Raphaël G. Git Repositories - youtubedl/blob - 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     'flac': 'flac',
  31     'm4a': 'ipod',
  32     'mka': 'matroska',
  33     'mkv': 'matroska',
  34     'mpg': 'mpeg',
  35     'ogv': 'ogg',
  36     'ts': 'mpegts',
  37     'wma': 'asf',
  38     'wmv': 'asf',
  39 }
  40 ACODECS = {
  41     'mp3': 'libmp3lame',
  42     'aac': 'aac',
  43     'flac': 'flac',
  44     'm4a': 'aac',
  45     'opus': 'opus',
  46     'vorbis': 'libvorbis',
  47     'wav': None,
  48 }
  49
  50
  51 class FFmpegPostProcessorError(PostProcessingError):
  52     pass
  53
  54
  55 class FFmpegPostProcessor(PostProcessor):
  56     def __init__(self, downloader=None):
  57         PostProcessor.__init__(self, downloader)
  58         self._determine_executables()
  59
  60     def check_version(self):
  61         if not self.available:
  62             raise FFmpegPostProcessorError('ffmpeg or avconv not found. Please install one.')
  63
  64         required_version = '10-0' if self.basename == 'avconv' else '1.0'
  65         if is_outdated_version(
  66                 self._versions[self.basename], required_version):
  67             warning = 'Your copy of %s is outdated, update %s to version %s or newer if you encounter any errors.' % (
  68                 self.basename, self.basename, required_version)
  69             if self._downloader:
  70                 self._downloader.report_warning(warning)
  71
  72     @staticmethod
  73     def get_versions(downloader=None):
  74         return FFmpegPostProcessor(downloader)._versions
  75
  76     def _determine_executables(self):
  77         programs = ['avprobe', 'avconv', 'ffmpeg', 'ffprobe']
  78         prefer_ffmpeg = False
  79
  80         self.basename = None
  81         self.probe_basename = None
  82
  83         self._paths = None
  84         self._versions = None
  85         if self._downloader:
  86             prefer_ffmpeg = self._downloader.params.get('prefer_ffmpeg', False)
  87             location = self._downloader.params.get('ffmpeg_location')
  88             if location is not None:
  89                 if not os.path.exists(location):
  90                     self._downloader.report_warning(
  91                         'ffmpeg-location %s does not exist! '
  92                         'Continuing without avconv/ffmpeg.' % (location))
  93                     self._versions = {}
  94                     return
  95                 elif not os.path.isdir(location):
  96                     basename = os.path.splitext(os.path.basename(location))[0]
  97                     if basename not in programs:
  98                         self._downloader.report_warning(
  99                             'Cannot identify executable %s, its basename should be one of %s. '
 100                             'Continuing without avconv/ffmpeg.' %
 101                             (location, ', '.join(programs)))
 102                         self._versions = {}
 103                         return None
 104                     location = os.path.dirname(os.path.abspath(location))
 105                     if basename in ('ffmpeg', 'ffprobe'):
 106                         prefer_ffmpeg = True
 107
 108                 self._paths = dict(
 109                     (p, os.path.join(location, p)) for p in programs)
 110                 self._versions = dict(
 111                     (p, get_exe_version(self._paths[p], args=['-version']))
 112                     for p in programs)
 113         if self._versions is None:
 114             self._versions = dict(
 115                 (p, get_exe_version(p, args=['-version'])) for p in programs)
 116             self._paths = dict((p, p) for p in programs)
 117
 118         if prefer_ffmpeg:
 119             prefs = ('ffmpeg', 'avconv')
 120         else:
 121             prefs = ('avconv', 'ffmpeg')
 122         for p in prefs:
 123             if self._versions[p]:
 124                 self.basename = p
 125                 break
 126
 127         if prefer_ffmpeg:
 128             prefs = ('ffprobe', 'avprobe')
 129         else:
 130             prefs = ('avprobe', 'ffprobe')
 131         for p in prefs:
 132             if self._versions[p]:
 133                 self.probe_basename = p
 134                 break
 135
 136     @property
 137     def available(self):
 138         return self.basename is not None
 139
 140     @property
 141     def executable(self):
 142         return self._paths[self.basename]
 143
 144     @property
 145     def probe_available(self):
 146         return self.probe_basename is not None
 147
 148     @property
 149     def probe_executable(self):
 150         return self._paths[self.probe_basename]
 151
 152     def get_audio_codec(self, path):
 153         if not self.probe_available:
 154             raise PostProcessingError('ffprobe or avprobe not found. Please install one.')
