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  43             warning = 'Your copy of %s is outdated, update %s to version %s or newer if you encounter any errors.'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 available(self):
  55         return self._executable is not None
  56
  57     @property
  58     def _executable(self):
  59         if self._downloader.params.get('prefer_ffmpeg', False):
  60             prefs = ('ffmpeg', 'avconv')
  61         else:
  62             prefs = ('avconv', 'ffmpeg')
  63         for p in prefs:
  64             if self._versions[p]:
  65                 return p
  66         return None
  67
  68     @property
  69     def _probe_executable(self):
  70         if self._downloader.params.get('prefer_ffmpeg', False):
  71             prefs = ('ffprobe', 'avprobe')
  72         else:
  73             prefs = ('avprobe', 'ffprobe')
  74         for p in prefs:
  75             if self._versions[p]:
  76                 return p
  77         return None
  78
  79     def _uses_avconv(self):
  80         return self._executable == 'avconv'
  81
  82     def run_ffmpeg_multiple_files(self, input_paths, out_path, opts):
  83         self.check_version()
  84
  85         oldest_mtime = min(
  86             os.stat(encodeFilename(path)).st_mtime for path in input_paths)
  87
  88         files_cmd = []
  89         for path in input_paths:
  90             files_cmd.extend([encodeArgument('-i'), encodeFilename(path, True)])
  91         cmd = ([encodeFilename(self._executable, True), encodeArgument('-y')] +
  92                files_cmd +
  93                [encodeArgument(o) for o in opts] +
  94                [encodeFilename(self._ffmpeg_filename_argument(out_path), True)])
  95
  96         if self._downloader.params.get('verbose', False):
  97             self._downloader.to_screen('[debug] ffmpeg command line: %s' % shell_quote(cmd))
  98         p = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
  99         stdout, stderr = p.communicate()
 100         if p.returncode != 0:
 101             stderr = stderr.decode('utf-8', 'replace')
 102             msg = stderr.strip().split('\n')[-1]
 103             raise FFmpegPostProcessorError(msg)
 104         os.utime(encodeFilename(out_path), (oldest_mtime, oldest_mtime))
 105         if self._deletetempfiles:
 106             for ipath in input_paths:
 107                 os.remove(ipath)
 108
 109     def run_ffmpeg(self, path, out_path, opts):
 110         self.run_ffmpeg_multiple_files([path], out_path, opts)
 111
 112     def _ffmpeg_filename_argument(self, fn):
 113         # ffmpeg broke --, see https://ffmpeg.org/trac/ffmpeg/ticket/2127 for details
 114         if fn.startswith('-'):
 115             return './' + fn
 116         return fn
 117
 118
 119 class FFmpegExtractAudioPP(FFmpegPostProcessor):
 120     def __init__(self, downloader=None, preferredcodec=None, preferredquality=None, nopostoverwrites=False):
 121         FFmpegPostProcessor.__init__(self, downloader)
 122         if preferredcodec is None:
 123             preferredcodec = 'best'
 124         self._preferredcodec = preferredcodec
 125         self._preferredquality = preferredquality
 126         self._nopostoverwrites = nopostoverwrites
 127
 128     def get_audio_codec(self, path):
 129
 130         if not self._probe_executable:
 131             raise PostProcessingError('ffprobe or avprobe not found. Please install one.')
