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  57             warning = 'Your copy of %s is outdated, update %s to version %s or newer if you encounter any errors.'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 run_ffmpeg_multiple_files(self, input_paths, out_path, opts):
 143         self.check_version()
 144
 145         oldest_mtime = min(
 146             os.stat(encodeFilename(path)).st_mtime for path in input_paths)
 147
 148         opts += self._configuration_args()
 149
 150         files_cmd = []
 151         for path in input_paths:
 152             files_cmd.extend([
 153                 encodeArgument('-i'),
 154                 encodeFilename(self._ffmpeg_filename_argument(path), True)
 155             ])
 156         cmd = ([encodeFilename(self.executable, True), encodeArgument('-y')] +
 157                files_cmd +
 158                [encodeArgument(o) for o in opts] +
 159                [encodeFilename(self._ffmpeg_filename_argument(out_path), True)])
 160
 161         if self._downloader.params.get('verbose', False):
 162             self._downloader.to_screen('[debug] ffmpeg command line: %s' % shell_quote(cmd))
 163         p = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE, stdin=subprocess.PIPE)
 164         stdout, stderr = p.communicate()
 165         if p.returncode != 0:
 166             stderr = stderr.decode('utf-8', 'replace')
 167             msg = stderr.strip().split('\n')[-1]
 168             raise FFmpegPostProcessorError(msg)
 169         self.try_utime(out_path, oldest_mtime, oldest_mtime)
 170
 171     def run_ffmpeg(self, path, out_path, opts):
 172         self.run_ffmpeg_multiple_files([path], out_path, opts)
 173
 174     def _ffmpeg_filename_argument(self, fn):
 175         # Always use 'file:' because the filename may contain ':' (ffmpeg
 176         # interprets that as a protocol) or can start with '-' (-- is broken in
 177         # ffmpeg, see https://ffmpeg.org/trac/ffmpeg/ticket/2127 for details)
 178         # Also leave '-' intact in order not to break streaming to stdout.
 179         return 'file:' + fn if fn != '-' else fn
 180
 181
 182 class FFmpegExtractAudioPP(FFmpegPostProcessor):
 183     def __init__(self, downloader=None, preferredcodec=None, preferredquality=None, nopostoverwrites=False):
 184         FFmpegPostProcessor.__init__(self, downloader)
 185         if preferredcodec is None:
 186             preferredcodec = 'best'
 187         self._preferredcodec = preferredcodec
 188         self._preferredquality = preferredquality
 189         self._nopostoverwrites = nopostoverwrites
 190
 191     def get_audio_codec(self, path):
 192
 193         if not self.probe_available:
 194             raise PostProcessingError('ffprobe or avprobe not found. Please install one.')
 195         try:
 196             cmd = [
 197                 encodeFilename(self.probe_executable, True),
 198                 encodeArgument('-show_streams'),
 199                 encodeFilename(self._ffmpeg_filename_argument(path), True)]
 200             if self._downloader.params.get('verbose', False):
 201                 self._downloader.to_screen('[debug] %s command line: %s' % (self.basename, shell_quote(cmd)))
 202             handle = subprocess.Popen(cmd, stderr=compat_subprocess_get_DEVNULL(), stdout=subprocess.PIPE, stdin=subprocess.PIPE)
 203             output = handle.communicate()[0]
 204             if handle.wait() != 0:
 205                 return None
 206         except (IOError, OSError):
 207             return None
 208         audio_codec = None
 209         for line in output.decode('ascii', 'ignore').split('\n'):
 210             if line.startswith('codec_name='):
 211                 audio_codec = line.split('=')[1].strip()
 212             elif line.strip() == 'codec_type=audio' and audio_codec is not None:
 213                 return audio_codec
 214         return None
 215
 216     def run_ffmpeg(self, path, out_path, codec, more_opts):
 217         if codec is None:
 218             acodec_opts = []
 219         else:
 220             acodec_opts = ['-acodec', codec]
 221         opts = ['-vn'] + acodec_opts + more_opts
 222         try:
 223             FFmpegPostProcessor.run_ffmpeg(self, path, out_path, opts)
 224         except FFmpegPostProcessorError as err:
 225             raise AudioConversionError(err.msg)
 226
 227     def run(self, information):
 228         path = information['filepath']
 229
 230         filecodec = self.get_audio_codec(path)
 231         if filecodec is None:
 232             raise PostProcessingError('WARNING: unable to obtain file audio codec with ffprobe')
 233
 234         more_opts = []
 235         if self._preferredcodec == 'best' or self._preferredcodec == filecodec or (self._preferredcodec == 'm4a' and filecodec == 'aac'):
 236             if filecodec == 'aac' and self._preferredcodec in ['m4a', 'best']:
 237                 # Lossless, but in another container
 238                 acodec = 'copy'
 239                 extension = 'm4a'
 240                 more_opts = ['-bsf:a', 'aac_adtstoasc']
 241             elif filecodec in ['aac', 'mp3', 'vorbis', 'opus']:
 242                 # Lossless if possible
 243                 acodec = 'copy'
 244                 extension = filecodec
 245                 if filecodec == 'aac':
 246                     more_opts = ['-f', 'adts']
 247                 if filecodec == 'vorbis':
 248                     extension = 'ogg'
 249             else:
 250                 # MP3 otherwise.
