youtube_dl/postprocessor/ffmpeg.py

   1 from __future__ import unicode_literals
   2
   3 import io
   4 import os
   5 import subprocess
   6 import time
   7
   8
   9 from .common import AudioConversionError, PostProcessor
  10
  11 from ..compat import (
  12     compat_subprocess_get_DEVNULL,
  13 )
  14 from ..utils import (
  15     encodeArgument,
  16     encodeFilename,
  17     get_exe_version,
  18     is_outdated_version,
  19     PostProcessingError,
  20     prepend_extension,
  21     shell_quote,
  22     subtitles_filename,
  23     dfxp2srt,
  24     ISO639Utils,
  25 )
  26
  27
  28 EXT_TO_OUT_FORMATS = {
  29     "aac": "adts",
  30     "m4a": "ipod",
  31     "mka": "matroska",
  32     "mkv": "matroska",
  33     "mpg": "mpeg",
  34     "ogv": "ogg",
  35     "ts": "mpegts",
  36     "wma": "asf",
  37     "wmv": "asf",
  38 }
  39
  40
  41 class FFmpegPostProcessorError(PostProcessingError):
  42     pass
  43
  44
  45 class FFmpegPostProcessor(PostProcessor):
  46     def __init__(self, downloader=None):
  47         PostProcessor.__init__(self, downloader)
  48         self._determine_executables()
  49
  50     def check_version(self):
  51         if not self.available:
  52             raise FFmpegPostProcessorError('ffmpeg or avconv not found. Please install one.')
  53
  54         required_version = '10-0' if self.basename == 'avconv' else '1.0'
  55         if is_outdated_version(
  56                 self._versions[self.basename], required_version):
  57             warning = 'Your copy of %s is outdated, update %s to version %s or newer if you encounter any errors.' % (
  58                 self.basename, self.basename, required_version)
  59             if self._downloader:
  60                 self._downloader.report_warning(warning)
  61
  62     @staticmethod
  63     def get_versions(downloader=None):
  64         return FFmpegPostProcessor(downloader)._versions
  65
  66     def _determine_executables(self):
  67         programs = ['avprobe', 'avconv', 'ffmpeg', 'ffprobe']
  68         prefer_ffmpeg = False
  69
  70         self.basename = None
  71         self.probe_basename = None
  72
  73         self._paths = None
  74         self._versions = None
  75         if self._downloader:
  76             prefer_ffmpeg = self._downloader.params.get('prefer_ffmpeg', False)
  77             location = self._downloader.params.get('ffmpeg_location')
  78             if location is not None:
  79                 if not os.path.exists(location):
  80                     self._downloader.report_warning(
  81                         'ffmpeg-location %s does not exist! '
  82                         'Continuing without avconv/ffmpeg.' % (location))
  83                     self._versions = {}
  84                     return
  85                 elif not os.path.isdir(location):
  86                     basename = os.path.splitext(os.path.basename(location))[0]
  87                     if basename not in programs:
  88                         self._downloader.report_warning(
  89                             'Cannot identify executable %s, its basename should be one of %s. '
  90                             'Continuing without avconv/ffmpeg.' %
  91                             (location, ', '.join(programs)))
  92                         self._versions = {}
  93                         return None
  94                     location = os.path.dirname(os.path.abspath(location))
  95                     if basename in ('ffmpeg', 'ffprobe'):
  96                         prefer_ffmpeg = True
  97
  98                 self._paths = dict(
  99                     (p, os.path.join(location, p)) for p in programs)
 100                 self._versions = dict(
 101                     (p, get_exe_version(self._paths[p], args=['-version']))
 102                     for p in programs)
 103         if self._versions is None:
 104             self._versions = dict(
 105                 (p, get_exe_version(p, args=['-version'])) for p in programs)
 106             self._paths = dict((p, p) for p in programs)
 107
 108         if prefer_ffmpeg:
 109             prefs = ('ffmpeg', 'avconv')
 110         else:
 111             prefs = ('avconv', 'ffmpeg')
 112         for p in prefs:
 113             if self._versions[p]:
 114                 self.basename = p
 115                 break
 116
 117         if prefer_ffmpeg:
 118             prefs = ('ffprobe', 'avprobe')
 119         else:
 120             prefs = ('avprobe', 'ffprobe')
 121         for p in prefs:
 122             if self._versions[p]:
 123                 self.probe_basename = p
 124                 break
 125
 126     @property
 127     def available(self):
 128         return self.basename is not None
 129
 130     @property
 131     def executable(self):
 132         return self._paths[self.basename]
 133
 134     @property
 135     def probe_available(self):
 136         return self.probe_basename is not None
 137
 138     @property
 139     def probe_executable(self):
 140         return self._paths[self.probe_basename]
 141
 142     def get_audio_codec(self, path):
 143         if not self.probe_available:
 144             raise PostProcessingError('ffprobe or avprobe not found. Please install one.')