 155         try:
 156             cmd = [
 157                 encodeFilename(self.probe_executable, True),
 158                 encodeArgument('-show_streams'),
 159                 encodeFilename(self._ffmpeg_filename_argument(path), True)]
 160             if self._downloader.params.get('verbose', False):
 161                 self._downloader.to_screen('[debug] %s command line: %s' % (self.basename, shell_quote(cmd)))
 162             handle = subprocess.Popen(cmd, stderr=compat_subprocess_get_DEVNULL(), stdout=subprocess.PIPE, stdin=subprocess.PIPE)
 163             output = handle.communicate()[0]
 164             if handle.wait() != 0:
 165                 return None
 166         except (IOError, OSError):
 167             return None
 168         audio_codec = None
 169         for line in output.decode('ascii', 'ignore').split('\n'):
 170             if line.startswith('codec_name='):
 171                 audio_codec = line.split('=')[1].strip()
 172             elif line.strip() == 'codec_type=audio' and audio_codec is not None:
 173                 return audio_codec
 174         return None
 175
 176     def run_ffmpeg_multiple_files(self, input_paths, out_path, opts):
 177         self.check_version()
 178
 179         oldest_mtime = min(
 180             os.stat(encodeFilename(path)).st_mtime for path in input_paths)
 181
 182         opts += self._configuration_args()
 183
 184         files_cmd = []
 185         for path in input_paths:
 186             files_cmd.extend([
 187                 encodeArgument('-i'),
 188                 encodeFilename(self._ffmpeg_filename_argument(path), True)
 189             ])
 190         cmd = ([encodeFilename(self.executable, True), encodeArgument('-y')] +
 191                files_cmd +
 192                [encodeArgument(o) for o in opts] +
 193                [encodeFilename(self._ffmpeg_filename_argument(out_path), True)])
 194
 195         if self._downloader.params.get('verbose', False):
 196             self._downloader.to_screen('[debug] ffmpeg command line: %s' % shell_quote(cmd))
 197         p = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE, stdin=subprocess.PIPE)
 198         stdout, stderr = p.communicate()
 199         if p.returncode != 0:
 200             stderr = stderr.decode('utf-8', 'replace')
 201             msg = stderr.strip().split('\n')[-1]
 202             raise FFmpegPostProcessorError(msg)
 203         self.try_utime(out_path, oldest_mtime, oldest_mtime)
 204
 205     def run_ffmpeg(self, path, out_path, opts):
 206         self.run_ffmpeg_multiple_files([path], out_path, opts)
 207
 208     def _ffmpeg_filename_argument(self, fn):
 209         # Always use 'file:' because the filename may contain ':' (ffmpeg
 210         # interprets that as a protocol) or can start with '-' (-- is broken in
 211         # ffmpeg, see https://ffmpeg.org/trac/ffmpeg/ticket/2127 for details)
 212         # Also leave '-' intact in order not to break streaming to stdout.