 251                 acodec = 'libmp3lame'
 252                 extension = 'mp3'
 253                 more_opts = []
 254                 if self._preferredquality is not None:
 255                     if int(self._preferredquality) < 10:
 256                         more_opts += ['-q:a', self._preferredquality]
 257                     else:
 258                         more_opts += ['-b:a', self._preferredquality + 'k']
 259         else:
 260             # We convert the audio (lossy)
 261             acodec = {'mp3': 'libmp3lame', 'aac': 'aac', 'm4a': 'aac', 'opus': 'opus', 'vorbis': 'libvorbis', 'wav': None}[self._preferredcodec]
 262             extension = self._preferredcodec
 263             more_opts = []
 264             if self._preferredquality is not None:
 265                 # The opus codec doesn't support the -aq option
 266                 if int(self._preferredquality) < 10 and extension != 'opus':
 267                     more_opts += ['-q:a', self._preferredquality]
 268                 else:
 269                     more_opts += ['-b:a', self._preferredquality + 'k']
 270             if self._preferredcodec == 'aac':
 271                 more_opts += ['-f', 'adts']
 272             if self._preferredcodec == 'm4a':
 273                 more_opts += ['-bsf:a', 'aac_adtstoasc']
 274             if self._preferredcodec == 'vorbis':
 275                 extension = 'ogg'
 276             if self._preferredcodec == 'wav':
 277                 extension = 'wav'
 278                 more_opts += ['-f', 'wav']
 279
 280         prefix, sep, ext = path.rpartition('.')  # not os.path.splitext, since the latter does not work on unicode in all setups
 281         new_path = prefix + sep + extension
 282
 283         # If we download foo.mp3 and convert it to... foo.mp3, then don't delete foo.mp3, silly.
 284         if (new_path == path or
 285                 (self._nopostoverwrites and os.path.exists(encodeFilename(new_path)))):
 286             self._downloader.to_screen('[ffmpeg] Post-process file %s exists, skipping' % new_path)
 287             return [], information
 288
 289         try:
 290             self._downloader.to_screen('[ffmpeg] Destination: ' + new_path)
 291             self.run_ffmpeg(path, new_path, acodec, more_opts)
 292         except AudioConversionError as e:
 293             raise PostProcessingError(
 294                 'audio conversion failed: ' + e.msg)
 295         except Exception:
 296             raise PostProcessingError('error running ' + self.basename)
 297
 298         # Try to update the date time for extracted audio file.
 299         if information.get('filetime') is not None:
 300             self.try_utime(
 301                 new_path, time.time(), information['filetime'],
 302                 errnote='Cannot update utime of audio file')
 303
 304         information['filepath'] = new_path
 305         information['ext'] = extension
 306
 307         return [path], information
 308
 309
 310 class FFmpegVideoConvertorPP(FFmpegPostProcessor):
 311     def __init__(self, downloader=None, preferedformat=None):
 312         super(FFmpegVideoConvertorPP, self).__init__(downloader)
 313         self._preferedformat = preferedformat
 314
 315     def run(self, information):
 316         path = information['filepath']
 317         if information['ext'] == self._preferedformat:
 318             self._downloader.to_screen('[ffmpeg] Not converting video file %s - already is in target format %s' % (path, self._preferedformat))
 319             return [], information
 320         options = []
 321         if self._preferedformat == 'avi':
 322             options.extend(['-c:v', 'libxvid', '-vtag', 'XVID'])
 323         prefix, sep, ext = path.rpartition('.')
