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 class FFmpegPostProcessorError(PostProcessingError):
  29     pass
  30
  31
  32 class FFmpegPostProcessor(PostProcessor):
  33     def __init__(self, downloader=None):
  34         PostProcessor.__init__(self, downloader)
  35         self._determine_executables()
  36
  37     def check_version(self):
  38         if not self.available:
  39             raise FFmpegPostProcessorError('ffmpeg or avconv not found. Please install one.')
  40
  41         required_version = '10-0' if self.basename == 'avconv' else '1.0'
  42         if is_outdated_version(
  43                 self._versions[self.basename], required_version):
  44             warning = 'Your copy of %s is outdated, update %s to version %s or newer if you encounter any errors.' % (
  45                 self.basename, self.basename, required_version)
  46             if self._downloader:
  47                 self._downloader.report_warning(warning)
  48
  49     @staticmethod
  50     def get_versions(downloader=None):
  51         return FFmpegPostProcessor(downloader)._versions
  52
  53     def _determine_executables(self):
  54         programs = ['avprobe', 'avconv', 'ffmpeg', 'ffprobe']
  55         prefer_ffmpeg = False
  56
  57         self.basename = None
  58         self.probe_basename = None
  59
  60         self._paths = None
  61         self._versions = None
  62         if self._downloader:
  63             prefer_ffmpeg = self._downloader.params.get('prefer_ffmpeg', False)
  64             location = self._downloader.params.get('ffmpeg_location')
  65             if location is not None:
  66                 if not os.path.exists(location):
  67                     self._downloader.report_warning(
  68                         'ffmpeg-location %s does not exist! '
  69                         'Continuing without avconv/ffmpeg.' % (location))
  70                     self._versions = {}
  71                     return
  72                 elif not os.path.isdir(location):
  73                     basename = os.path.splitext(os.path.basename(location))[0]
  74                     if basename not in programs:
  75                         self._downloader.report_warning(
  76                             'Cannot identify executable %s, its basename should be one of %s. '
  77                             'Continuing without avconv/ffmpeg.' %
  78                             (location, ', '.join(programs)))
  79                         self._versions = {}
  80                         return None
  81                     location = os.path.dirname(os.path.abspath(location))
  82                     if basename in ('ffmpeg', 'ffprobe'):
  83                         prefer_ffmpeg = True
  84
  85                 self._paths = dict(
  86                     (p, os.path.join(location, p)) for p in programs)
  87                 self._versions = dict(
  88                     (p, get_exe_version(self._paths[p], args=['-version']))
  89                     for p in programs)
  90         if self._versions is None:
  91             self._versions = dict(
  92                 (p, get_exe_version(p, args=['-version'])) for p in programs)
  93             self._paths = dict((p, p) for p in programs)
  94
  95         if prefer_ffmpeg:
  96             prefs = ('ffmpeg', 'avconv')
  97         else:
  98             prefs = ('avconv', 'ffmpeg')
  99         for p in prefs:
 100             if self._versions[p]:
 101                 self.basename = p
 102                 break
 103
 104         if prefer_ffmpeg:
 105             prefs = ('ffprobe', 'avprobe')
 106         else:
 107             prefs = ('avprobe', 'ffprobe')
 108         for p in prefs:
 109             if self._versions[p]:
 110                 self.probe_basename = p
 111                 break
 112
 113     @property
 114     def available(self):
 115         return self.basename is not None
 116
 117     @property
 118     def executable(self):
 119         return self._paths[self.basename]
 120
 121     @property
 122     def probe_available(self):
 123         return self.probe_basename is not None
 124
 125     @property
 126     def probe_executable(self):
 127         return self._paths[self.probe_basename]
 128
 129     def run_ffmpeg_multiple_files(self, input_paths, out_path, opts):
 130         self.check_version()
 131
 132         oldest_mtime = min(
 133             os.stat(encodeFilename(path)).st_mtime for path in input_paths)
 134
 135         opts += self._configuration_args()
 136
 137         files_cmd = []
 138         for path in input_paths:
 139             files_cmd.extend([
 140                 encodeArgument('-i'),
 141                 encodeFilename(self._ffmpeg_filename_argument(path), True)