 213         return 'file:' + fn if fn != '-' else fn
 214
 215
 216 class FFmpegExtractAudioPP(FFmpegPostProcessor):
 217     def __init__(self, downloader=None, preferredcodec=None, preferredquality=None, nopostoverwrites=False):
 218         FFmpegPostProcessor.__init__(self, downloader)
 219         if preferredcodec is None:
 220             preferredcodec = 'best'
 221         self._preferredcodec = preferredcodec
 222         self._preferredquality = preferredquality
 223         self._nopostoverwrites = nopostoverwrites
 224
 225     def run_ffmpeg(self, path, out_path, codec, more_opts):
 226         if codec is None:
 227             acodec_opts = []
 228         else:
 229             acodec_opts = ['-acodec', codec]
 230         opts = ['-vn'] + acodec_opts + more_opts
 231         try:
 232             FFmpegPostProcessor.run_ffmpeg(self, path, out_path, opts)
 233         except FFmpegPostProcessorError as err:
 234             raise AudioConversionError(err.msg)
 235
 236     def run(self, information):
 237         path = information['filepath']
 238
 239         filecodec = self.get_audio_codec(path)
 240         if filecodec is None:
 241             raise PostProcessingError('WARNING: unable to obtain file audio codec with ffprobe')
 242
 243         more_opts = []
 244         if self._preferredcodec == 'best' or self._preferredcodec == filecodec or (self._preferredcodec == 'm4a' and filecodec == 'aac'):
 245             if filecodec == 'aac' and self._preferredcodec in ['m4a', 'best']:
 246                 # Lossless, but in another container
 247                 acodec = 'copy'
 248                 extension = 'm4a'
 249                 more_opts = ['-bsf:a', 'aac_adtstoasc']
 250             elif filecodec in ['aac', 'flac', 'mp3', 'vorbis', 'opus']:
 251                 # Lossless if possible
 252                 acodec = 'copy'
 253                 extension = filecodec
 254                 if filecodec == 'aac':
 255                     more_opts = ['-f', 'adts']
 256                 if filecodec == 'vorbis':
 257                     extension = 'ogg'
 258             else:
 259                 # MP3 otherwise.
 260                 acodec = 'libmp3lame'
 261                 extension = 'mp3'
 262                 more_opts = []
 263                 if self._preferredquality is not None:
 264                     if int(self._preferredquality) < 10:
 265                         more_opts += ['-q:a', self._preferredquality]
 266                     else:
 267                         more_opts += ['-b:a', self._preferredquality + 'k']
 268         else:
 269             # We convert the audio (lossy if codec is lossy)
 270             acodec = ACODECS[self._preferredcodec]
 271             extension = self._preferredcodec
 272             more_opts = []
 273             if self._preferredquality is not None:
 274                 # The opus codec doesn't support the -aq option
 275                 if int(self._preferredquality) < 10 and extension != 'opus':
 276                     more_opts += ['-q:a', self._preferredquality]
 277                 else:
 278                     more_opts += ['-b:a', self._preferredquality + 'k']
 279             if self._preferredcodec == 'aac':
 280                 more_opts += ['-f', 'adts']
 281             if self._preferredcodec == 'm4a':
 282                 more_opts += ['-bsf:a', 'aac_adtstoasc']
 283             if self._preferredcodec == 'vorbis':
 284                 extension = 'ogg'
 285             if self._preferredcodec == 'wav':
 286                 extension = 'wav'
 287                 more_opts += ['-f', 'wav']
 288
 289         prefix, sep, ext = path.rpartition('.')  # not os.path.splitext, since the latter does not work on unicode in all setups
 290         new_path = prefix + sep + extension
 291
 292         information['filepath'] = new_path
 293         information['ext'] = extension
 294
 295         # If we download foo.mp3 and convert it to... foo.mp3, then don't delete foo.mp3, silly.
 296         if (new_path == path or
 297                 (self._nopostoverwrites and os.path.exists(encodeFilename(new_path)))):
 298             self._downloader.to_screen('[ffmpeg] Post-process file %s exists, skipping' % new_path)
 299             return [], information
 300
 301         try:
 302             self._downloader.to_screen('[ffmpeg] Destination: ' + new_path)
 303             self.run_ffmpeg(path, new_path, acodec, more_opts)
 304         except AudioConversionError as e:
 305             raise PostProcessingError(
 306                 'audio conversion failed: ' + e.msg)
 307         except Exception:
 308             raise PostProcessingError('error running ' + self.basename)
 309
 310         # Try to update the date time for extracted audio file.