 132         try:
 133             cmd = [
 134                 encodeFilename(self._probe_executable, True),
 135                 encodeArgument('-show_streams'),
 136                 encodeFilename(self._ffmpeg_filename_argument(path), True)]
 137             handle = subprocess.Popen(cmd, stderr=compat_subprocess_get_DEVNULL(), stdout=subprocess.PIPE)
 138             output = handle.communicate()[0]
 139             if handle.wait() != 0:
 140                 return None
 141         except (IOError, OSError):
 142             return None
 143         audio_codec = None
 144         for line in output.decode('ascii', 'ignore').split('\n'):
 145             if line.startswith('codec_name='):
 146                 audio_codec = line.split('=')[1].strip()
 147             elif line.strip() == 'codec_type=audio' and audio_codec is not None:
 148                 return audio_codec
 149         return None
 150
 151     def run_ffmpeg(self, path, out_path, codec, more_opts):
 152         if codec is None:
 153             acodec_opts = []
 154         else:
 155             acodec_opts = ['-acodec', codec]
 156         opts = ['-vn'] + acodec_opts + more_opts
 157         try:
 158             FFmpegPostProcessor.run_ffmpeg(self, path, out_path, opts)
 159         except FFmpegPostProcessorError as err:
 160             raise AudioConversionError(err.msg)
 161
 162     def run(self, information):
 163         path = information['filepath']
 164
 165         filecodec = self.get_audio_codec(path)
 166         if filecodec is None:
 167             raise PostProcessingError('WARNING: unable to obtain file audio codec with ffprobe')
 168
 169         more_opts = []
 170         if self._preferredcodec == 'best' or self._preferredcodec == filecodec or (self._preferredcodec == 'm4a' and filecodec == 'aac'):
 171             if filecodec == 'aac' and self._preferredcodec in ['m4a', 'best']:
 172                 # Lossless, but in another container
 173                 acodec = 'copy'
 174                 extension = 'm4a'
 175                 more_opts = ['-bsf:a', 'aac_adtstoasc']
 176             elif filecodec in ['aac', 'mp3', 'vorbis', 'opus']:
 177                 # Lossless if possible
 178                 acodec = 'copy'
 179                 extension = filecodec
 180                 if filecodec == 'aac':
 181                     more_opts = ['-f', 'adts']
 182                 if filecodec == 'vorbis':
 183                     extension = 'ogg'
 184             else:
 185                 # MP3 otherwise.
 186                 acodec = 'libmp3lame'
 187                 extension = 'mp3'
 188                 more_opts = []
 189                 if self._preferredquality is not None:
 190                     if int(self._preferredquality) < 10:
 191                         more_opts += ['-q:a', self._preferredquality]
 192                     else:
 193                         more_opts += ['-b:a', self._preferredquality + 'k']
 194         else:
 195             # We convert the audio (lossy)
 196             acodec = {'mp3': 'libmp3lame', 'aac': 'aac', 'm4a': 'aac', 'opus': 'opus', 'vorbis': 'libvorbis', 'wav': None}[self._preferredcodec]
 197             extension = self._preferredcodec
 198             more_opts = []
 199             if self._preferredquality is not None:
 200                 # The opus codec doesn't support the -aq option
 201                 if int(self._preferredquality) < 10 and extension != 'opus':
 202                     more_opts += ['-q:a', self._preferredquality]
 203                 else:
 204                     more_opts += ['-b:a', self._preferredquality + 'k']
 205             if self._preferredcodec == 'aac':
 206                 more_opts += ['-f', 'adts']
 207             if self._preferredcodec == 'm4a':
 208                 more_opts += ['-bsf:a', 'aac_adtstoasc']
 209             if self._preferredcodec == 'vorbis':
 210                 extension = 'ogg'
 211             if self._preferredcodec == 'wav':
 212                 extension = 'wav'
 213                 more_opts += ['-f', 'wav']
 214
 215         prefix, sep, ext = path.rpartition('.')  # not os.path.splitext, since the latter does not work on unicode in all setups
 216         new_path = prefix + sep + extension
 217
 218         # If we download foo.mp3 and convert it to... foo.mp3, then don't delete foo.mp3, silly.
 219         if new_path == path:
 220             self._nopostoverwrites = True
 221
 222         try:
 223             if self._nopostoverwrites and os.path.exists(encodeFilename(new_path)):
 224                 self._downloader.to_screen('[youtube] Post-process file %s exists, skipping' % new_path)
 225             else:
 226                 self._downloader.to_screen('[' + self._executable + '] Destination: ' + new_path)
 227                 self.run_ffmpeg(path, new_path, acodec, more_opts)
 228         except:
 229             etype, e, tb = sys.exc_info()
 230             if isinstance(e, AudioConversionError):
 231                 msg = 'audio conversion failed: ' + e.msg
 232             else:
 233                 msg = 'error running ' + self._executable
 234             raise PostProcessingError(msg)
 235
 236         # Try to update the date time for extracted audio file.
