youtube_dl/postprocessor/ffmpeg.py

   1 from __future__ import unicode_literals
   2
   3 import os
   4 import subprocess
   5 import sys
   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 )
  24
  25
  26 class FFmpegPostProcessorError(PostProcessingError):
  27     pass
  28
  29
  30 class FFmpegPostProcessor(PostProcessor):
  31     def __init__(self, downloader=None, deletetempfiles=False):
  32         PostProcessor.__init__(self, downloader)
  33         self._deletetempfiles = deletetempfiles
  34         self._determine_executables()
  35
  36     def check_version(self):
  37         if not self.available():
  38             raise FFmpegPostProcessorError('ffmpeg or avconv not found. Please install one.')
  39
  40         required_version = '10-0' if self._uses_avconv() else '1.0'
  41         if is_outdated_version(
  42                 self._versions[self.basename], required_version):
  43             warning = 'Your copy of %s is outdated, update %s to version %s or newer if you encounter any errors.' % (
  44                 self.basename, self.basename, required_version)
  45             if self._downloader:
  46                 self._downloader.report_warning(warning)
  47
  48     @staticmethod
  49     def get_versions(downloader=None):
  50         return FFmpegPostProcessor(downloader)._versions
  51
  52     def _determine_executables(self):
  53         programs = ['avprobe', 'avconv', 'ffmpeg', 'ffprobe']
  54         prefer_ffmpeg = self._downloader.params.get('prefer_ffmpeg', False)
  55
  56         self.basename = None
  57         self.probe_basename = None
  58
  59         self._paths = None
  60         self._versions = None
  61         if self._downloader:
  62             location = self._downloader.params.get('ffmpeg_location')
  63             if location is not None:
  64                 if not os.path.exists(location):
  65                     self._downloader.report_warning(
  66                         'ffmpeg-location %s does not exist! '
  67                         'Continuing without avconv/ffmpeg.' % (location))
  68                     self._versions = {}
  69                     return
  70                 elif not os.path.isdir(location):
  71                     basename = os.path.splitext(os.path.basename(location))[0]
  72                     if basename not in programs:
  73                         self._downloader.report_warning(
  74                             'Cannot identify executable %s, its basename should be one of %s. '
  75                             'Continuing without avconv/ffmpeg.' %
  76                             (location, ', '.join(programs)))
  77                         self._versions = {}
  78                         return None
  79                     location = os.path.dirname(os.path.abspath(location))
  80                     if basename in ('ffmpeg', 'ffprobe'):
  81                         prefer_ffmpeg = True
  82
  83                 self._paths = dict(
  84                     (p, os.path.join(location, p)) for p in programs)
  85                 self._versions = dict(
  86                     (p, get_exe_version(self._paths[p], args=['-version']))
  87                     for p in programs)
  88         if self._versions is None:
  89             self._versions = dict(
  90                 (p, get_exe_version(p, args=['-version'])) for p in programs)
  91             self._paths = dict((p, p) for p in programs)
  92
  93         if prefer_ffmpeg:
  94             prefs = ('ffmpeg', 'avconv')
  95         else:
  96             prefs = ('avconv', 'ffmpeg')
  97         for p in prefs:
  98             if self._versions[p]:
  99                 self.basename = p
 100                 break
 101
 102         if prefer_ffmpeg:
 103             prefs = ('ffprobe', 'avprobe')
 104         else:
 105             prefs = ('avprobe', 'ffprobe')
 106         for p in prefs:
 107             if self._versions[p]:
 108                 self.probe_basename = p
 109                 break
 110
 111     def available(self):
 112         return self.basename is not None
 113
 114     def _uses_avconv(self):
 115         return self.basename == 'avconv'
 116
 117     @property
 118     def executable(self):
 119         return self._paths[self.basename]
 120
 121     @property
 122     def probe_executable(self):
 123         return self._paths[self.probe_basename]
 124
 125     def run_ffmpeg_multiple_files(self, input_paths, out_path, opts):
 126         self.check_version()
 127
 128         oldest_mtime = min(
 129             os.stat(encodeFilename(path)).st_mtime for path in input_paths)
 130
 131         files_cmd = []
 132         for path in input_paths:
 133             files_cmd.extend([encodeArgument('-i'), encodeFilename(path, True)])
 134         cmd = ([encodeFilename(self.executable, True), encodeArgument('-y')] +
 135                files_cmd +
 136                [encodeArgument(o) for o in opts] +
 137                [encodeFilename(self._ffmpeg_filename_argument(out_path), True)])
 138
 139         if self._downloader.params.get('verbose', False):
 140             self._downloader.to_screen('[debug] ffmpeg command line: %s' % shell_quote(cmd))
 141         p = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE, stdin=subprocess.PIPE)
 142         stdout, stderr = p.communicate()
 143         if p.returncode != 0:
 144             stderr = stderr.decode('utf-8', 'replace')
 145             msg = stderr.strip().split('\n')[-1]
 146             raise FFmpegPostProcessorError(msg)
 147         os.utime(encodeFilename(out_path), (oldest_mtime, oldest_mtime))
 148         if self._deletetempfiles:
 149             for ipath in input_paths:
 150                 os.remove(ipath)
 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_executable:
 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:
 265             self._nopostoverwrites = True
 266
 267         try:
 268             if self._nopostoverwrites and os.path.exists(encodeFilename(new_path)):
 269                 self._downloader.to_screen('[youtube] Post-process file %s exists, skipping' % new_path)
 270             else:
 271                 self._downloader.to_screen('[' + self.basename + '] Destination: ' + new_path)
 272                 self.run_ffmpeg(path, new_path, acodec, more_opts)
 273         except:
 274             etype, e, tb = sys.exc_info()
 275             if isinstance(e, AudioConversionError):
 276                 msg = 'audio conversion failed: ' + e.msg
 277             else:
 278                 msg = 'error running ' + self.basename
 279             raise PostProcessingError(msg)
 280
 281         # Try to update the date time for extracted audio file.