 142             ])
 143         cmd = ([encodeFilename(self.executable, True), encodeArgument('-y')] +
 144                files_cmd +
 145                [encodeArgument(o) for o in opts] +
 146                [encodeFilename(self._ffmpeg_filename_argument(out_path), True)])
 147
 148         if self._downloader.params.get('verbose', False):
 149             self._downloader.to_screen('[debug] ffmpeg command line: %s' % shell_quote(cmd))
 150         p = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE, stdin=subprocess.PIPE)
 151         stdout, stderr = p.communicate()
 152         if p.returncode != 0:
 153             stderr = stderr.decode('utf-8', 'replace')
 154             msg = stderr.strip().split('\n')[-1]
 155             raise FFmpegPostProcessorError(msg)
 156         self.try_utime(out_path, oldest_mtime, oldest_mtime)
 157
 158     def run_ffmpeg(self, path, out_path, opts):
 159         self.run_ffmpeg_multiple_files([path], out_path, opts)
 160
 161     def _ffmpeg_filename_argument(self, fn):
 162         # Always use 'file:' because the filename may contain ':' (ffmpeg
 163         # interprets that as a protocol) or can start with '-' (-- is broken in
 164         # ffmpeg, see https://ffmpeg.org/trac/ffmpeg/ticket/2127 for details)
 165         return 'file:' + fn
 166
 167
 168 class FFmpegExtractAudioPP(FFmpegPostProcessor):
 169     def __init__(self, downloader=None, preferredcodec=None, preferredquality=None, nopostoverwrites=False):
 170         FFmpegPostProcessor.__init__(self, downloader)
 171         if preferredcodec is None:
 172             preferredcodec = 'best'
 173         self._preferredcodec = preferredcodec
 174         self._preferredquality = preferredquality
 175         self._nopostoverwrites = nopostoverwrites
 176
 177     def get_audio_codec(self, path):
 178
 179         if not self.probe_available:
 180             raise PostProcessingError('ffprobe or avprobe not found. Please install one.')
 181         try:
 182             cmd = [
 183                 encodeFilename(self.probe_executable, True),
 184                 encodeArgument('-show_streams'),
 185                 encodeFilename(self._ffmpeg_filename_argument(path), True)]
 186             if self._downloader.params.get('verbose', False):
 187                 self._downloader.to_screen('[debug] %s command line: %s' % (self.basename, shell_quote(cmd)))
 188             handle = subprocess.Popen(cmd, stderr=compat_subprocess_get_DEVNULL(), stdout=subprocess.PIPE, stdin=subprocess.PIPE)
 189             output = handle.communicate()[0]
 190             if handle.wait() != 0:
 191                 return None
 192         except (IOError, OSError):
 193             return None
 194         audio_codec = None
 195         for line in output.decode('ascii', 'ignore').split('\n'):
 196             if line.startswith('codec_name='):
 197                 audio_codec = line.split('=')[1].strip()
 198             elif line.strip() == 'codec_type=audio' and audio_codec is not None:
 199                 return audio_codec
 200         return None
 201
 202     def run_ffmpeg(self, path, out_path, codec, more_opts):
 203         if codec is None:
 204             acodec_opts = []
 205         else:
 206             acodec_opts = ['-acodec', codec]
 207         opts = ['-vn'] + acodec_opts + more_opts
 208         try:
 209             FFmpegPostProcessor.run_ffmpeg(self, path, out_path, opts)
 210         except FFmpegPostProcessorError as err:
 211             raise AudioConversionError(err.msg)
 212
 213     def run(self, information):
 214         path = information['filepath']
 215
 216         filecodec = self.get_audio_codec(path)
 217         if filecodec is None:
 218             raise PostProcessingError('WARNING: unable to obtain file audio codec with ffprobe')
 219
 220         more_opts = []
 221         if self._preferredcodec == 'best' or self._preferredcodec == filecodec or (self._preferredcodec == 'm4a' and filecodec == 'aac'):
 222             if filecodec == 'aac' and self._preferredcodec in ['m4a', 'best']:
 223                 # Lossless, but in another container
 224                 acodec = 'copy'
 225                 extension = 'm4a'
 226                 more_opts = ['-bsf:a', 'aac_adtstoasc']
 227             elif filecodec in ['aac', 'mp3', 'vorbis', 'opus']:
 228                 # Lossless if possible
 229                 acodec = 'copy'
 230                 extension = filecodec
 231                 if filecodec == 'aac':
 232                     more_opts = ['-f', 'adts']
 233                 if filecodec == 'vorbis':
 234                     extension = 'ogg'
 235             else:
 236                 # MP3 otherwise.
