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.basename == '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  43             warning = 'Your copy of %s is outdated, update %s to version %s or newer if you encounter any errors.'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     @property
 112     def available(self):
 113         return self.basename is not None
 114
 115     @property
 116     def executable(self):
 117         return self._paths[self.basename]
 118
 119     @property
 120     def probe_executable(self):
 121         return self._paths[self.probe_basename]
 122
 123     def run_ffmpeg_multiple_files(self, input_paths, out_path, opts):
 124         self.check_version()
 125
 126         oldest_mtime = min(
 127             os.stat(encodeFilename(path)).st_mtime for path in input_paths)
 128
 129         files_cmd = []
 130         for path in input_paths:
 131             files_cmd.extend([encodeArgument('-i'), encodeFilename(path, True)])
 132         cmd = ([encodeFilename(self.executable, True), encodeArgument('-y')] +
 133                files_cmd +
 134                [encodeArgument(o) for o in opts] +
 135                [encodeFilename(self._ffmpeg_filename_argument(out_path), True)])
 136
 137         if self._downloader.params.get('verbose', False):
 138             self._downloader.to_screen('[debug] ffmpeg command line: %s' % shell_quote(cmd))
 139         p = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE, stdin=subprocess.PIPE)
 140         stdout, stderr = p.communicate()
 141         if p.returncode != 0:
 142             stderr = stderr.decode('utf-8', 'replace')
 143             msg = stderr.strip().split('\n')[-1]
 144             raise FFmpegPostProcessorError(msg)
 145         os.utime(encodeFilename(out_path), (oldest_mtime, oldest_mtime))
 146         if self._deletetempfiles:
 147             for ipath in input_paths:
 148                 os.remove(ipath)
 149
 150     def run_ffmpeg(self, path, out_path, opts):
 151         self.run_ffmpeg_multiple_files([path], out_path, opts)
 152
 153     def _ffmpeg_filename_argument(self, fn):
 154         # ffmpeg broke --, see https://ffmpeg.org/trac/ffmpeg/ticket/2127 for details
 155         if fn.startswith('-'):
 156             return './' + fn
 157         return fn
 158
 159
 160 class FFmpegExtractAudioPP(FFmpegPostProcessor):
 161     def __init__(self, downloader=None, preferredcodec=None, preferredquality=None, nopostoverwrites=False):
 162         FFmpegPostProcessor.__init__(self, downloader)
 163         if preferredcodec is None:
 164             preferredcodec = 'best'
 165         self._preferredcodec = preferredcodec
 166         self._preferredquality = preferredquality
 167         self._nopostoverwrites = nopostoverwrites
 168
 169     def get_audio_codec(self, path):
 170
 171         if not self.probe_executable:
 172             raise PostProcessingError('ffprobe or avprobe not found. Please install one.')
 173         try:
 174             cmd = [
 175                 encodeFilename(self.probe_executable, True),
 176                 encodeArgument('-show_streams'),
 177                 encodeFilename(self._ffmpeg_filename_argument(path), True)]
 178             if self._downloader.params.get('verbose', False):
 179                 self._downloader.to_screen('[debug] %s command line: %s' % (self.basename, shell_quote(cmd)))
 180             handle = subprocess.Popen(cmd, stderr=compat_subprocess_get_DEVNULL(), stdout=subprocess.PIPE, stdin=subprocess.PIPE)
 181             output = handle.communicate()[0]
 182             if handle.wait() != 0:
 183                 return None
 184         except (IOError, OSError):
 185             return None
 186         audio_codec = None
 187         for line in output.decode('ascii', 'ignore').split('\n'):
 188             if line.startswith('codec_name='):
 189                 audio_codec = line.split('=')[1].strip()
 190             elif line.strip() == 'codec_type=audio' and audio_codec is not None:
 191                 return audio_codec
 192         return None
 193
 194     def run_ffmpeg(self, path, out_path, codec, more_opts):
 195         if codec is None:
 196             acodec_opts = []
 197         else:
 198             acodec_opts = ['-acodec', codec]
 199         opts = ['-vn'] + acodec_opts + more_opts
 200         try:
 201             FFmpegPostProcessor.run_ffmpeg(self, path, out_path, opts)
 202         except FFmpegPostProcessorError as err:
 203             raise AudioConversionError(err.msg)
 204
 205     def run(self, information):
 206         path = information['filepath']
 207
 208         filecodec = self.get_audio_codec(path)
 209         if filecodec is None:
 210             raise PostProcessingError('WARNING: unable to obtain file audio codec with ffprobe')
 211
 212         more_opts = []
 213         if self._preferredcodec == 'best' or self._preferredcodec == filecodec or (self._preferredcodec == 'm4a' and filecodec == 'aac'):
 214             if filecodec == 'aac' and self._preferredcodec in ['m4a', 'best']:
 215                 # Lossless, but in another container
 216                 acodec = 'copy'
 217                 extension = 'm4a'
 218                 more_opts = ['-bsf:a', 'aac_adtstoasc']
 219             elif filecodec in ['aac', 'mp3', 'vorbis', 'opus']:
 220                 # Lossless if possible
 221                 acodec = 'copy'
 222                 extension = filecodec
 223                 if filecodec == 'aac':
 224                     more_opts = ['-f', 'adts']
 225                 if filecodec == 'vorbis':
 226                     extension = 'ogg'
 227             else:
 228                 # MP3 otherwise.