 237         if information.get('filetime') is not None:
 238             try:
 239                 os.utime(encodeFilename(new_path), (time.time(), information['filetime']))
 240             except:
 241                 self._downloader.report_warning('Cannot update utime of audio file')
 242
 243         information['filepath'] = new_path
 244         return self._nopostoverwrites, information
 245
 246
 247 class FFmpegVideoConvertorPP(FFmpegPostProcessor):
 248     def __init__(self, downloader=None, preferedformat=None):
 249         super(FFmpegVideoConvertorPP, self).__init__(downloader)
 250         self._preferedformat = preferedformat
 251
 252     def run(self, information):
 253         path = information['filepath']
 254         prefix, sep, ext = path.rpartition('.')
 255         outpath = prefix + sep + self._preferedformat
 256         if information['ext'] == self._preferedformat:
 257             self._downloader.to_screen('[ffmpeg] Not converting video file %s - already is in target format %s' % (path, self._preferedformat))
 258             return True, information
 259         self._downloader.to_screen('[' + 'ffmpeg' + '] Converting video from %s to %s, Destination: ' % (information['ext'], self._preferedformat) + outpath)
 260         self.run_ffmpeg(path, outpath, [])
 261         information['filepath'] = outpath
 262         information['format'] = self._preferedformat
 263         information['ext'] = self._preferedformat
 264         return False, information
 265
 266
 267 class FFmpegEmbedSubtitlePP(FFmpegPostProcessor):
 268     # See http://www.loc.gov/standards/iso639-2/ISO-639-2_utf-8.txt
 269     _lang_map = {
 270         'aa': 'aar',
 271         'ab': 'abk',
 272         'ae': 'ave',
 273         'af': 'afr',
 274         'ak': 'aka',
 275         'am': 'amh',
 276         'an': 'arg',
 277         'ar': 'ara',
 278         'as': 'asm',
 279         'av': 'ava',
 280         'ay': 'aym',
 281         'az': 'aze',
 282         'ba': 'bak',
 283         'be': 'bel',
 284         'bg': 'bul',
 285         'bh': 'bih',
 286         'bi': 'bis',
 287         'bm': 'bam',
 288         'bn': 'ben',
 289         'bo': 'bod',
 290         'br': 'bre',
 291         'bs': 'bos',
 292         'ca': 'cat',
 293         'ce': 'che',
 294         'ch': 'cha',
 295         'co': 'cos',
 296         'cr': 'cre',
 297         'cs': 'ces',
 298         'cu': 'chu',
 299         'cv': 'chv',
 300         'cy': 'cym',
 301         'da': 'dan',
 302         'de': 'deu',
 303         'dv': 'div',
 304         'dz': 'dzo',
 305         'ee': 'ewe',
 306         'el': 'ell',
 307         'en': 'eng',
 308         'eo': 'epo',
 309         'es': 'spa',
 310         'et': 'est',
 311         'eu': 'eus',
 312         'fa': 'fas',
 313         'ff': 'ful',
 314         'fi': 'fin',
 315         'fj': 'fij',
 316         'fo': 'fao',
 317         'fr': 'fra',
 318         'fy': 'fry',
 319         'ga': 'gle',
 320         'gd': 'gla',
 321         'gl': 'glg',
 322         'gn': 'grn',
 323         'gu': 'guj',
 324         'gv': 'glv',
 325         'ha': 'hau',
 326         'he': 'heb',
 327         'hi': 'hin',
 328         'ho': 'hmo',
 329         'hr': 'hrv',
 330         'ht': 'hat',
 331         'hu': 'hun',
 332         'hy': 'hye',
 333         'hz': 'her',
 334         'ia': 'ina',
 335         'id': 'ind',
 336         'ie': 'ile',
 337         'ig': 'ibo',
 338         'ii': 'iii',
 339         'ik': 'ipk',
 340         'io': 'ido',
 341         'is': 'isl',
 342         'it': 'ita',
 343         'iu': 'iku',
 344         'ja': 'jpn',
 345         'jv': 'jav',
 346         'ka': 'kat',
 347         'kg': 'kon',
 348         'ki': 'kik',
 349         'kj': 'kua',
 350         'kk': 'kaz',
 351         'kl': 'kal',
 352         'km': 'khm',
 353         'kn': 'kan',
 354         'ko': 'kor',
 355         'kr': 'kau',
 356         'ks': 'kas',
 357         'ku': 'kur',
 358         'kv': 'kom',
 359         'kw': 'cor',
 360         'ky': 'kir',
 361         'la': 'lat',
 362         'lb': 'ltz',
 363         'lg': 'lug',
 364         'li': 'lim',
 365         'ln': 'lin',