 282         if information.get('filetime') is not None:
 283             try:
 284                 os.utime(encodeFilename(new_path), (time.time(), information['filetime']))
 285             except:
 286                 self._downloader.report_warning('Cannot update utime of audio file')
 287
 288         information['filepath'] = new_path
 289         return self._nopostoverwrites, 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         prefix, sep, ext = path.rpartition('.')
 300         outpath = prefix + sep + self._preferedformat
 301         if information['ext'] == self._preferedformat:
 302             self._downloader.to_screen('[ffmpeg] Not converting video file %s - already is in target format %s' % (path, self._preferedformat))
 303             return True, information
 304         self._downloader.to_screen('[' + 'ffmpeg' + '] Converting video from %s to %s, Destination: ' % (information['ext'], self._preferedformat) + outpath)
 305         self.run_ffmpeg(path, outpath, [])
 306         information['filepath'] = outpath
 307         information['format'] = self._preferedformat
 308         information['ext'] = self._preferedformat
 309         return False, information
 310
 311
 312 class FFmpegEmbedSubtitlePP(FFmpegPostProcessor):
 313     # See http://www.loc.gov/standards/iso639-2/ISO-639-2_utf-8.txt
 314     _lang_map = {
 315         'aa': 'aar',
 316         'ab': 'abk',
 317         'ae': 'ave',
 318         'af': 'afr',
 319         'ak': 'aka',
 320         'am': 'amh',
 321         'an': 'arg',
 322         'ar': 'ara',
 323         'as': 'asm',
 324         'av': 'ava',
 325         'ay': 'aym',
 326         'az': 'aze',
 327         'ba': 'bak',
 328         'be': 'bel',
 329         'bg': 'bul',
 330         'bh': 'bih',
 331         'bi': 'bis',
 332         'bm': 'bam',
 333         'bn': 'ben',
 334         'bo': 'bod',
 335         'br': 'bre',
 336         'bs': 'bos',
 337         'ca': 'cat',
 338         'ce': 'che',
 339         'ch': 'cha',
 340         'co': 'cos',
 341         'cr': 'cre',
 342         'cs': 'ces',
 343         'cu': 'chu',
 344         'cv': 'chv',
 345         'cy': 'cym',
 346         'da': 'dan',
 347         'de': 'deu',
 348         'dv': 'div',
 349         'dz': 'dzo',
 350         'ee': 'ewe',
 351         'el': 'ell',
 352         'en': 'eng',
 353         'eo': 'epo',
 354         'es': 'spa',
 355         'et': 'est',
 356         'eu': 'eus',
 357         'fa': 'fas',
 358         'ff': 'ful',
 359         'fi': 'fin',
 360         'fj': 'fij',
 361         'fo': 'fao',
 362         'fr': 'fra',
 363         'fy': 'fry',
 364         'ga': 'gle',
 365         'gd': 'gla',
 366         'gl': 'glg',
 367         'gn': 'grn',
 368         'gu': 'guj',
 369         'gv': 'glv',
 370         'ha': 'hau',
 371         'he': 'heb',
 372         'hi': 'hin',
 373         'ho': 'hmo',
 374         'hr': 'hrv',
 375         'ht': 'hat',
 376         'hu': 'hun',
 377         'hy': 'hye',
 378         'hz': 'her',
 379         'ia': 'ina',
 380         'id': 'ind',
 381         'ie': 'ile',
 382         'ig': 'ibo',
 383         'ii': 'iii',
 384         'ik': 'ipk',
 385         'io': 'ido',
 386         'is': 'isl',
 387         'it': 'ita',
 388         'iu': 'iku',
 389         'ja': 'jpn',
 390         'jv': 'jav',
 391         'ka': 'kat',
 392         'kg': 'kon',
 393         'ki': 'kik',
 394         'kj': 'kua',
 395         'kk': 'kaz',
 396         'kl': 'kal',
 397         'km': 'khm',
 398         'kn': 'kan',
 399         'ko': 'kor',
 400         'kr': 'kau',
 401         'ks': 'kas',
 402         'ku': 'kur',
 403         'kv': 'kom',
 404         'kw': 'cor',
 405         'ky': 'kir',
 406         'la': 'lat',
 407         'lb': 'ltz',
 408         'lg': 'lug',
 409         'li': 'lim',
 410         'ln': 'lin',
 411         'lo': 'lao',
 412         'lt': 'lit',
 413         'lu': 'lub',
 414         'lv': 'lav',
 415         'mg': 'mlg',
 416         'mh': 'mah',
 417         'mi': 'mri',
 418         'mk': 'mkd',