 145         try:
 146             cmd = [
 147                 encodeFilename(self.probe_executable, True),
 148                 encodeArgument('-show_streams'),
 149                 encodeFilename(self._ffmpeg_filename_argument(path), True)]
 150             if self._downloader.params.get('verbose', False):
 151                 self._downloader.to_screen('[debug] %s command line: %s' % (self.basename, shell_quote(cmd)))
 152             handle = subprocess.Popen(cmd, stderr=compat_subprocess_get_DEVNULL(), stdout=subprocess.PIPE, stdin=subprocess.PIPE)
 153             output = handle.communicate()[0]
 154             if handle.wait() != 0:
 155                 return None
 156         except (IOError, OSError):
 157             return None
 158         audio_codec = None
 159         for line in output.decode('ascii', 'ignore').split('\n'):
 160             if line.startswith('codec_name='):
 161                 audio_codec = line.split('=')[1].strip()
 162             elif line.strip() == 'codec_type=audio' and audio_codec is not None:
 163                 return audio_codec
 164         return None
 165
 166     def run_ffmpeg_multiple_files(self, input_paths, out_path, opts):
 167         self.check_version()
 168
 169         oldest_mtime = min(
 170             os.stat(encodeFilename(path)).st_mtime for path in input_paths)
 171
 172         opts += self._configuration_args()
 173
 174         files_cmd = []
 175         for path in input_paths:
 176             files_cmd.extend([
 177                 encodeArgument('-i'),
 178                 encodeFilename(self._ffmpeg_filename_argument(path), True)
 179             ])
 180         cmd = ([encodeFilename(self.executable, True), encodeArgument('-y')] +
 181                files_cmd +
 182                [encodeArgument(o) for o in opts] +
 183                [encodeFilename(self._ffmpeg_filename_argument(out_path), True)])
 184
 185         if self._downloader.params.get('verbose', False):
 186             self._downloader.to_screen('[debug] ffmpeg command line: %s' % shell_quote(cmd))
 187         p = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE, stdin=subprocess.PIPE)
 188         stdout, stderr = p.communicate()
 189         if p.returncode != 0:
 190             stderr = stderr.decode('utf-8', 'replace')
 191             msg = stderr.strip().split('\n')[-1]
 192             raise FFmpegPostProcessorError(msg)
 193         self.try_utime(out_path, oldest_mtime, oldest_mtime)
 194
 195     def run_ffmpeg(self, path, out_path, opts):
 196         self.run_ffmpeg_multiple_files([path], out_path, opts)
 197
 198     def _ffmpeg_filename_argument(self, fn):
 199         # Always use 'file:' because the filename may contain ':' (ffmpeg
 200         # interprets that as a protocol) or can start with '-' (-- is broken in
 201         # ffmpeg, see https://ffmpeg.org/trac/ffmpeg/ticket/2127 for details)
 202         # Also leave '-' intact in order not to break streaming to stdout.