 311         if information.get('filetime') is not None:
 312             self.try_utime(
 313                 new_path, time.time(), information['filetime'],
 314                 errnote='Cannot update utime of audio file')
 315
 316         return [path], information
 317
 318
 319 class FFmpegVideoConvertorPP(FFmpegPostProcessor):
 320     def __init__(self, downloader=None, preferedformat=None):
 321         super(FFmpegVideoConvertorPP, self).__init__(downloader)
 322         self._preferedformat = preferedformat
 323
 324     def run(self, information):
 325         path = information['filepath']
 326         if information['ext'] == self._preferedformat:
 327             self._downloader.to_screen('[ffmpeg] Not converting video file %s - already is in target format %s' % (path, self._preferedformat))
 328             return [], information
 329         options = []
 330         if self._preferedformat == 'avi':
 331             options.extend(['-c:v', 'libxvid', '-vtag', 'XVID'])
 332         prefix, sep, ext = path.rpartition('.')
 333         outpath = prefix + sep + self._preferedformat
 334         self._downloader.to_screen('[' + 'ffmpeg' + '] Converting video from %s to %s, Destination: ' % (information['ext'], self._preferedformat) + outpath)
 335         self.run_ffmpeg(path, outpath, options)
 336         information['filepath'] = outpath
 337         information['format'] = self._preferedformat
 338         information['ext'] = self._preferedformat
 339         return [path], information
 340
 341
 342 class FFmpegEmbedSubtitlePP(FFmpegPostProcessor):
 343     def run(self, information):
 344         if information['ext'] not in ('mp4', 'webm', 'mkv'):
 345             self._downloader.to_screen('[ffmpeg] Subtitles can only be embedded in mp4, webm or mkv files')
 346             return [], information
 347         subtitles = information.get('requested_subtitles')
 348         if not subtitles:
 349             self._downloader.to_screen('[ffmpeg] There aren\'t any subtitles to embed')
 350             return [], information
 351
 352         filename = information['filepath']
 353
 354         ext = information['ext']
 355         sub_langs = []
 356         sub_filenames = []
 357         webm_vtt_warn = False
 358
 359         for lang, sub_info in subtitles.items():
 360             sub_ext = sub_info['ext']
 361             if ext != 'webm' or ext == 'webm' and sub_ext == 'vtt':
 362                 sub_langs.append(lang)
 363                 sub_filenames.append(subtitles_filename(filename, lang, sub_ext))
 364             else:
 365                 if not webm_vtt_warn and ext == 'webm' and sub_ext != 'vtt':
 366                     webm_vtt_warn = True
 367                     self._downloader.to_screen('[ffmpeg] Only WebVTT subtitles can be embedded in webm files')
 368
 369         if not sub_langs:
 370             return [], information
 371
 372         input_files = [filename] + sub_filenames
 373
 374         opts = [
 375             '-map', '0',
 376             '-c', 'copy',
 377             # Don't copy the existing subtitles, we may be running the
 378             # postprocessor a second time
 379             '-map', '-0:s',
 380         ]
 381         if information['ext'] == 'mp4':
 382             opts += ['-c:s', 'mov_text']
 383         for (i, lang) in enumerate(sub_langs):
 384             opts.extend(['-map', '%d:0' % (i + 1)])
 385             lang_code = ISO639Utils.short2long(lang)
 386             if lang_code is not None:
 387                 opts.extend(['-metadata:s:s:%d' % i, 'language=%s' % lang_code])
 388
 389         temp_filename = prepend_extension(filename, 'temp')
 390         self._downloader.to_screen('[ffmpeg] Embedding subtitles in \'%s\'' % filename)
 391         self.run_ffmpeg_multiple_files(input_files, temp_filename, opts)
 392         os.remove(encodeFilename(filename))
 393         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 394
 395         return sub_filenames, information
 396