 419         'ml': 'mal',
 420         'mn': 'mon',
 421         'mr': 'mar',
 422         'ms': 'msa',
 423         'mt': 'mlt',
 424         'my': 'mya',
 425         'na': 'nau',
 426         'nb': 'nob',
 427         'nd': 'nde',
 428         'ne': 'nep',
 429         'ng': 'ndo',
 430         'nl': 'nld',
 431         'nn': 'nno',
 432         'no': 'nor',
 433         'nr': 'nbl',
 434         'nv': 'nav',
 435         'ny': 'nya',
 436         'oc': 'oci',
 437         'oj': 'oji',
 438         'om': 'orm',
 439         'or': 'ori',
 440         'os': 'oss',
 441         'pa': 'pan',
 442         'pi': 'pli',
 443         'pl': 'pol',
 444         'ps': 'pus',
 445         'pt': 'por',
 446         'qu': 'que',
 447         'rm': 'roh',
 448         'rn': 'run',
 449         'ro': 'ron',
 450         'ru': 'rus',
 451         'rw': 'kin',
 452         'sa': 'san',
 453         'sc': 'srd',
 454         'sd': 'snd',
 455         'se': 'sme',
 456         'sg': 'sag',
 457         'si': 'sin',
 458         'sk': 'slk',
 459         'sl': 'slv',
 460         'sm': 'smo',
 461         'sn': 'sna',
 462         'so': 'som',
 463         'sq': 'sqi',
 464         'sr': 'srp',
 465         'ss': 'ssw',
 466         'st': 'sot',
 467         'su': 'sun',
 468         'sv': 'swe',
 469         'sw': 'swa',
 470         'ta': 'tam',
 471         'te': 'tel',
 472         'tg': 'tgk',
 473         'th': 'tha',
 474         'ti': 'tir',
 475         'tk': 'tuk',
 476         'tl': 'tgl',
 477         'tn': 'tsn',
 478         'to': 'ton',
 479         'tr': 'tur',
 480         'ts': 'tso',
 481         'tt': 'tat',
 482         'tw': 'twi',
 483         'ty': 'tah',
 484         'ug': 'uig',
 485         'uk': 'ukr',
 486         'ur': 'urd',
 487         'uz': 'uzb',
 488         've': 'ven',
 489         'vi': 'vie',
 490         'vo': 'vol',
 491         'wa': 'wln',
 492         'wo': 'wol',
 493         'xh': 'xho',
 494         'yi': 'yid',
 495         'yo': 'yor',
 496         'za': 'zha',
 497         'zh': 'zho',
 498         'zu': 'zul',
 499     }
 500
 501     def __init__(self, downloader=None, subtitlesformat='srt'):
 502         super(FFmpegEmbedSubtitlePP, self).__init__(downloader)
 503         self._subformat = subtitlesformat
 504
 505     @classmethod
 506     def _conver_lang_code(cls, code):
 507         """Convert language code from ISO 639-1 to ISO 639-2/T"""
 508         return cls._lang_map.get(code[:2])
 509
 510     def run(self, information):
 511         if information['ext'] != 'mp4':
 512             self._downloader.to_screen('[ffmpeg] Subtitles can only be embedded in mp4 files')
 513             return True, information
 514         if not information.get('subtitles'):
 515             self._downloader.to_screen('[ffmpeg] There aren\'t any subtitles to embed')
 516             return True, information
 517
 518         sub_langs = [key for key in information['subtitles']]
 519         filename = information['filepath']
 520         input_files = [filename] + [subtitles_filename(filename, lang, self._subformat) for lang in sub_langs]
 521
 522         opts = [
 523             '-map', '0',
 524             '-c', 'copy',
 525             # Don't copy the existing subtitles, we may be running the
 526             # postprocessor a second time
 527             '-map', '-0:s',
 528             '-c:s', 'mov_text',
 529         ]
 530         for (i, lang) in enumerate(sub_langs):
 531             opts.extend(['-map', '%d:0' % (i + 1)])
 532             lang_code = self._conver_lang_code(lang)
 533             if lang_code is not None:
 534                 opts.extend(['-metadata:s:s:%d' % i, 'language=%s' % lang_code])
 535
 536         temp_filename = prepend_extension(filename, 'temp')
 537         self._downloader.to_screen('[ffmpeg] Embedding subtitles in \'%s\'' % filename)
 538         self.run_ffmpeg_multiple_files(input_files, temp_filename, opts)
 539         os.remove(encodeFilename(filename))
 540         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 541
