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._versions = self.get_versions()
  34         self._deletetempfiles = deletetempfiles
  35
  36     def check_version(self):
  37         if not self._executable:
  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._executable], required_version):
  43             warning = 'Your copy of %s is outdated, update %s to version %s or newer if you encounter any errors.' % (
  44                 self._executable, self._executable, required_version)
  45             if self._downloader:
  46                 self._downloader.report_warning(warning)
  47
  48     @staticmethod
  49     def get_versions():
  50         programs = ['avprobe', 'avconv', 'ffmpeg', 'ffprobe']
  51         return dict((p, get_exe_version(p, args=['-version'])) for p in programs)
  52
  53     @property
  54     def _executable(self):
  55         if self._downloader.params.get('prefer_ffmpeg', False):
  56             prefs = ('ffmpeg', 'avconv')
  57         else:
  58             prefs = ('avconv', 'ffmpeg')
  59         for p in prefs:
  60             if self._versions[p]:
  61                 return p
  62         return None
  63
  64     @property
  65     def _probe_executable(self):
  66         if self._downloader.params.get('prefer_ffmpeg', False):
  67             prefs = ('ffprobe', 'avprobe')
  68         else:
  69             prefs = ('avprobe', 'ffprobe')
  70         for p in prefs:
  71             if self._versions[p]:
  72                 return p
  73         return None
  74
  75     def _uses_avconv(self):
  76         return self._executable == 'avconv'
  77
  78     def run_ffmpeg_multiple_files(self, input_paths, out_path, opts):
  79         self.check_version()
  80
  81         files_cmd = []
  82         for path in input_paths:
  83             files_cmd.extend([encodeArgument('-i'), encodeFilename(path, True)])
  84         cmd = ([encodeFilename(self._executable, True), encodeArgument('-y')] +
  85                files_cmd
  86                + [encodeArgument(o) for o in opts] +
  87                [encodeFilename(self._ffmpeg_filename_argument(out_path), True)])
  88
  89         if self._downloader.params.get('verbose', False):
  90             self._downloader.to_screen('[debug] ffmpeg command line: %s' % shell_quote(cmd))
  91         p = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
  92         stdout, stderr = p.communicate()
  93         if p.returncode != 0:
  94             stderr = stderr.decode('utf-8', 'replace')
  95             msg = stderr.strip().split('\n')[-1]
  96             raise FFmpegPostProcessorError(msg)
  97         if self._deletetempfiles:
  98             for ipath in input_paths:
  99                 os.remove(ipath)
 100
 101     def run_ffmpeg(self, path, out_path, opts):
 102         self.run_ffmpeg_multiple_files([path], out_path, opts)
 103
 104     def _ffmpeg_filename_argument(self, fn):
 105         # ffmpeg broke --, see https://ffmpeg.org/trac/ffmpeg/ticket/2127 for details
 106         if fn.startswith('-'):
 107             return './' + fn
 108         return fn
 109
 110
 111 class FFmpegExtractAudioPP(FFmpegPostProcessor):
 112     def __init__(self, downloader=None, preferredcodec=None, preferredquality=None, nopostoverwrites=False):
 113         FFmpegPostProcessor.__init__(self, downloader)
 114         if preferredcodec is None:
 115             preferredcodec = 'best'
 116         self._preferredcodec = preferredcodec
 117         self._preferredquality = preferredquality
 118         self._nopostoverwrites = nopostoverwrites
 119
 120     def get_audio_codec(self, path):
 121
 122         if not self._probe_executable:
 123             raise PostProcessingError('ffprobe or avprobe not found. Please install one.')