 397
 398 class FFmpegMetadataPP(FFmpegPostProcessor):
 399     def run(self, info):
 400         metadata = {}
 401
 402         def add(meta_list, info_list=None):
 403             if not info_list:
 404                 info_list = meta_list
 405             if not isinstance(meta_list, (list, tuple)):
 406                 meta_list = (meta_list,)
 407             if not isinstance(info_list, (list, tuple)):
 408                 info_list = (info_list,)
 409             for info_f in info_list:
 410                 if info.get(info_f) is not None:
 411                     for meta_f in meta_list:
 412                         metadata[meta_f] = info[info_f]
 413                     break
 414
 415         add('title', ('track', 'title'))
 416         add('date', 'upload_date')
 417         add(('description', 'comment'), 'description')
 418         add('purl', 'webpage_url')
 419         add('track', 'track_number')
 420         add('artist', ('artist', 'creator', 'uploader', 'uploader_id'))
 421         add('genre')
 422         add('album')
 423         add('album_artist')
 424         add('disc', 'disc_number')
 425
 426         if not metadata:
 427             self._downloader.to_screen('[ffmpeg] There isn\'t any metadata to add')
 428             return [], info
 429
 430         filename = info['filepath']
 431         temp_filename = prepend_extension(filename, 'temp')
 432
 433         if info['ext'] == 'm4a':
 434             options = ['-vn', '-acodec', 'copy']
 435         else:
 436             options = ['-c', 'copy']
 437
 438         for (name, value) in metadata.items():
 439             options.extend(['-metadata', '%s=%s' % (name, value)])
 440
 441         self._downloader.to_screen('[ffmpeg] Adding metadata to \'%s\'' % filename)
 442         self.run_ffmpeg(filename, temp_filename, options)
 443         os.remove(encodeFilename(filename))
 444         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 445         return [], info
 446
 447
 448 class FFmpegMergerPP(FFmpegPostProcessor):
 449     def run(self, info):
 450         filename = info['filepath']
 451         temp_filename = prepend_extension(filename, 'temp')
 452         args = ['-c', 'copy', '-map', '0:v:0', '-map', '1:a:0']
 453         self._downloader.to_screen('[ffmpeg] Merging formats into "%s"' % filename)
 454         self.run_ffmpeg_multiple_files(info['__files_to_merge'], temp_filename, args)
 455         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 456         return info['__files_to_merge'], info
 457
 458     def can_merge(self):
 459         # TODO: figure out merge-capable ffmpeg version
 460         if self.basename != 'avconv':
 461             return True
 462
 463         required_version = '10-0'
 464         if is_outdated_version(
 465                 self._versions[self.basename], required_version):
 466             warning = ('Your copy of %s is outdated and unable to properly mux separate video and audio files, '
 467                        'youtube-dl will download single file media. '
 468                        'Update %s to version %s or newer to fix this.') % (
 469                            self.basename, self.basename, required_version)
 470             if self._downloader:
 471                 self._downloader.report_warning(warning)
 472             return False
 473         return True
 474
 475
 476 class FFmpegFixupStretchedPP(FFmpegPostProcessor):
 477     def run(self, info):
 478         stretched_ratio = info.get('stretched_ratio')
 479         if stretched_ratio is None or stretched_ratio == 1:
 480             return [], info
 481
 482         filename = info['filepath']
 483         temp_filename = prepend_extension(filename, 'temp')
 484
 485         options = ['-c', 'copy', '-aspect', '%f' % stretched_ratio]
 486         self._downloader.to_screen('[ffmpeg] Fixing aspect ratio in "%s"' % filename)
 487         self.run_ffmpeg(filename, temp_filename, options)
 488
 489         os.remove(encodeFilename(filename))