 324         outpath = prefix + sep + self._preferedformat
 325         self._downloader.to_screen('[' + 'ffmpeg' + '] Converting video from %s to %s, Destination: ' % (information['ext'], self._preferedformat) + outpath)
 326         self.run_ffmpeg(path, outpath, options)
 327         information['filepath'] = outpath
 328         information['format'] = self._preferedformat
 329         information['ext'] = self._preferedformat
 330         return [path], information
 331
 332
 333 class FFmpegEmbedSubtitlePP(FFmpegPostProcessor):
 334     def run(self, information):
 335         if information['ext'] not in ('mp4', 'webm', 'mkv'):
 336             self._downloader.to_screen('[ffmpeg] Subtitles can only be embedded in mp4, webm or mkv files')
 337             return [], information
 338         subtitles = information.get('requested_subtitles')
 339         if not subtitles:
 340             self._downloader.to_screen('[ffmpeg] There aren\'t any subtitles to embed')
 341             return [], information
 342
 343         filename = information['filepath']
 344
 345         ext = information['ext']
 346         sub_langs = []
 347         sub_filenames = []
 348         webm_vtt_warn = False
 349
 350         for lang, sub_info in subtitles.items():
 351             sub_ext = sub_info['ext']
 352             if ext != 'webm' or ext == 'webm' and sub_ext == 'vtt':
 353                 sub_langs.append(lang)
 354                 sub_filenames.append(subtitles_filename(filename, lang, sub_ext))
 355             else:
 356                 if not webm_vtt_warn and ext == 'webm' and sub_ext != 'vtt':
 357                     webm_vtt_warn = True
 358                     self._downloader.to_screen('[ffmpeg] Only WebVTT subtitles can be embedded in webm files')
 359
 360         if not sub_langs:
 361             return [], information
 362
 363         input_files = [filename] + sub_filenames
 364
 365         opts = [
 366             '-map', '0:v',
 367             '-c:v', 'copy',
 368             '-map', '0:a',
 369             '-c:a', 'copy',
 370             # Don't copy the existing subtitles, we may be running the
 371             # postprocessor a second time
 372             '-map', '-0:s',
 373         ]
 374         if information['ext'] == 'mp4':
 375             opts += ['-c:s', 'mov_text']
 376         for (i, lang) in enumerate(sub_langs):
 377             opts.extend(['-map', '%d:0' % (i + 1)])
 378             lang_code = ISO639Utils.short2long(lang)
 379             if lang_code is not None:
 380                 opts.extend(['-metadata:s:s:%d' % i, 'language=%s' % lang_code])
 381
 382         temp_filename = prepend_extension(filename, 'temp')
 383         self._downloader.to_screen('[ffmpeg] Embedding subtitles in \'%s\'' % filename)
 384         self.run_ffmpeg_multiple_files(input_files, temp_filename, opts)
 385         os.remove(encodeFilename(filename))
 386         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 387
 388         return sub_filenames, information
 389
 390
 391 class FFmpegMetadataPP(FFmpegPostProcessor):
 392     def run(self, info):
 393         metadata = {}
 394
 395         def add(meta_list, info_list=None):
 396             if not info_list:
 397                 info_list = meta_list
 398             if not isinstance(meta_list, (list, tuple)):
 399                 meta_list = (meta_list,)
 400             if not isinstance(info_list, (list, tuple)):
 401                 info_list = (info_list,)
 402             for info_f in info_list:
 403                 if info.get(info_f) is not None:
 404                     for meta_f in meta_list:
 405                         metadata[meta_f] = info[info_f]
 406                     break
 407
 408         add('title', ('track', 'title'))
 409         add('date', 'upload_date')
 410         add(('description', 'comment'), 'description')
 411         add('purl', 'webpage_url')
 412         add('track', 'track_number')
 413         add('artist', ('artist', 'creator', 'uploader', 'uploader_id'))
 414         add('genre')
 415         add('album')
 416         add('album_artist')
 417         add('disc', 'disc_number')
 418
 419         if not metadata:
 420             self._downloader.to_screen('[ffmpeg] There isn\'t any metadata to add')
 421             return [], info
 422
 423         filename = info['filepath']
 424         temp_filename = prepend_extension(filename, 'temp')
 425
 426         if info['ext'] == 'm4a':
 427             options = ['-vn', '-acodec', 'copy']
 428         else:
 429             options = ['-c', 'copy']
 430
 431         for (name, value) in metadata.items():
 432             options.extend(['-metadata', '%s=%s' % (name, value)])
 433
 434         self._downloader.to_screen('[ffmpeg] Adding metadata to \'%s\'' % filename)
 435         self.run_ffmpeg(filename, temp_filename, options)
 436         os.remove(encodeFilename(filename))
 437         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 438         return [], info
 439
 440
 441 class FFmpegMergerPP(FFmpegPostProcessor):
 442     def run(self, info):
 443         filename = info['filepath']
 444         temp_filename = prepend_extension(filename, 'temp')
 445         args = ['-c', 'copy', '-map', '0:v:0', '-map', '1:a:0']
 446         self._downloader.to_screen('[ffmpeg] Merging formats into "%s"' % filename)
 447         self.run_ffmpeg_multiple_files(info['__files_to_merge'], temp_filename, args)