 124         try:
 125             cmd = [
 126                 encodeFilename(self._probe_executable, True),
 127                 encodeArgument('-show_streams'),
 128                 encodeFilename(self._ffmpeg_filename_argument(path), True)]
 129             handle = subprocess.Popen(cmd, stderr=compat_subprocess_get_DEVNULL(), stdout=subprocess.PIPE)
 130             output = handle.communicate()[0]
 131             if handle.wait() != 0:
 132                 return None
 133         except (IOError, OSError):
 134             return None
 135         audio_codec = None
 136         for line in output.decode('ascii', 'ignore').split('\n'):
 137             if line.startswith('codec_name='):
 138                 audio_codec = line.split('=')[1].strip()
 139             elif line.strip() == 'codec_type=audio' and audio_codec is not None:
 140                 return audio_codec
 141         return None
 142
 143     def run_ffmpeg(self, path, out_path, codec, more_opts):
 144         if codec is None:
 145             acodec_opts = []
 146         else:
 147             acodec_opts = ['-acodec', codec]
 148         opts = ['-vn'] + acodec_opts + more_opts
 149         try:
 150             FFmpegPostProcessor.run_ffmpeg(self, path, out_path, opts)
 151         except FFmpegPostProcessorError as err:
 152             raise AudioConversionError(err.msg)
 153
 154     def run(self, information):
 155         path = information['filepath']
 156
 157         filecodec = self.get_audio_codec(path)
 158         if filecodec is None:
 159             raise PostProcessingError('WARNING: unable to obtain file audio codec with ffprobe')
 160
 161         uses_avconv = self._uses_avconv()
 162         more_opts = []
 163         if self._preferredcodec == 'best' or self._preferredcodec == filecodec or (self._preferredcodec == 'm4a' and filecodec == 'aac'):
 164             if filecodec == 'aac' and self._preferredcodec in ['m4a', 'best']:
 165                 # Lossless, but in another container
 166                 acodec = 'copy'
 167                 extension = 'm4a'
 168                 more_opts = ['-bsf:a' if uses_avconv else '-absf', 'aac_adtstoasc']
 169             elif filecodec in ['aac', 'mp3', 'vorbis', 'opus']:
 170                 # Lossless if possible
 171                 acodec = 'copy'
 172                 extension = filecodec
 173                 if filecodec == 'aac':
 174                     more_opts = ['-f', 'adts']
 175                 if filecodec == 'vorbis':
 176                     extension = 'ogg'
 177             else:
 178                 # MP3 otherwise.
 179                 acodec = 'libmp3lame'
 180                 extension = 'mp3'
 181                 more_opts = []
 182                 if self._preferredquality is not None:
 183                     if int(self._preferredquality) < 10:
 184                         more_opts += ['-q:a' if uses_avconv else '-aq', self._preferredquality]
 185                     else:
 186                         more_opts += ['-b:a' if uses_avconv else '-ab', self._preferredquality + 'k']
 187         else:
 188             # We convert the audio (lossy)
 189             acodec = {'mp3': 'libmp3lame', 'aac': 'aac', 'm4a': 'aac', 'opus': 'opus', 'vorbis': 'libvorbis', 'wav': None}[self._preferredcodec]
 190             extension = self._preferredcodec
 191             more_opts = []
 192             if self._preferredquality is not None:
 193                 # The opus codec doesn't support the -aq option
 194                 if int(self._preferredquality) < 10 and extension != 'opus':
 195                     more_opts += ['-q:a' if uses_avconv else '-aq', self._preferredquality]
 196                 else:
 197                     more_opts += ['-b:a' if uses_avconv else '-ab', self._preferredquality + 'k']
 198             if self._preferredcodec == 'aac':
 199                 more_opts += ['-f', 'adts']
 200             if self._preferredcodec == 'm4a':
 201                 more_opts += ['-bsf:a' if uses_avconv else '-absf', 'aac_adtstoasc']
 202             if self._preferredcodec == 'vorbis':
 203                 extension = 'ogg'
 204             if self._preferredcodec == 'wav':
 205                 extension = 'wav'
 206                 more_opts += ['-f', 'wav']
 207
 208         prefix, sep, ext = path.rpartition('.')  # not os.path.splitext, since the latter does not work on unicode in all setups
 209         new_path = prefix + sep + extension
 210
 211         # If we download foo.mp3 and convert it to... foo.mp3, then don't delete foo.mp3, silly.
