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  57             warning = 'Your copy of %s is outdated, update %s to version %s or newer if you encounter any errors.'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         return 'file:' + fn
 179
 180
 181 class FFmpegExtractAudioPP(FFmpegPostProcessor):
 182     def __init__(self, downloader=None, preferredcodec=None, preferredquality=None, nopostoverwrites=False):
 183         FFmpegPostProcessor.__init__(self, downloader)
 184         if preferredcodec is None:
 185             preferredcodec = 'best'
 186         self._preferredcodec = preferredcodec
 187         self._preferredquality = preferredquality
 188         self._nopostoverwrites = nopostoverwrites
 189
 190     def get_audio_codec(self, path):
 191
 192         if not self.probe_available:
 193             raise PostProcessingError('ffprobe or avprobe not found. Please install one.')
 194         try:
 195             cmd = [
 196                 encodeFilename(self.probe_executable, True),
 197                 encodeArgument('-show_streams'),
 198                 encodeFilename(self._ffmpeg_filename_argument(path), True)]
 199             if self._downloader.params.get('verbose', False):
 200                 self._downloader.to_screen('[debug] %s command line: %s' % (self.basename, shell_quote(cmd)))
 201             handle = subprocess.Popen(cmd, stderr=compat_subprocess_get_DEVNULL(), stdout=subprocess.PIPE, stdin=subprocess.PIPE)
 202             output = handle.communicate()[0]
 203             if handle.wait() != 0:
 204                 return None
 205         except (IOError, OSError):
 206             return None
 207         audio_codec = None
 208         for line in output.decode('ascii', 'ignore').split('\n'):
 209             if line.startswith('codec_name='):
 210                 audio_codec = line.split('=')[1].strip()
 211             elif line.strip() == 'codec_type=audio' and audio_codec is not None:
 212                 return audio_codec
 213         return None
 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         # If we download foo.mp3 and convert it to... foo.mp3, then don't delete foo.mp3, silly.
 283         if (new_path == path or
 284                 (self._nopostoverwrites and os.path.exists(encodeFilename(new_path)))):
 285             self._downloader.to_screen('[ffmpeg] Post-process file %s exists, skipping' % new_path)
 286             return [], information
 287
 288         try:
 289             self._downloader.to_screen('[ffmpeg] Destination: ' + new_path)
 290             self.run_ffmpeg(path, new_path, acodec, more_opts)
 291         except AudioConversionError as e:
 292             raise PostProcessingError(
 293                 'audio conversion failed: ' + e.msg)
 294         except Exception:
 295             raise PostProcessingError('error running ' + self.basename)
 296
 297         # Try to update the date time for extracted audio file.
 298         if information.get('filetime') is not None:
 299             self.try_utime(
 300                 new_path, time.time(), information['filetime'],
 301                 errnote='Cannot update utime of audio file')
 302
 303         information['filepath'] = new_path
 304         information['ext'] = extension
 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', 'mkv']:
 335             self._downloader.to_screen('[ffmpeg] Subtitles can only be embedded in mp4 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         sub_langs = list(subtitles.keys())
 343         filename = information['filepath']
 344         sub_filenames = [subtitles_filename(filename, lang, sub_info['ext']) for lang, sub_info in subtitles.items()]
 345         input_files = [filename] + sub_filenames
 346
 347         opts = [
 348             '-map', '0',
 349             '-c', 'copy',
 350             # Don't copy the existing subtitles, we may be running the
 351             # postprocessor a second time
 352             '-map', '-0:s',
 353         ]
 354         if information['ext'] == 'mp4':
 355             opts += ['-c:s', 'mov_text']
 356         for (i, lang) in enumerate(sub_langs):
 357             opts.extend(['-map', '%d:0' % (i + 1)])
 358             lang_code = ISO639Utils.short2long(lang)
 359             if lang_code is not None:
 360                 opts.extend(['-metadata:s:s:%d' % i, 'language=%s' % lang_code])
 361
 362         temp_filename = prepend_extension(filename, 'temp')
 363         self._downloader.to_screen('[ffmpeg] Embedding subtitles in \'%s\'' % filename)
 364         self.run_ffmpeg_multiple_files(input_files, temp_filename, opts)
 365         os.remove(encodeFilename(filename))
 366         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 367
 368         return sub_filenames, information
 369
 370
 371 class FFmpegMetadataPP(FFmpegPostProcessor):
 372     def run(self, info):
 373         metadata = {}
 374         if info.get('title') is not None:
 375             metadata['title'] = info['title']
 376         if info.get('upload_date') is not None:
 377             metadata['date'] = info['upload_date']
 378         if info.get('artist') is not None:
 379             metadata['artist'] = info['artist']
 380         elif info.get('uploader') is not None:
 381             metadata['artist'] = info['uploader']
 382         elif info.get('uploader_id') is not None:
 383             metadata['artist'] = info['uploader_id']
 384         if info.get('description') is not None:
 385             metadata['description'] = info['description']
 386             metadata['comment'] = info['description']
 387         if info.get('webpage_url') is not None:
 388             metadata['purl'] = info['webpage_url']
 389         if info.get('album') is not None:
 390             metadata['album'] = info['album']
 391
 392         if not metadata:
 393             self._downloader.to_screen('[ffmpeg] There isn\'t any metadata to add')
 394             return [], info
 395
 396         filename = info['filepath']
 397         temp_filename = prepend_extension(filename, 'temp')
 398
 399         if info['ext'] == 'm4a':
 400             options = ['-vn', '-acodec', 'copy']
 401         else:
 402             options = ['-c', 'copy']
 403
 404         for (name, value) in metadata.items():