 366         'lo': 'lao',
 367         'lt': 'lit',
 368         'lu': 'lub',
 369         'lv': 'lav',
 370         'mg': 'mlg',
 371         'mh': 'mah',
 372         'mi': 'mri',
 373         'mk': 'mkd',
 374         'ml': 'mal',
 375         'mn': 'mon',
 376         'mr': 'mar',
 377         'ms': 'msa',
 378         'mt': 'mlt',
 379         'my': 'mya',
 380         'na': 'nau',
 381         'nb': 'nob',
 382         'nd': 'nde',
 383         'ne': 'nep',
 384         'ng': 'ndo',
 385         'nl': 'nld',
 386         'nn': 'nno',
 387         'no': 'nor',
 388         'nr': 'nbl',
 389         'nv': 'nav',
 390         'ny': 'nya',
 391         'oc': 'oci',
 392         'oj': 'oji',
 393         'om': 'orm',
 394         'or': 'ori',
 395         'os': 'oss',
 396         'pa': 'pan',
 397         'pi': 'pli',
 398         'pl': 'pol',
 399         'ps': 'pus',
 400         'pt': 'por',
 401         'qu': 'que',
 402         'rm': 'roh',
 403         'rn': 'run',
 404         'ro': 'ron',
 405         'ru': 'rus',
 406         'rw': 'kin',
 407         'sa': 'san',
 408         'sc': 'srd',
 409         'sd': 'snd',
 410         'se': 'sme',
 411         'sg': 'sag',
 412         'si': 'sin',
 413         'sk': 'slk',
 414         'sl': 'slv',
 415         'sm': 'smo',
 416         'sn': 'sna',
 417         'so': 'som',
 418         'sq': 'sqi',
 419         'sr': 'srp',
 420         'ss': 'ssw',
 421         'st': 'sot',
 422         'su': 'sun',
 423         'sv': 'swe',
 424         'sw': 'swa',
 425         'ta': 'tam',
 426         'te': 'tel',
 427         'tg': 'tgk',
 428         'th': 'tha',
 429         'ti': 'tir',
 430         'tk': 'tuk',
 431         'tl': 'tgl',
 432         'tn': 'tsn',
 433         'to': 'ton',
 434         'tr': 'tur',
 435         'ts': 'tso',
 436         'tt': 'tat',
 437         'tw': 'twi',
 438         'ty': 'tah',
 439         'ug': 'uig',
 440         'uk': 'ukr',
 441         'ur': 'urd',
 442         'uz': 'uzb',
 443         've': 'ven',
 444         'vi': 'vie',
 445         'vo': 'vol',
 446         'wa': 'wln',
 447         'wo': 'wol',
 448         'xh': 'xho',
 449         'yi': 'yid',
 450         'yo': 'yor',
 451         'za': 'zha',
 452         'zh': 'zho',
 453         'zu': 'zul',
 454     }
 455
 456     def __init__(self, downloader=None, subtitlesformat='srt'):
 457         super(FFmpegEmbedSubtitlePP, self).__init__(downloader)
 458         self._subformat = subtitlesformat
 459
 460     @classmethod
 461     def _conver_lang_code(cls, code):
 462         """Convert language code from ISO 639-1 to ISO 639-2/T"""
 463         return cls._lang_map.get(code[:2])
 464
 465     def run(self, information):
 466         if information['ext'] != 'mp4':
 467             self._downloader.to_screen('[ffmpeg] Subtitles can only be embedded in mp4 files')
 468             return True, information
 469         if not information.get('subtitles'):
 470             self._downloader.to_screen('[ffmpeg] There aren\'t any subtitles to embed')
 471             return True, information
 472
 473         sub_langs = [key for key in information['subtitles']]
 474         filename = information['filepath']
 475         input_files = [filename] + [subtitles_filename(filename, lang, self._subformat) for lang in sub_langs]
 476
 477         opts = [
 478             '-map', '0',
 479             '-c', 'copy',
 480             # Don't copy the existing subtitles, we may be running the
 481             # postprocessor a second time
 482             '-map', '-0:s',
 483             '-c:s', 'mov_text',
 484         ]
 485         for (i, lang) in enumerate(sub_langs):
 486             opts.extend(['-map', '%d:0' % (i + 1)])
 487             lang_code = self._conver_lang_code(lang)
 488             if lang_code is not None:
 489                 opts.extend(['-metadata:s:s:%d' % i, 'language=%s' % lang_code])
 490
 491         temp_filename = prepend_extension(filename, 'temp')
 492         self._downloader.to_screen('[ffmpeg] Embedding subtitles in \'%s\'' % filename)
 493         self.run_ffmpeg_multiple_files(input_files, temp_filename, opts)
 494         os.remove(encodeFilename(filename))