 212         if new_path == path:
 213             self._nopostoverwrites = True
 214
 215         try:
 216             if self._nopostoverwrites and os.path.exists(encodeFilename(new_path)):
 217                 self._downloader.to_screen('[youtube] Post-process file %s exists, skipping' % new_path)
 218             else:
 219                 self._downloader.to_screen('[' + self._executable + '] Destination: ' + new_path)
 220                 self.run_ffmpeg(path, new_path, acodec, more_opts)
 221         except:
 222             etype, e, tb = sys.exc_info()
 223             if isinstance(e, AudioConversionError):
 224                 msg = 'audio conversion failed: ' + e.msg
 225             else:
 226                 msg = 'error running ' + self._executable
 227             raise PostProcessingError(msg)
 228
 229         # Try to update the date time for extracted audio file.
 230         if information.get('filetime') is not None:
 231             try:
 232                 os.utime(encodeFilename(new_path), (time.time(), information['filetime']))
 233             except:
 234                 self._downloader.report_warning('Cannot update utime of audio file')
 235
 236         information['filepath'] = new_path
 237         return self._nopostoverwrites, information
 238
 239
 240 class FFmpegVideoConvertorPP(FFmpegPostProcessor):
 241     def __init__(self, downloader=None, preferedformat=None):
 242         super(FFmpegVideoConvertorPP, self).__init__(downloader)
 243         self._preferedformat = preferedformat
 244
 245     def run(self, information):
 246         path = information['filepath']
 247         prefix, sep, ext = path.rpartition('.')
 248         outpath = prefix + sep + self._preferedformat
 249         if information['ext'] == self._preferedformat:
 250             self._downloader.to_screen('[ffmpeg] Not converting video file %s - already is in target format %s' % (path, self._preferedformat))
 251             return True, information
 252         self._downloader.to_screen('[' + 'ffmpeg' + '] Converting video from %s to %s, Destination: ' % (information['ext'], self._preferedformat) + outpath)
 253         self.run_ffmpeg(path, outpath, [])
 254         information['filepath'] = outpath
 255         information['format'] = self._preferedformat
 256         information['ext'] = self._preferedformat
 257         return False, information
 258
 259
 260 class FFmpegEmbedSubtitlePP(FFmpegPostProcessor):
 261     # See http://www.loc.gov/standards/iso639-2/ISO-639-2_utf-8.txt
 262     _lang_map = {
 263         'aa': 'aar',
 264         'ab': 'abk',
 265         'ae': 'ave',
 266         'af': 'afr',
 267         'ak': 'aka',
 268         'am': 'amh',
 269         'an': 'arg',
 270         'ar': 'ara',
 271         'as': 'asm',
 272         'av': 'ava',
 273         'ay': 'aym',
 274         'az': 'aze',
 275         'ba': 'bak',
 276         'be': 'bel',
 277         'bg': 'bul',
 278         'bh': 'bih',
 279         'bi': 'bis',
 280         'bm': 'bam',
 281         'bn': 'ben',
 282         'bo': 'bod',
 283         'br': 'bre',
 284         'bs': 'bos',
 285         'ca': 'cat',
 286         'ce': 'che',
 287         'ch': 'cha',
 288         'co': 'cos',
 289         'cr': 'cre',
 290         'cs': 'ces',
 291         'cu': 'chu',
 292         'cv': 'chv',
 293         'cy': 'cym',
 294         'da': 'dan',
 295         'de': 'deu',
 296         'dv': 'div',
 297         'dz': 'dzo',
 298         'ee': 'ewe',
 299         'el': 'ell',
 300         'en': 'eng',
 301         'eo': 'epo',
 302         'es': 'spa',
 303         'et': 'est',
 304         'eu': 'eus',
 305         'fa': 'fas',
 306         'ff': 'ful',
 307         'fi': 'fin',
 308         'fj': 'fij',
 309         'fo': 'fao',
 310         'fr': 'fra',
 311         'fy': 'fry',
 312         'ga': 'gle',