 237                 acodec = 'libmp3lame'
 238                 extension = 'mp3'
 239                 more_opts = []
 240                 if self._preferredquality is not None:
 241                     if int(self._preferredquality) < 10:
 242                         more_opts += ['-q:a', self._preferredquality]
 243                     else:
 244                         more_opts += ['-b:a', self._preferredquality + 'k']
 245         else:
 246             # We convert the audio (lossy)
 247             acodec = {'mp3': 'libmp3lame', 'aac': 'aac', 'm4a': 'aac', 'opus': 'opus', 'vorbis': 'libvorbis', 'wav': None}[self._preferredcodec]
 248             extension = self._preferredcodec
 249             more_opts = []
 250             if self._preferredquality is not None:
 251                 # The opus codec doesn't support the -aq option
 252                 if int(self._preferredquality) < 10 and extension != 'opus':
 253                     more_opts += ['-q:a', self._preferredquality]
 254                 else:
 255                     more_opts += ['-b:a', self._preferredquality + 'k']
 256             if self._preferredcodec == 'aac':
 257                 more_opts += ['-f', 'adts']
 258             if self._preferredcodec == 'm4a':
 259                 more_opts += ['-bsf:a', 'aac_adtstoasc']
 260             if self._preferredcodec == 'vorbis':
 261                 extension = 'ogg'
 262             if self._preferredcodec == 'wav':
 263                 extension = 'wav'
 264                 more_opts += ['-f', 'wav']
 265
 266         prefix, sep, ext = path.rpartition('.')  # not os.path.splitext, since the latter does not work on unicode in all setups
 267         new_path = prefix + sep + extension
 268
 269         # If we download foo.mp3 and convert it to... foo.mp3, then don't delete foo.mp3, silly.
 270         if (new_path == path or
 271                 (self._nopostoverwrites and os.path.exists(encodeFilename(new_path)))):
 272             self._downloader.to_screen('[ffmpeg] Post-process file %s exists, skipping' % new_path)
 273             return [], information
 274
 275         try:
 276             self._downloader.to_screen('[ffmpeg] Destination: ' + new_path)
 277             self.run_ffmpeg(path, new_path, acodec, more_opts)
 278         except AudioConversionError as e:
 279             raise PostProcessingError(
 280                 'audio conversion failed: ' + e.msg)
 281         except Exception:
 282             raise PostProcessingError('error running ' + self.basename)
 283
 284         # Try to update the date time for extracted audio file.
 285         if information.get('filetime') is not None:
 286             self.try_utime(
 287                 new_path, time.time(), information['filetime'],
 288                 errnote='Cannot update utime of audio file')
 289
 290         information['filepath'] = new_path
 291         information['ext'] = extension
 292
 293         return [path], information
 294
 295
 296 class FFmpegVideoConvertorPP(FFmpegPostProcessor):
 297     def __init__(self, downloader=None, preferedformat=None):
 298         super(FFmpegVideoConvertorPP, self).__init__(downloader)
 299         self._preferedformat = preferedformat
 300
 301     def run(self, information):
 302         path = information['filepath']
 303         if information['ext'] == self._preferedformat:
 304             self._downloader.to_screen('[ffmpeg] Not converting video file %s - already is in target format %s' % (path, self._preferedformat))
 305             return [], information
 306         options = []
 307         if self._preferedformat == 'avi':
 308             options.extend(['-c:v', 'libxvid', '-vtag', 'XVID'])
 309         prefix, sep, ext = path.rpartition('.')
