubtitlesConvertorPP(FFmpegPostProcessor):
 929     SUPPORTED_EXTS = ('srt', 'vtt', 'ass', 'lrc')
 930
 931     def __init__(self, downloader=None, format=None):
 932         super(FFmpegSubtitlesConvertorPP, self).__init__(downloader)
 933         self.format = format
 934
 935     def run(self, info):
 936         subs = info.get('requested_subtitles')
 937         new_ext = self.format
 938         new_format = new_ext
 939         if new_format == 'vtt':
 940             new_format = 'webvtt'
 941         if subs is None:
 942             self.to_screen('There aren\'t any subtitles to convert')
 943             return [], info
 944         self.to_screen('Converting subtitles')
 945         sub_filenames = []
 946         for lang, sub in subs.items():
 947             if not os.path.exists(sub.get('filepath', '')):
 948                 self.report_warning(f'Skipping embedding {lang} subtitle because the file is missing')
 949                 continue
 950             ext = sub['ext']
 951             if ext == new_ext:
 952                 self.to_screen('Subtitle file for %s is already in the requested format' % new_ext)
 953                 continue
 954             elif ext == 'json':
 955                 self.to_screen(
 956                     'You have requested to convert json subtitles into another format, '
 957                     'which is currently not possible')
 958                 continue
 959             old_file = sub['filepath']
 960             sub_filenames.append(old_file)
 961             new_file = replace_extension(old_file, new_ext)
 962
 963             if ext in ('dfxp', 'ttml', 'tt'):
 964                 self.report_warning(
 965                     'You have requested to convert dfxp (TTML) subtitles into another format, '
 966                     'which results in style information loss')
 967
 968                 dfxp_file = old_file
 969                 srt_file = replace_extension(old_file, 'srt')
 970
 971                 with open(dfxp_file, 'rb') as f:
 972                     srt_data = dfxp2srt(f.read())
 973
 974                 with io.open(srt_file, 'wt', encoding='utf-8') as f:
 975                     f.write(srt_data)
 976                 old_file = srt_file
 977
 978                 subs[lang] = {
 979                     'ext': 'srt',
 980                     'data': srt_data,
 981                     'filepath': srt_file,
 982                 }
 983
 984                 if new_ext == 'srt':
 985                     continue
 986                 else:
 987                     sub_filenames.append(srt_file)
 988
 989             self.run_ffmpeg(old_file, new_file, ['-f', new_format])
 990
 991             with io.open(new_file, 'rt', encoding='utf-8') as f:
 992                 subs[lang] = {
 993                     'ext': new_ext,
 994                     'data': f.read(),
 995                     'filepath': new_file,
 996                 }
 997
 998             info['__files_to_move'][new_file] = replace_extension(
 999                 info['__files_to_move'][sub['filepath']], new_ext)
1000
1001         return sub_filenames, info
1002
1003
1004 class FFmpegSplitChaptersPP(FFmpegPostProcessor):
1005     def __init__(self, downloader, force_keyframes=False):
1006         FFmpegPostProcessor.__init__(self, downloader)
1007         self._force_keyframes = force_keyframes
1008
1009     def _prepare_filename(self, number, chapter, info):
1010         info = info.copy()
1011         info.update({
1012             'section_number': number,
1013             'section_title': chapter.get('title'),
1014             'section_start': chapter.get('start_time'),
1015             'section_end': chapter.get('end_time'),
1016         })
1017         return self._downloader.prepare_filename(info, 'chapter')
1018
1019     def _ffmpeg_args_for_chapter(self, number, chapter, info):
1020         destination = self._prepare_filename(number, chapter, info)
1021         if not self._downloader._ensure_dir_exists(encodeFilename(destination)):
1022             return
1023
1024         chapter['filepath'] = destination
1025         self.to_screen('Chapter %03d; Destination: %s' % (number, destination))
1026         return (
1027             destination,
1028             ['-ss', compat_str(chapter['start_time']),