 313         'gd': 'gla',
 314         'gl': 'glg',
 315         'gn': 'grn',
 316         'gu': 'guj',
 317         'gv': 'glv',
 318         'ha': 'hau',
 319         'he': 'heb',
 320         'hi': 'hin',
 321         'ho': 'hmo',
 322         'hr': 'hrv',
 323         'ht': 'hat',
 324         'hu': 'hun',
 325         'hy': 'hye',
 326         'hz': 'her',
 327         'ia': 'ina',
 328         'id': 'ind',
 329         'ie': 'ile',
 330         'ig': 'ibo',
 331         'ii': 'iii',
 332         'ik': 'ipk',
 333         'io': 'ido',
 334         'is': 'isl',
 335         'it': 'ita',
 336         'iu': 'iku',
 337         'ja': 'jpn',
 338         'jv': 'jav',
 339         'ka': 'kat',
 340         'kg': 'kon',
 341         'ki': 'kik',
 342         'kj': 'kua',
 343         'kk': 'kaz',
 344         'kl': 'kal',
 345         'km': 'khm',
 346         'kn': 'kan',
 347         'ko': 'kor',
 348         'kr': 'kau',
 349         'ks': 'kas',
 350         'ku': 'kur',
 351         'kv': 'kom',
 352         'kw': 'cor',
 353         'ky': 'kir',
 354         'la': 'lat',
 355         'lb': 'ltz',
 356         'lg': 'lug',
 357         'li': 'lim',
 358         'ln': 'lin',
 359         'lo': 'lao',
 360         'lt': 'lit',
 361         'lu': 'lub',
 362         'lv': 'lav',
 363         'mg': 'mlg',
 364         'mh': 'mah',
 365         'mi': 'mri',
 366         'mk': 'mkd',
 367         'ml': 'mal',
 368         'mn': 'mon',
 369         'mr': 'mar',
 370         'ms': 'msa',
 371         'mt': 'mlt',
 372         'my': 'mya',
 373         'na': 'nau',
 374         'nb': 'nob',
 375         'nd': 'nde',
 376         'ne': 'nep',
 377         'ng': 'ndo',
 378         'nl': 'nld',
 379         'nn': 'nno',
 380         'no': 'nor',
 381         'nr': 'nbl',
 382         'nv': 'nav',
 383         'ny': 'nya',
 384         'oc': 'oci',
 385         'oj': 'oji',
 386         'om': 'orm',
 387         'or': 'ori',
 388         'os': 'oss',
 389         'pa': 'pan',
 390         'pi': 'pli',
 391         'pl': 'pol',
 392         'ps': 'pus',
 393         'pt': 'por',
 394         'qu': 'que',
 395         'rm': 'roh',
 396         'rn': 'run',
 397         'ro': 'ron',
 398         'ru': 'rus',
 399         'rw': 'kin',
 400         'sa': 'san',
 401         'sc': 'srd',
 402         'sd': 'snd',
 403         'se': 'sme',
 404         'sg': 'sag',
 405         'si': 'sin',
 406         'sk': 'slk',
 407         'sl': 'slv',
 408         'sm': 'smo',
 409         'sn': 'sna',
 410         'so': 'som',
 411         'sq': 'sqi',
 412         'sr': 'srp',
 413         'ss': 'ssw',
 414         'st': 'sot',
 415         'su': 'sun',
 416         'sv': 'swe',
 417         'sw': 'swa',
 418         'ta': 'tam',
 419         'te': 'tel',
 420         'tg': 'tgk',
 421         'th': 'tha',
 422         'ti': 'tir',
 423         'tk': 'tuk',
 424         'tl': 'tgl',
 425         'tn': 'tsn',
 426         'to': 'ton',
 427         'tr': 'tur',
 428         'ts': 'tso',
 429         'tt': 'tat',
 430         'tw': 'twi',
 431         'ty': 'tah',
 432         'ug': 'uig',
 433         'uk': 'ukr',
 434         'ur': 'urd',
 435         'uz': 'uzb',
 436         've': 'ven',
 437         'vi': 'vie',
 438         'vo': 'vol',
 439         'wa': 'wln',
 440         'wo': 'wol',
 441         'xh': 'xho',
 442         'yi': 'yid',
 443         'yo': 'yor',
 444         'za': 'zha',
 445         'zh': 'zho',
 446         'zu': 'zul',
 447     }
 448
 449     def __init__(self, downloader=None, subtitlesformat='srt'):
 450         super(FFmpegEmbedSubtitlePP, self).__init__(downloader)
 451         self._subformat = subtitlesformat
 452
 453     @classmethod
 454     def _conver_lang_code(cls, code):
 455         """Convert language code from ISO 639-1 to ISO 639-2/T"""
 456         return cls._lang_map.get(code[:2])
