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  44             warning = 'Your copy of %s is outdated, update %s to version %s or newer if you encounter any errors.'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