 490         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 491
 492         return [], info
 493
 494
 495 class FFmpegFixupM4aPP(FFmpegPostProcessor):
 496     def run(self, info):
 497         if info.get('container') != 'm4a_dash':
 498             return [], info
 499
 500         filename = info['filepath']
 501         temp_filename = prepend_extension(filename, 'temp')
 502
 503         options = ['-c', 'copy', '-f', 'mp4']
 504         self._downloader.to_screen('[ffmpeg] Correcting container in "%s"' % filename)
 505         self.run_ffmpeg(filename, temp_filename, options)
 506
 507         os.remove(encodeFilename(filename))
 508         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 509
 510         return [], info
 511
 512
 513 class FFmpegFixupM3u8PP(FFmpegPostProcessor):
 514     def run(self, info):
 515         filename = info['filepath']
 516         if self.get_audio_codec(filename) == 'aac':
 517             temp_filename = prepend_extension(filename, 'temp')
 518
 519             options = ['-c', 'copy', '-f', 'mp4', '-bsf:a', 'aac_adtstoasc']
 520             self._downloader.to_screen('[ffmpeg] Fixing malformated aac bitstream in "%s"' % filename)
 521             self.run_ffmpeg(filename, temp_filename, options)
 522
 523             os.remove(encodeFilename(filename))
 524             os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 525         return [], info
 526
 527
 528 class FFmpegSubtitlesConvertorPP(FFmpegPostProcessor):
 529     def __init__(self, downloader=None, format=None):
 530         super(FFmpegSubtitlesConvertorPP, self).__init__(downloader)
 531         self.format = format
 532
 533     def run(self, info):
 534         subs = info.get('requested_subtitles')
 535         filename = info['filepath']
 536         new_ext = self.format
 537         new_format = new_ext
 538         if new_format == 'vtt':
 539             new_format = 'webvtt'
 540         if subs is None:
 541             self._downloader.to_screen('[ffmpeg] There aren\'t any subtitles to convert')
 542             return [], info
 543         self._downloader.to_screen('[ffmpeg] Converting subtitles')
 544         sub_filenames = []
 545         for lang, sub in subs.items():
 546             ext = sub['ext']
 547             if ext == new_ext:
 548                 self._downloader.to_screen(
 549                     '[ffmpeg] Subtitle file for %s is already in the requested format' % new_ext)
 550                 continue
 551             old_file = subtitles_filename(filename, lang, ext)
 552             sub_filenames.append(old_file)
 553             new_file = subtitles_filename(filename, lang, new_ext)
 554
 555             if ext == 'dfxp' or ext == 'ttml' or ext == 'tt':
 556                 self._downloader.report_warning(
 557                     'You have requested to convert dfxp (TTML) subtitles into another format, '
 558                     'which results in style information loss')
 559
 560                 dfxp_file = old_file
 561                 srt_file = subtitles_filename(filename, lang, 'srt')
 562
 563                 with io.open(dfxp_file, 'rt', encoding='utf-8') as f:
 564                     srt_data = dfxp2srt(f.read())
 565
 566                 with io.open(srt_file, 'wt', encoding='utf-8') as f:
 567                     f.write(srt_data)
 568                 old_file = srt_file
 569
 570                 subs[lang] = {
 571                     'ext': 'srt',
 572                     'data': srt_data
 573                 }
 574
 575                 if new_ext == 'srt':
 576                     continue
 577                 else:
 578                     sub_filenames.append(srt_file)
 579
 580             self.run_ffmpeg(old_file, new_file, ['-f', new_format])
 581
 582             with io.open(new_file, 'rt', encoding='utf-8') as f:
 583                 subs[lang] = {
 584                     'ext': new_ext,
 585                     'data': f.read(),
 586                 }
 587
 588         return sub_filenames, info