1029              '-t', compat_str(chapter['end_time'] - chapter['start_time'])])
1030
1031     @PostProcessor._restrict_to(images=False)
1032     def run(self, info):
1033         chapters = info.get('chapters') or []
1034         if not chapters:
1035             self.to_screen('Chapter information is unavailable')
1036             return [], info
1037
1038         in_file = info['filepath']
1039         if self._force_keyframes and len(chapters) > 1:
1040             in_file = self.force_keyframes(in_file, (c['start_time'] for c in chapters))
1041         self.to_screen('Splitting video by chapters; %d chapters found' % len(chapters))
1042         for idx, chapter in enumerate(chapters):
1043             destination, opts = self._ffmpeg_args_for_chapter(idx + 1, chapter, info)
1044             self.real_run_ffmpeg([(in_file, opts)], [(destination, ['-c', 'copy'])])
1045         if in_file != info['filepath']:
1046             os.remove(in_file)
1047         return [], info
1048
1049
1050 class FFmpegThumbnailsConvertorPP(FFmpegPostProcessor):
1051     SUPPORTED_EXTS = ('jpg', 'png')
1052
1053     def __init__(self, downloader=None, format=None):
1054         super(FFmpegThumbnailsConvertorPP, self).__init__(downloader)
1055         self.format = format
1056
1057     @staticmethod
1058     def is_webp(path):
1059         with open(encodeFilename(path), 'rb') as f:
1060             b = f.read(12)
1061         return b[0:4] == b'RIFF' and b[8:] == b'WEBP'
1062
1063     def fixup_webp(self, info, idx=-1):
1064         thumbnail_filename = info['thumbnails'][idx]['filepath']
1065         _, thumbnail_ext = os.path.splitext(thumbnail_filename)
1066         if thumbnail_ext:
1067             thumbnail_ext = thumbnail_ext[1:].lower()
1068             if thumbnail_ext != 'webp' and self.is_webp(thumbnail_filename):
1069                 self.to_screen('Correcting thumbnail "%s" extension to webp' % thumbnail_filename)
1070                 webp_filename = replace_extension(thumbnail_filename, 'webp')
1071                 os.replace(thumbnail_filename, webp_filename)
1072                 info['thumbnails'][idx]['filepath'] = webp_filename
1073                 info['__files_to_move'][webp_filename] = replace_extension(
1074                     info['__files_to_move'].pop(thumbnail_filename), 'webp')
1075
1076     @staticmethod
1077     def _options(target_ext):
1078         if target_ext == 'jpg':
1079             return ['-bsf:v', 'mjpeg2jpeg']
1080         return []
1081
1082     def convert_thumbnail(self, thumbnail_filename, target_ext):
1083         thumbnail_conv_filename = replace_extension(thumbnail_filename, target_ext)
1084
1085         self.to_screen('Converting thumbnail "%s" to %s' % (thumbnail_filename, target_ext))
1086         self.real_run_ffmpeg(
1087             [(thumbnail_filename, ['-f', 'image2', '-pattern_type', 'none'])],
1088             [(thumbnail_conv_filename.replace('%', '%%'), self._options(target_ext))])
1089         return thumbnail_conv_filename
1090
1091     def run(self, info):
1092         files_to_delete = []
1093         has_thumbnail = False
1094
1095         for idx, thumbnail_dict in enumerate(info['thumbnails']):
1096             if 'filepath' not in thumbnail_dict:
1097                 continue
1098             has_thumbnail = True
1099             self.fixup_webp(info, idx)
1100             original_thumbnail = thumbnail_dict['filepath']
1101             _, thumbnail_ext = os.path.splitext(original_thumbnail)
1102             if thumbnail_ext:
1103                 thumbnail_ext = thumbnail_ext[1:].lower()
1104             if thumbnail_ext == 'jpeg':
1105                 thumbnail_ext = 'jpg'
1106             if thumbnail_ext == self.format:
1107                 self.to_screen('Thumbnail "%s" is already in the requested format' % original_thumbnail)
1108                 continue
1109             thumbnail_dict['filepath'] = self.convert_thumbnail(original_thumbnail, self.format)
1110             files_to_delete.append(original_thumbnail)
1111             info['__files_to_move'][thumbnail_dict['filepath']] = replace_extension(
1112                 info['__files_to_move'][original_thumbnail], self.format)
1113
1114         if not has_thumbnail:
1115             self.to_screen('There aren\'t any thumbnails to convert')
1116         return files_to_delete, info