 448         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 449         return info['__files_to_merge'], info
 450
 451     def can_merge(self):
 452         # TODO: figure out merge-capable ffmpeg version
 453         if self.basename != 'avconv':
 454             return True
 455
 456         required_version = '10-0'
 457         if is_outdated_version(
 458                 self._versions[self.basename], required_version):
 459             warning = ('Your copy of %s is outdated and unable to properly mux separate video and audio files, '
 460                        'youtube-dl will download single file media. '
 461                        'Update %s to version %s or newer to fix this.') % (
 462                            self.basename, self.basename, required_version)
 463             if self._downloader:
 464                 self._downloader.report_warning(warning)
 465             return False
 466         return True
 467
 468
 469 class FFmpegFixupStretchedPP(FFmpegPostProcessor):
 470     def run(self, info):
 471         stretched_ratio = info.get('stretched_ratio')
 472         if stretched_ratio is None or stretched_ratio == 1:
 473             return [], info
 474
 475         filename = info['filepath']
 476         temp_filename = prepend_extension(filename, 'temp')
 477
 478         options = ['-c', 'copy', '-aspect', '%f' % stretched_ratio]
 479         self._downloader.to_screen('[ffmpeg] Fixing aspect ratio in "%s"' % filename)
 480         self.run_ffmpeg(filename, temp_filename, options)
 481
 482         os.remove(encodeFilename(filename))
 483         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 484
 485         return [], info
 486
 487
 488 class FFmpegFixupM4aPP(FFmpegPostProcessor):
 489     def run(self, info):
 490         if info.get('container') != 'm4a_dash':
 491             return [], info
 492
 493         filename = info['filepath']
 494         temp_filename = prepend_extension(filename, 'temp')
 495
 496         options = ['-c', 'copy', '-f', 'mp4']
 497         self._downloader.to_screen('[ffmpeg] Correcting container in "%s"' % filename)
 498         self.run_ffmpeg(filename, temp_filename, options)
 499
 500         os.remove(encodeFilename(filename))
 501         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 502
 503         return [], info
 504
 505
 506 class FFmpegFixupM3u8PP(FFmpegPostProcessor):
 507     def run(self, info):
 508         filename = info['filepath']
 509         temp_filename = prepend_extension(filename, 'temp')
 510
 511         options = ['-c', 'copy', '-f', 'mp4', '-bsf:a', 'aac_adtstoasc']
 512         self._downloader.to_screen('[ffmpeg] Fixing malformated aac bitstream in "%s"' % filename)
 513         self.run_ffmpeg(filename, temp_filename, options)
 514
 515         os.remove(encodeFilename(filename))
 516         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 517
 518         return [], info
 519
 520
 521 class FFmpegSubtitlesConvertorPP(FFmpegPostProcessor):
 522     def __init__(self, downloader=None, format=None):
 523         super(FFmpegSubtitlesConvertorPP, self).__init__(downloader)
 524         self.format = format
 525
 526     def run(self, info):
 527         subs = info.get('requested_subtitles')
 528         filename = info['filepath']
 529         new_ext = self.format
 530         new_format = new_ext
 531         if new_format == 'vtt':
 532             new_format = 'webvtt'
 533         if subs is None:
 534             self._downloader.to_screen('[ffmpeg] There aren\'t any subtitles to convert')
 535             return [], info
 536         self._downloader.to_screen('[ffmpeg] Converting subtitles')
 537         sub_filenames = []
 538         for lang, sub in subs.items():
 539             ext = sub['ext']
 540             if ext == new_ext:
 541                 self._downloader.to_screen(
 542                     '[ffmpeg] Subtitle file for %s is already in the requested'
 543                     'format' % new_ext)
 544                 continue
 545             old_file = subtitles_filename(filename, lang, ext)
 546             sub_filenames.append(old_file)
 547             new_file = subtitles_filename(filename, lang, new_ext)
 548
 549             if ext == 'dfxp' or ext == 'ttml' or ext == 'tt':
 550                 self._downloader.report_warning(
 551                     'You have requested to convert dfxp (TTML) subtitles into another format, '
 552                     'which results in style information loss')
 553
 554                 dfxp_file = old_file
 555                 srt_file = subtitles_filename(filename, lang, 'srt')
 556
 557                 with io.open(dfxp_file, 'rt', encoding='utf-8') as f:
 558                     srt_data = dfxp2srt(f.read())
 559
 560                 with io.open(srt_file, 'wt', encoding='utf-8') as f:
 561                     f.write(srt_data)
 562                 old_file = srt_file
 563
 564                 subs[lang] = {
 565                     'ext': 'srt',
 566                     'data': srt_data
 567                 }
 568
 569                 if new_ext == 'srt':
 570                     continue
 571                 else:
 572                     sub_filenames.append(srt_file)
 573
 574             self.run_ffmpeg(old_file, new_file, ['-f', new_format])
 575
 576             with io.open(new_file, 'rt', encoding='utf-8') as f:
 577                 subs[lang] = {
 578                     'ext': new_ext,
 579                     'data': f.read(),
 580                 }
 581
 582         return sub_filenames, info