 310         outpath = prefix + sep + self._preferedformat
 311         self._downloader.to_screen('[' + 'ffmpeg' + '] Converting video from %s to %s, Destination: ' % (information['ext'], self._preferedformat) + outpath)
 312         self.run_ffmpeg(path, outpath, options)
 313         information['filepath'] = outpath
 314         information['format'] = self._preferedformat
 315         information['ext'] = self._preferedformat
 316         return [path], information
 317
 318
 319 class FFmpegEmbedSubtitlePP(FFmpegPostProcessor):
 320     def run(self, information):
 321         if information['ext'] not in ['mp4', 'mkv']:
 322             self._downloader.to_screen('[ffmpeg] Subtitles can only be embedded in mp4 or mkv files')
 323             return [], information
 324         subtitles = information.get('requested_subtitles')
 325         if not subtitles:
 326             self._downloader.to_screen('[ffmpeg] There aren\'t any subtitles to embed')
 327             return [], information
 328
 329         sub_langs = list(subtitles.keys())
 330         filename = information['filepath']
 331         sub_filenames = [subtitles_filename(filename, lang, sub_info['ext']) for lang, sub_info in subtitles.items()]
 332         input_files = [filename] + sub_filenames
 333
 334         opts = [
 335             '-map', '0',
 336             '-c', 'copy',
 337             # Don't copy the existing subtitles, we may be running the
 338             # postprocessor a second time
 339             '-map', '-0:s',
 340         ]
 341         if information['ext'] == 'mp4':
 342             opts += ['-c:s', 'mov_text']
 343         for (i, lang) in enumerate(sub_langs):
 344             opts.extend(['-map', '%d:0' % (i + 1)])
 345             lang_code = ISO639Utils.short2long(lang)
 346             if lang_code is not None:
 347                 opts.extend(['-metadata:s:s:%d' % i, 'language=%s' % lang_code])
 348
 349         temp_filename = prepend_extension(filename, 'temp')
 350         self._downloader.to_screen('[ffmpeg] Embedding subtitles in \'%s\'' % filename)
 351         self.run_ffmpeg_multiple_files(input_files, temp_filename, opts)
 352         os.remove(encodeFilename(filename))
 353         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 354
 355         return sub_filenames, information
 356
 357
 358 class FFmpegMetadataPP(FFmpegPostProcessor):
 359     def run(self, info):
 360         metadata = {}
 361         if info.get('title') is not None:
 362             metadata['title'] = info['title']
 363         if info.get('upload_date') is not None:
 364             metadata['date'] = info['upload_date']
 365         if info.get('artist') is not None:
 366             metadata['artist'] = info['artist']
 367         elif info.get('uploader') is not None:
 368             metadata['artist'] = info['uploader']
 369         elif info.get('uploader_id') is not None:
 370             metadata['artist'] = info['uploader_id']
 371         if info.get('description') is not None:
 372             metadata['description'] = info['description']
 373             metadata['comment'] = info['description']
 374         if info.get('webpage_url') is not None:
 375             metadata['purl'] = info['webpage_url']
 376         if info.get('album') is not None:
 377             metadata['album'] = info['album']
 378
 379         if not metadata:
 380             self._downloader.to_screen('[ffmpeg] There isn\'t any metadata to add')
 381             return [], info
 382
 383         filename = info['filepath']
 384         temp_filename = prepend_extension(filename, 'temp')
 385
 386         if info['ext'] == 'm4a':
 387             options = ['-vn', '-acodec', 'copy']
 388         else:
 389             options = ['-c', 'copy']
 390
 391         for (name, value) in metadata.items():
 392             options.extend(['-metadata', '%s=%s' % (name, value)])
 393
 394         # https://github.com/rg3/youtube-dl/issues/8350
 395         if info.get('protocol') == 'm3u8_native' or info.get('protocol') == 'm3u8' and self._downloader.params.get('hls_prefer_native', False):
 396             options.extend(['-bsf:a', 'aac_adtstoasc'])
 397
 398         self._downloader.to_screen('[ffmpeg] Adding metadata to \'%s\'' % filename)
 399         self.run_ffmpeg(filename, temp_filename, options)
 400         os.remove(encodeFilename(filename))
 401         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 402         return [], info
 403
 404
 405 class FFmpegMergerPP(FFmpegPostProcessor):
 406     def run(self, info):
 407         filename = info['filepath']
 408         temp_filename = prepend_extension(filename, 'temp')
 409         args = ['-c', 'copy', '-map', '0:v:0', '-map', '1:a:0']
 410         self._downloader.to_screen('[ffmpeg] Merging formats into "%s"' % filename)
 411         self.run_ffmpeg_multiple_files(info['__files_to_merge'], temp_filename, args)