 495         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 496
 497         return True, information
 498
 499
 500 class FFmpegMetadataPP(FFmpegPostProcessor):
 501     def run(self, info):
 502         metadata = {}
 503         if info.get('title') is not None:
 504             metadata['title'] = info['title']
 505         if info.get('upload_date') is not None:
 506             metadata['date'] = info['upload_date']
 507         if info.get('uploader') is not None:
 508             metadata['artist'] = info['uploader']
 509         elif info.get('uploader_id') is not None:
 510             metadata['artist'] = info['uploader_id']
 511         if info.get('description') is not None:
 512             metadata['description'] = info['description']
 513             metadata['comment'] = info['description']
 514         if info.get('webpage_url') is not None:
 515             metadata['purl'] = info['webpage_url']
 516
 517         if not metadata:
 518             self._downloader.to_screen('[ffmpeg] There isn\'t any metadata to add')
 519             return True, info
 520
 521         filename = info['filepath']
 522         temp_filename = prepend_extension(filename, 'temp')
 523
 524         if info['ext'] == 'm4a':
 525             options = ['-vn', '-acodec', 'copy']
 526         else:
 527             options = ['-c', 'copy']
 528
 529         for (name, value) in metadata.items():
 530             options.extend(['-metadata', '%s=%s' % (name, value)])
 531
 532         self._downloader.to_screen('[ffmpeg] Adding metadata to \'%s\'' % filename)
 533         self.run_ffmpeg(filename, temp_filename, options)
 534         os.remove(encodeFilename(filename))
 535         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 536         return True, info
 537
 538
 539 class FFmpegMergerPP(FFmpegPostProcessor):
 540     def run(self, info):
 541         filename = info['filepath']
 542         args = ['-c', 'copy', '-map', '0:v:0', '-map', '1:a:0']
 543         self._downloader.to_screen('[ffmpeg] Merging formats into "%s"' % filename)
 544         self.run_ffmpeg_multiple_files(info['__files_to_merge'], filename, args)
 545         return True, info
 546
 547
 548 class FFmpegAudioFixPP(FFmpegPostProcessor):
 549     def run(self, info):
 550         filename = info['filepath']
 551         temp_filename = prepend_extension(filename, 'temp')
 552
 553         options = ['-vn', '-acodec', 'copy']
 554         self._downloader.to_screen('[ffmpeg] Fixing audio file "%s"' % filename)
 555         self.run_ffmpeg(filename, temp_filename, options)
 556
 557         os.remove(encodeFilename(filename))
 558         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 559
 560         return True, info
 561
 562
 563 class FFmpegFixupStretchedPP(FFmpegPostProcessor):
 564     def run(self, info):
 565         stretched_ratio = info.get('stretched_ratio')
 566         if stretched_ratio is None or stretched_ratio == 1:
 567             return True, info
 568
 569         filename = info['filepath']
 570         temp_filename = prepend_extension(filename, 'temp')
 571
 572         options = ['-c', 'copy', '-aspect', '%f' % stretched_ratio]
 573         self._downloader.to_screen('[ffmpeg] Fixing aspect ratio in "%s"' % filename)
 574         self.run_ffmpeg(filename, temp_filename, options)
 575
 576         os.remove(encodeFilename(filename))
 577         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 578
 579         return True, info
 580
 581
 582 class FFmpegFixupM4aPP(FFmpegPostProcessor):
 583     def run(self, info):
 584         if info.get('container') != 'm4a_dash':
 585             return True, info
 586
 587         filename = info['filepath']
 588         temp_filename = prepend_extension(filename, 'temp')
 589
 590         options = ['-c', 'copy', '-f', 'mp4']
 591         self._downloader.to_screen('[ffmpeg] Correcting container in "%s"' % filename)
 592         self.run_ffmpeg(filename, temp_filename, options)
 593
 594         os.remove(encodeFilename(filename))
 595         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 596
 597         return True, info