 203         return 'file:' + fn if fn != '-' else fn
 204
 205
 206 class FFmpegExtractAudioPP(FFmpegPostProcessor):
 207     def __init__(self, downloader=None, preferredcodec=None, preferredquality=None, nopostoverwrites=False):
 208         FFmpegPostProcessor.__init__(self, downloader)
 209         if preferredcodec is None:
 210             preferredcodec = 'best'
 211         self._preferredcodec = preferredcodec
 212         self._preferredquality = preferredquality
 213         self._nopostoverwrites = nopostoverwrites
 214
 215     def run_ffmpeg(self, path, out_path, codec, more_opts):
 216         if codec is None:
 217             acodec_opts = []
 218         else:
 219             acodec_opts = ['-acodec', codec]
 220         opts = ['-vn'] + acodec_opts + more_opts
 221         try:
 222             FFmpegPostProcessor.run_ffmpeg(self, path, out_path, opts)
 223         except FFmpegPostProcessorError as err:
 224             raise AudioConversionError(err.msg)
 225
 226     def run(self, information):
 227         path = information['filepath']
 228
 229         filecodec = self.get_audio_codec(path)
 230         if filecodec is None:
 231             raise PostProcessingError('WARNING: unable to obtain file audio codec with ffprobe')
 232
 233         more_opts = []
 234         if self._preferredcodec == 'best' or self._preferredcodec == filecodec or (self._preferredcodec == 'm4a' and filecodec == 'aac'):
 235             if filecodec == 'aac' and self._preferredcodec in ['m4a', 'best']:
 236                 # Lossless, but in another container
 237                 acodec = 'copy'
 238                 extension = 'm4a'
 239                 more_opts = ['-bsf:a', 'aac_adtstoasc']
 240             elif filecodec in ['aac', 'mp3', 'vorbis', 'opus']:
 241                 # Lossless if possible
 242                 acodec = 'copy'
 243                 extension = filecodec
 244                 if filecodec == 'aac':
 245                     more_opts = ['-f', 'adts']
 246                 if filecodec == 'vorbis':
 247                     extension = 'ogg'
 248             else:
 249                 # MP3 otherwise.
 250                 acodec = 'libmp3lame'
 251                 extension = 'mp3'
 252                 more_opts = []
 253                 if self._preferredquality is not None:
 254                     if int(self._preferredquality) < 10:
 255                         more_opts += ['-q:a', self._preferredquality]
 256                     else:
 257                         more_opts += ['-b:a', self._preferredquality + 'k']
 258         else:
 259             # We convert the audio (lossy)
 260             acodec = {'mp3': 'libmp3lame', 'aac': 'aac', 'm4a': 'aac', 'opus': 'opus', 'vorbis': 'libvorbis', 'wav': None}[self._preferredcodec]
 261             extension = self._preferredcodec
 262             more_opts = []
 263             if self._preferredquality is not None:
 264                 # The opus codec doesn't support the -aq option
 265                 if int(self._preferredquality) < 10 and extension != 'opus':
 266                     more_opts += ['-q:a', self._preferredquality]
 267                 else:
 268                     more_opts += ['-b:a', self._preferredquality + 'k']
 269             if self._preferredcodec == 'aac':
 270                 more_opts += ['-f', 'adts']
 271             if self._preferredcodec == 'm4a':
 272                 more_opts += ['-bsf:a', 'aac_adtstoasc']
 273             if self._preferredcodec == 'vorbis':
 274                 extension = 'ogg'
 275             if self._preferredcodec == 'wav':
 276                 extension = 'wav'
 277                 more_opts += ['-f', 'wav']
 278
 279         prefix, sep, ext = path.rpartition('.')  # not os.path.splitext, since the latter does not work on unicode in all setups
 280         new_path = prefix + sep + extension
 281
 282         information['filepath'] = new_path
 283         information['ext'] = extension
 284
 285         # If we download foo.mp3 and convert it to... foo.mp3, then don't delete foo.mp3, silly.