 405             options.extend(['-metadata', '%s=%s' % (name, value)])
 406
 407         self._downloader.to_screen('[ffmpeg] Adding metadata to \'%s\'' % filename)
 408         self.run_ffmpeg(filename, temp_filename, options)
 409         os.remove(encodeFilename(filename))
 410         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 411         return [], info
 412
 413
 414 class FFmpegMergerPP(FFmpegPostProcessor):
 415     def run(self, info):
 416         filename = info['filepath']
 417         temp_filename = prepend_extension(filename, 'temp')
 418         args = ['-c', 'copy', '-map', '0:v:0', '-map', '1:a:0']
 419         self._downloader.to_screen('[ffmpeg] Merging formats into "%s"' % filename)
 420         self.run_ffmpeg_multiple_files(info['__files_to_merge'], temp_filename, args)
 421         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 422         return info['__files_to_merge'], info
 423
 424     def can_merge(self):
 425         # TODO: figure out merge-capable ffmpeg version
 426         if self.basename != 'avconv':
 427             return True
 428
 429         required_version = '10-0'
 430         if is_outdated_version(
 431                 self._versions[self.basename], required_version):
 432             warning = ('Your copy of %s is outdated and unable to properly mux separate video and audio files, '
 433                        'youtube-dl will download single file media. '
 434                        'Update %s to version %s or newer to fix this.') % (
 435                            self.basename, self.basename, required_version)
 436             if self._downloader:
 437                 self._downloader.report_warning(warning)
 438             return False
 439         return True
 440
 441
 442 class FFmpegFixupStretchedPP(FFmpegPostProcessor):
 443     def run(self, info):
 444         stretched_ratio = info.get('stretched_ratio')
 445         if stretched_ratio is None or stretched_ratio == 1:
 446             return [], info
 447
 448         filename = info['filepath']
 449         temp_filename = prepend_extension(filename, 'temp')
 450
 451         options = ['-c', 'copy', '-aspect', '%f' % stretched_ratio]
 452         self._downloader.to_screen('[ffmpeg] Fixing aspect ratio in "%s"' % filename)
 453         self.run_ffmpeg(filename, temp_filename, options)
 454
 455         os.remove(encodeFilename(filename))
 456         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 457
 458         return [], info
 459
 460
 461 class FFmpegFixupM4aPP(FFmpegPostProcessor):
 462     def run(self, info):
 463         if info.get('container') != 'm4a_dash':
 464             return [], info
 465
 466         filename = info['filepath']
 467         temp_filename = prepend_extension(filename, 'temp')
 468
 469         options = ['-c', 'copy', '-f', 'mp4']
 470         self._downloader.to_screen('[ffmpeg] Correcting container in "%s"' % filename)
 471         self.run_ffmpeg(filename, temp_filename, options)
 472
 473         os.remove(encodeFilename(filename))
 474         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 475
 476         return [], info
 477
 478
 479 class FFmpegFixupM3u8PP(FFmpegPostProcessor):
 480     def run(self, info):
 481         filename = info['filepath']
 482         temp_filename = prepend_extension(filename, 'temp')
 483
 484         options = ['-c', 'copy', '-f', 'mp4', '-bsf:a', 'aac_adtstoasc']
 485         self._downloader.to_screen('[ffmpeg] Fixing malformated aac bitstream in "%s"' % filename)
 486         self.run_ffmpeg(filename, temp_filename, options)
 487
 488         os.remove(encodeFilename(filename))
 489         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 490
 491         return [], info
 492
 493
 494 class FFmpegSubtitlesConvertorPP(FFmpegPostProcessor):
 495     def __init__(self, downloader=None, format=None):
 496         super(FFmpegSubtitlesConvertorPP, self).__init__(downloader)
 497         self.format = format
 498
 499     def run(self, info):
 500         subs = info.get('requested_subtitles')
 501         filename = info['filepath']
 502         new_ext = self.format
 503         new_format = new_ext
 504         if new_format == 'vtt':
 505             new_format = 'webvtt'
 506         if subs is None:
 507             self._downloader.to_screen('[ffmpeg] There aren\'t any subtitles to convert')
 508             return [], info
 509         self._downloader.to_screen('[ffmpeg] Converting subtitles')
 510         sub_filenames = []
 511         for lang, sub in subs.items():
 512             ext = sub['ext']
 513             if ext == new_ext:
 514                 self._downloader.to_screen(
 515                     '[ffmpeg] Subtitle file for %s is already in the requested'
 516                     'format' % new_ext)
 517                 continue
 518             old_file = subtitles_filename(filename, lang, ext)
 519             sub_filenames.append(old_file)
 520             new_file = subtitles_filename(filename, lang, new_ext)
 521
 522             if ext == 'dfxp' or ext == 'ttml':
 523                 self._downloader.report_warning(
 524                     'You have requested to convert dfxp (TTML) subtitles into another format, '
 525                     'which results in style information loss')
 526
 527                 dfxp_file = old_file
 528                 srt_file = subtitles_filename(filename, lang, 'srt')
 529
 530                 with io.open(dfxp_file, 'rt', encoding='utf-8') as f:
 531                     srt_data = dfxp2srt(f.read())
 532
 533                 with io.open(srt_file, 'wt', encoding='utf-8') as f:
 534                     f.write(srt_data)
 535                 old_file = srt_file
 536
 537                 subs[lang] = {
 538                     'ext': 'srt',
 539                     'data': srt_data
 540                 }
 541
 542                 if new_ext == 'srt':
 543                     continue
 544                 else:
 545                     sub_filenames.append(srt_file)
 546
 547             self.run_ffmpeg(old_file, new_file, ['-f', new_format])
 548
 549             with io.open(new_file, 'rt', encoding='utf-8') as f:
 550                 subs[lang] = {
 551                     'ext': new_ext,
 552                     'data': f.read(),
 553                 }
 554
 555         return sub_filenames, info