 542         return True, information
 543
 544
 545 class FFmpegMetadataPP(FFmpegPostProcessor):
 546     def run(self, info):
 547         metadata = {}
 548         if info.get('title') is not None:
 549             metadata['title'] = info['title']
 550         if info.get('upload_date') is not None:
 551             metadata['date'] = info['upload_date']
 552         if info.get('uploader') is not None:
 553             metadata['artist'] = info['uploader']
 554         elif info.get('uploader_id') is not None:
 555             metadata['artist'] = info['uploader_id']
 556         if info.get('description') is not None:
 557             metadata['description'] = info['description']
 558             metadata['comment'] = info['description']
 559         if info.get('webpage_url') is not None:
 560             metadata['purl'] = info['webpage_url']
 561
 562         if not metadata:
 563             self._downloader.to_screen('[ffmpeg] There isn\'t any metadata to add')
 564             return True, info
 565
 566         filename = info['filepath']
 567         temp_filename = prepend_extension(filename, 'temp')
 568
 569         if info['ext'] == 'm4a':
 570             options = ['-vn', '-acodec', 'copy']
 571         else:
 572             options = ['-c', 'copy']
 573
 574         for (name, value) in metadata.items():
 575             options.extend(['-metadata', '%s=%s' % (name, value)])
 576
 577         self._downloader.to_screen('[ffmpeg] Adding metadata to \'%s\'' % filename)
 578         self.run_ffmpeg(filename, temp_filename, options)
 579         os.remove(encodeFilename(filename))
 580         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 581         return True, info
 582
 583
 584 class FFmpegMergerPP(FFmpegPostProcessor):
 585     def run(self, info):
 586         filename = info['filepath']
 587         args = ['-c', 'copy', '-map', '0:v:0', '-map', '1:a:0']
 588         self._downloader.to_screen('[ffmpeg] Merging formats into "%s"' % filename)
 589         self.run_ffmpeg_multiple_files(info['__files_to_merge'], filename, args)
 590         return True, info
 591
 592
 593 class FFmpegAudioFixPP(FFmpegPostProcessor):
 594     def run(self, info):
 595         filename = info['filepath']
 596         temp_filename = prepend_extension(filename, 'temp')
 597
 598         options = ['-vn', '-acodec', 'copy']
 599         self._downloader.to_screen('[ffmpeg] Fixing audio file "%s"' % filename)
 600         self.run_ffmpeg(filename, temp_filename, options)
 601
 602         os.remove(encodeFilename(filename))
 603         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 604
 605         return True, info
 606
 607
 608 class FFmpegFixupStretchedPP(FFmpegPostProcessor):
 609     def run(self, info):
 610         stretched_ratio = info.get('stretched_ratio')
 611         if stretched_ratio is None or stretched_ratio == 1:
 612             return True, info
 613
 614         filename = info['filepath']
 615         temp_filename = prepend_extension(filename, 'temp')
 616
 617         options = ['-c', 'copy', '-aspect', '%f' % stretched_ratio]
 618         self._downloader.to_screen('[ffmpeg] Fixing aspect ratio in "%s"' % filename)
 619         self.run_ffmpeg(filename, temp_filename, options)
 620
 621         os.remove(encodeFilename(filename))
 622         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 623
 624         return True, info
 625
 626
 627 class FFmpegFixupM4aPP(FFmpegPostProcessor):
 628     def run(self, info):
 629         if info.get('container') != 'm4a_dash':
 630             return True, info
 631
 632         filename = info['filepath']
 633         temp_filename = prepend_extension(filename, 'temp')
 634
 635         options = ['-c', 'copy', '-f', 'mp4']
 636         self._downloader.to_screen('[ffmpeg] Correcting container in "%s"' % filename)
 637         self.run_ffmpeg(filename, temp_filename, options)
 638
 639         os.remove(encodeFilename(filename))
 640         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 641
 642         return True, info