 286         if (new_path == path or
 287                 (self._nopostoverwrites and os.path.exists(encodeFilename(new_path)))):
 288             self._downloader.to_screen('[ffmpeg] Post-process file %s exists, skipping' % new_path)
 289             return [], information
 290
 291         try:
 292             self._downloader.to_screen('[ffmpeg] Destination: ' + new_path)
 293             self.run_ffmpeg(path, new_path, acodec, more_opts)
 294         except AudioConversionError as e:
 295             raise PostProcessingError(
 296                 'audio conversion failed: ' + e.msg)
 297         except Exception:
 298             raise PostProcessingError('error running ' + self.basename)
 299
 300         # Try to update the date time for extracted audio file.
 301         if information.get('filetime') is not None:
 302             self.try_utime(
 303                 new_path, time.time(), information['filetime'],
 304                 errnote='Cannot update utime of audio file')
 305
 306         return [path], information
 307
 308
 309 class FFmpegVideoConvertorPP(FFmpegPostProcessor):
 310     def __init__(self, downloader=None, preferedformat=None):
 311         super(FFmpegVideoConvertorPP, self).__init__(downloader)
 312         self._preferedformat = preferedformat
 313
 314     def run(self, information):
 315         path = information['filepath']
 316         if information['ext'] == self._preferedformat:
 317             self._downloader.to_screen('[ffmpeg] Not converting video file %s - already is in target format %s' % (path, self._preferedformat))
 318             return [], information
 319         options = []
 320         if self._preferedformat == 'avi':
 321             options.extend(['-c:v', 'libxvid', '-vtag', 'XVID'])
 322         prefix, sep, ext = path.rpartition('.')
 323         outpath = prefix + sep + self._preferedformat
 324         self._downloader.to_screen('[' + 'ffmpeg' + '] Converting video from %s to %s, Destination: ' % (information['ext'], self._preferedformat) + outpath)
 325         self.run_ffmpeg(path, outpath, options)
 326         information['filepath'] = outpath
 327         information['format'] = self._preferedformat
 328         information['ext'] = self._preferedformat
 329         return [path], information
 330
 331
 332 class FFmpegEmbedSubtitlePP(FFmpegPostProcessor):
 333     def run(self, information):
 334         if information['ext'] not in ('mp4', 'webm', 'mkv'):
 335             self._downloader.to_screen('[ffmpeg] Subtitles can only be embedded in mp4, webm or mkv files')
 336             return [], information
 337         subtitles = information.get('requested_subtitles')
 338         if not subtitles:
 339             self._downloader.to_screen('[ffmpeg] There aren\'t any subtitles to embed')
 340             return [], information
 341
 342         filename = information['filepath']
 343
 344         ext = information['ext']
 345         sub_langs = []
 346         sub_filenames = []
 347         webm_vtt_warn = False
 348
 349         for lang, sub_info in subtitles.items():
 350             sub_ext = sub_info['ext']
 351             if ext != 'webm' or ext == 'webm' and sub_ext == 'vtt':
 352                 sub_langs.append(lang)
 353                 sub_filenames.append(subtitles_filename(filename, lang, sub_ext))
 354             else:
 355                 if not webm_vtt_warn and ext == 'webm' and sub_ext != 'vtt':
 356                     webm_vtt_warn = True
 357                     self._downloader.to_screen('[ffmpeg] Only WebVTT subtitles can be embedded in webm files')
 358
 359         if not sub_langs:
 360             return [], information
 361
 362         input_files = [filename] + sub_filenames
 363
 364         opts = [
 365             '-map', '0',
 366             '-c', 'copy',
 367             # Don't copy the existing subtitles, we may be running the
 368             # postprocessor a second time
 369             '-map', '-0:s',
 370         ]
 371         if information['ext'] == 'mp4':
 372             opts += ['-c:s', 'mov_text']
 373         for (i, lang) in enumerate(sub_langs):
 374             opts.extend(['-map', '%d:0' % (i + 1)])
 375             lang_code = ISO639Utils.short2long(lang)
 376             if lang_code is not None:
 377                 opts.extend(['-metadata:s:s:%d' % i, 'language=%s' % lang_code])
 378
 379         temp_filename = prepend_extension(filename, 'temp')
 380         self._downloader.to_screen('[ffmpeg] Embedding subtitles in \'%s\'' % filename)
 381         self.run_ffmpeg_multiple_files(input_files, temp_filename, opts)
 382         os.remove(encodeFilename(filename))
 383         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 384
 385         return sub_filenames, information
 386
 387
 388 class FFmpegMetadataPP(FFmpegPostProcessor):
 389     def run(self, info):
 390         metadata = {}
 391
 392         def add(meta_list, info_list=None):
 393             if not info_list:
 394                 info_list = meta_list
 395             if not isinstance(meta_list, (list, tuple)):
 396                 meta_list = (meta_list,)
 397             if not isinstance(info_list, (list, tuple)):
 398                 info_list = (info_list,)
 399             for info_f in info_list:
 400                 if info.get(info_f) is not None:
 401                     for meta_f in meta_list:
 402                         metadata[meta_f] = info[info_f]
 403                     break
 404
 405         add('title', ('track', 'title'))
 406         add('date', 'upload_date')
 407         add(('description', 'comment'), 'description')
 408         add('purl', 'webpage_url')
 409         add('track', 'track_number')
 410         add('artist', ('artist', 'creator', 'uploader', 'uploader_id'))
 411         add('genre')
 412         add('album')
 413         add('album_artist')
 414         add('disc', 'disc_number')
 415
 416         if not metadata:
 417             self._downloader.to_screen('[ffmpeg] There isn\'t any metadata to add')
 418             return [], info
 419
 420         filename = info['filepath']
 421         temp_filename = prepend_extension(filename, 'temp')
 422
 423         if info['ext'] == 'm4a':
 424             options = ['-vn', '-acodec', 'copy']
 425         else:
 426             options = ['-c', 'copy']
 427
 428         for (name, value) in metadata.items():
 429             options.extend(['-metadata', '%s=%s' % (name, value)])
 430
 431         self._downloader.to_screen('[ffmpeg] Adding metadata to \'%s\'' % filename)
 432         self.run_ffmpeg(filename, temp_filename, options)
 433         os.remove(encodeFilename(filename))
 434         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 435         return [], info
 436
 437
 438 class FFmpegMergerPP(FFmpegPostProcessor):
 439     def run(self, info):
 440         filename = info['filepath']
 441         temp_filename = prepend_extension(filename, 'temp')
 442         args = ['-c', 'copy', '-map', '0:v:0', '-map', '1:a:0']
 443         self._downloader.to_screen('[ffmpeg] Merging formats into "%s"' % filename)
 444         self.run_ffmpeg_multiple_files(info['__files_to_merge'], temp_filename, args)
 445         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 446         return info['__files_to_merge'], info
 447
 448     def can_merge(self):
 449         # TODO: figure out merge-capable ffmpeg version
 450         if self.basename != 'avconv':
 451             return True
 452
 453         required_version = '10-0'
 454         if is_outdated_version(
 455                 self._versions[self.basename], required_version):
 456             warning = ('Your copy of %s is outdated and unable to properly mux separate video and audio files, '
 457                        'youtube-dl will download single file media. '
 458                        'Update %s to version %s or newer to fix this.') % (
 459                            self.basename, self.basename, required_version)
 460             if self._downloader:
 461                 self._downloader.report_warning(warning)
 462             return False
 463         return True
 464
 465
 466 class FFmpegFixupStretchedPP(FFmpegPostProcessor):
 467     def run(self, info):
 468         stretched_ratio = info.get('stretched_ratio')
 469         if stretched_ratio is None or stretched_ratio == 1:
 470             return [], info
 471
 472         filename = info['filepath']
 473         temp_filename = prepend_extension(filename, 'temp')
 474
 475         options = ['-c', 'copy', '-aspect', '%f' % stretched_ratio]
