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  44             warning = 'Your copy of %s is outdated, update %s to version %s or newer if you encounter any errors.'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