 457
 458     def run(self, information):
 459         if information['ext'] != 'mp4':
 460             self._downloader.to_screen('[ffmpeg] Subtitles can only be embedded in mp4 files')
 461             return True, information
 462         if not information.get('subtitles'):
 463             self._downloader.to_screen('[ffmpeg] There aren\'t any subtitles to embed')
 464             return True, information
 465
 466         sub_langs = [key for key in information['subtitles']]
 467         filename = information['filepath']
 468         input_files = [filename] + [subtitles_filename(filename, lang, self._subformat) for lang in sub_langs]
 469
 470         opts = ['-map', '0:0', '-map', '0:1', '-c:v', 'copy', '-c:a', 'copy']
 471         for (i, lang) in enumerate(sub_langs):
 472             opts.extend(['-map', '%d:0' % (i + 1), '-c:s:%d' % i, 'mov_text'])
 473             lang_code = self._conver_lang_code(lang)
 474             if lang_code is not None:
 475                 opts.extend(['-metadata:s:s:%d' % i, 'language=%s' % lang_code])
 476         opts.extend(['-f', 'mp4'])
 477
 478         temp_filename = filename + '.temp'
 479         self._downloader.to_screen('[ffmpeg] Embedding subtitles in \'%s\'' % filename)
 480         self.run_ffmpeg_multiple_files(input_files, temp_filename, opts)
 481         os.remove(encodeFilename(filename))
 482         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 483
 484         return True, information
 485
 486
 487 class FFmpegMetadataPP(FFmpegPostProcessor):
 488     def run(self, info):
 489         metadata = {}
 490         if info.get('title') is not None:
 491             metadata['title'] = info['title']
 492         if info.get('upload_date') is not None:
 493             metadata['date'] = info['upload_date']
 494         if info.get('uploader') is not None:
 495             metadata['artist'] = info['uploader']
 496         elif info.get('uploader_id') is not None:
 497             metadata['artist'] = info['uploader_id']
 498
 499         if not metadata:
 500             self._downloader.to_screen('[ffmpeg] There isn\'t any metadata to add')
 501             return True, info
 502
 503         filename = info['filepath']
 504         temp_filename = prepend_extension(filename, 'temp')
 505
 506         if info['ext'] == 'm4a':
 507             options = ['-vn', '-acodec', 'copy']
 508         else:
 509             options = ['-c', 'copy']
 510
 511         for (name, value) in metadata.items():
 512             options.extend(['-metadata', '%s=%s' % (name, value)])
 513
 514         self._downloader.to_screen('[ffmpeg] Adding metadata to \'%s\'' % filename)
 515         self.run_ffmpeg(filename, temp_filename, options)
 516         os.remove(encodeFilename(filename))
 517         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 518         return True, info
 519
 520
 521 class FFmpegMergerPP(FFmpegPostProcessor):
 522     def run(self, info):
 523         filename = info['filepath']
 524         args = ['-c', 'copy', '-map', '0:v:0', '-map', '1:a:0']
 525         self._downloader.to_screen('[ffmpeg] Merging formats into "%s"' % filename)
 526         self.run_ffmpeg_multiple_files(info['__files_to_merge'], filename, args)
 527         return True, info
 528
 529
 530 class FFmpegAudioFixPP(FFmpegPostProcessor):
 531     def run(self, info):
 532         filename = info['filepath']
 533         temp_filename = prepend_extension(filename, 'temp')
 534
 535         options = ['-vn', '-acodec', 'copy']
 536         self._downloader.to_screen('[ffmpeg] Fixing audio file "%s"' % filename)
 537         self.run_ffmpeg(filename, temp_filename, options)
 538
 539         os.remove(encodeFilename(filename))
 540         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 541
 542         return True, info