 476         self._downloader.to_screen('[ffmpeg] Fixing aspect ratio in "%s"' % filename)
 477         self.run_ffmpeg(filename, temp_filename, options)
 478
 479         os.remove(encodeFilename(filename))
 480         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 481
 482         return [], info
 483
 484
 485 class FFmpegFixupM4aPP(FFmpegPostProcessor):
 486     def run(self, info):
 487         if info.get('container') != 'm4a_dash':
 488             return [], info
 489
 490         filename = info['filepath']
 491         temp_filename = prepend_extension(filename, 'temp')
 492
 493         options = ['-c', 'copy', '-f', 'mp4']
 494         self._downloader.to_screen('[ffmpeg] Correcting container in "%s"' % filename)
 495         self.run_ffmpeg(filename, temp_filename, options)
 496
 497         os.remove(encodeFilename(filename))
 498         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 499
 500         return [], info
 501
 502
 503 class FFmpegFixupM3u8PP(FFmpegPostProcessor):
 504     def run(self, info):
 505         filename = info['filepath']
 506         if self.get_audio_codec(filename) == 'aac':
 507             temp_filename = prepend_extension(filename, 'temp')
 508
 509             options = ['-c', 'copy', '-f', 'mp4', '-bsf:a', 'aac_adtstoasc']
 510             self._downloader.to_screen('[ffmpeg] Fixing malformated aac bitstream in "%s"' % filename)
 511             self.run_ffmpeg(filename, temp_filename, options)
 512
 513             os.remove(encodeFilename(filename))
 514             os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 515         return [], info
 516
 517
 518 class FFmpegSubtitlesConvertorPP(FFmpegPostProcessor):
 519     def __init__(self, downloader=None, format=None):
 520         super(FFmpegSubtitlesConvertorPP, self).__init__(downloader)
 521         self.format = format
 522
 523     def run(self, info):
 524         subs = info.get('requested_subtitles')
 525         filename = info['filepath']
 526         new_ext = self.format
 527         new_format = new_ext
 528         if new_format == 'vtt':
 529             new_format = 'webvtt'
 530         if subs is None:
 531             self._downloader.to_screen('[ffmpeg] There aren\'t any subtitles to convert')
 532             return [], info
 533         self._downloader.to_screen('[ffmpeg] Converting subtitles')
 534         sub_filenames = []
 535         for lang, sub in subs.items():
 536             ext = sub['ext']
 537             if ext == new_ext:
 538                 self._downloader.to_screen(
 539                     '[ffmpeg] Subtitle file for %s is already in the requested'
 540                     'format' % new_ext)
 541                 continue
 542             old_file = subtitles_filename(filename, lang, ext)
 543             sub_filenames.append(old_file)
 544             new_file = subtitles_filename(filename, lang, new_ext)
 545
 546             if ext == 'dfxp' or ext == 'ttml' or ext == 'tt':
 547                 self._downloader.report_warning(
 548                     'You have requested to convert dfxp (TTML) subtitles into another format, '
 549                     'which results in style information loss')
 550
 551                 dfxp_file = old_file
 552                 srt_file = subtitles_filename(filename, lang, 'srt')
 553
 554                 with io.open(dfxp_file, 'rt', encoding='utf-8') as f:
 555                     srt_data = dfxp2srt(f.read())
 556
 557                 with io.open(srt_file, 'wt', encoding='utf-8') as f:
 558                     f.write(srt_data)
 559                 old_file = srt_file
 560
 561                 subs[lang] = {
 562                     'ext': 'srt',
 563                     'data': srt_data
 564                 }
 565
 566                 if new_ext == 'srt':
 567                     continue
 568                 else:
 569                     sub_filenames.append(srt_file)
 570
 571             self.run_ffmpeg(old_file, new_file, ['-f', new_format])
 572
 573             with io.open(new_file, 'rt', encoding='utf-8') as f:
 574                 subs[lang] = {
 575                     'ext': new_ext,
 576                     'data': f.read(),
 577                 }
 578
 579         return sub_filenames, info