 412         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 413         return info['__files_to_merge'], info
 414
 415     def can_merge(self):
 416         # TODO: figure out merge-capable ffmpeg version
 417         if self.basename != 'avconv':
 418             return True
 419
 420         required_version = '10-0'
 421         if is_outdated_version(
 422                 self._versions[self.basename], required_version):
 423             warning = ('Your copy of %s is outdated and unable to properly mux separate video and audio files, '
 424                        'youtube-dl will download single file media. '
 425                        'Update %s to version %s or newer to fix this.') % (
 426                            self.basename, self.basename, required_version)
 427             if self._downloader:
 428                 self._downloader.report_warning(warning)
 429             return False
 430         return True
 431
 432
 433 class FFmpegFixupStretchedPP(FFmpegPostProcessor):
 434     def run(self, info):
 435         stretched_ratio = info.get('stretched_ratio')
 436         if stretched_ratio is None or stretched_ratio == 1:
 437             return [], info
 438
 439         filename = info['filepath']
 440         temp_filename = prepend_extension(filename, 'temp')
 441
 442         options = ['-c', 'copy', '-aspect', '%f' % stretched_ratio]
 443         self._downloader.to_screen('[ffmpeg] Fixing aspect ratio in "%s"' % filename)
 444         self.run_ffmpeg(filename, temp_filename, options)
 445
 446         os.remove(encodeFilename(filename))
 447         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 448
 449         return [], info
 450
 451
 452 class FFmpegFixupM4aPP(FFmpegPostProcessor):
 453     def run(self, info):
 454         if info.get('container') != 'm4a_dash':
 455             return [], info
 456
 457         filename = info['filepath']
 458         temp_filename = prepend_extension(filename, 'temp')
 459
 460         options = ['-c', 'copy', '-f', 'mp4']
 461         self._downloader.to_screen('[ffmpeg] Correcting container in "%s"' % filename)
 462         self.run_ffmpeg(filename, temp_filename, options)
 463
 464         os.remove(encodeFilename(filename))
 465         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 466
 467         return [], info
 468
 469
 470 class FFmpegSubtitlesConvertorPP(FFmpegPostProcessor):
 471     def __init__(self, downloader=None, format=None):
 472         super(FFmpegSubtitlesConvertorPP, self).__init__(downloader)
 473         self.format = format
 474
 475     def run(self, info):
 476         subs = info.get('requested_subtitles')
 477         filename = info['filepath']
 478         new_ext = self.format
 479         new_format = new_ext
 480         if new_format == 'vtt':
 481             new_format = 'webvtt'
 482         if subs is None:
 483             self._downloader.to_screen('[ffmpeg] There aren\'t any subtitles to convert')
 484             return [], info
 485         self._downloader.to_screen('[ffmpeg] Converting subtitles')
 486         sub_filenames = []
 487         for lang, sub in subs.items():
 488             ext = sub['ext']
 489             if ext == new_ext:
 490                 self._downloader.to_screen(
 491                     '[ffmpeg] Subtitle file for %s is already in the requested'
 492                     'format' % new_ext)
 493                 continue
 494             old_file = subtitles_filename(filename, lang, ext)
 495             sub_filenames.append(old_file)
 496             new_file = subtitles_filename(filename, lang, new_ext)
 497
 498             if ext == 'dfxp' or ext == 'ttml':
 499                 self._downloader.report_warning(
 500                     'You have requested to convert dfxp (TTML) subtitles into another format, '
 501                     'which results in style information loss')
 502
 503                 dfxp_file = old_file
 504                 srt_file = subtitles_filename(filename, lang, 'srt')
 505
 506                 with io.open(dfxp_file, 'rt', encoding='utf-8') as f:
 507                     srt_data = dfxp2srt(f.read())
 508
 509                 with io.open(srt_file, 'wt', encoding='utf-8') as f:
 510                     f.write(srt_data)
 511                 old_file = srt_file
 512
 513                 subs[lang] = {
 514                     'ext': 'srt',
 515                     'data': srt_data
 516                 }
 517
 518                 if new_ext == 'srt':
 519                     continue
 520                 else:
 521                     sub_filenames.append(srt_file)
 522
 523             self.run_ffmpeg(old_file, new_file, ['-f', new_format])
 524
 525             with io.open(new_file, 'rt', encoding='utf-8') as f:
 526                 subs[lang] = {
 527                     'ext': new_ext,
 528                     'data': f.read(),
 529                 }
 530
 531         return sub_filenames, info