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  44             warning = 'Your copy of %s is outdated, update %s to version %s or newer if you encounter any errors.'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