 229                 acodec = 'libmp3lame'
 230                 extension = 'mp3'
 231                 more_opts = []
 232                 if self._preferredquality is not None:
 233                     if int(self._preferredquality) < 10:
 234                         more_opts += ['-q:a', self._preferredquality]
 235                     else:
 236                         more_opts += ['-b:a', self._preferredquality + 'k']
 237         else:
 238             # We convert the audio (lossy)
 239             acodec = {'mp3': 'libmp3lame', 'aac': 'aac', 'm4a': 'aac', 'opus': 'opus', 'vorbis': 'libvorbis', 'wav': None}[self._preferredcodec]
 240             extension = self._preferredcodec
 241             more_opts = []
 242             if self._preferredquality is not None:
 243                 # The opus codec doesn't support the -aq option
 244                 if int(self._preferredquality) < 10 and extension != 'opus':
 245                     more_opts += ['-q:a', self._preferredquality]
 246                 else:
 247                     more_opts += ['-b:a', self._preferredquality + 'k']
 248             if self._preferredcodec == 'aac':
 249                 more_opts += ['-f', 'adts']
 250             if self._preferredcodec == 'm4a':
 251                 more_opts += ['-bsf:a', 'aac_adtstoasc']
 252             if self._preferredcodec == 'vorbis':
 253                 extension = 'ogg'
 254             if self._preferredcodec == 'wav':
 255                 extension = 'wav'
 256                 more_opts += ['-f', 'wav']
 257
 258         prefix, sep, ext = path.rpartition('.')  # not os.path.splitext, since the latter does not work on unicode in all setups
 259         new_path = prefix + sep + extension
 260
 261         # If we download foo.mp3 and convert it to... foo.mp3, then don't delete foo.mp3, silly.
 262         if new_path == path:
 263             self._nopostoverwrites = True
 264
 265         try:
 266             if self._nopostoverwrites and os.path.exists(encodeFilename(new_path)):
 267                 self._downloader.to_screen('[youtube] Post-process file %s exists, skipping' % new_path)
 268             else:
 269                 self._downloader.to_screen('[' + self.basename + '] Destination: ' + new_path)
 270                 self.run_ffmpeg(path, new_path, acodec, more_opts)
 271         except:
 272             etype, e, tb = sys.exc_info()
 273             if isinstance(e, AudioConversionError):
 274                 msg = 'audio conversion failed: ' + e.msg
 275             else:
 276                 msg = 'error running ' + self.basename
 277             raise PostProcessingError(msg)
 278
 279         # Try to update the date time for extracted audio file.
 280         if information.get('filetime') is not None:
 281             try:
 282                 os.utime(encodeFilename(new_path), (time.time(), information['filetime']))
 283             except:
 284                 self._downloader.report_warning('Cannot update utime of audio file')
 285
 286         information['filepath'] = new_path
 287         return self._nopostoverwrites, information
 288
 289
 290 class FFmpegVideoConvertorPP(FFmpegPostProcessor):
 291     def __init__(self, downloader=None, preferedformat=None):
 292         super(FFmpegVideoConvertorPP, self).__init__(downloader)
 293         self._preferedformat = preferedformat
 294
 295     def run(self, information):
 296         path = information['filepath']
 297         prefix, sep, ext = path.rpartition('.')
 298         outpath = prefix + sep + self._preferedformat
 299         if information['ext'] == self._preferedformat:
 300             self._downloader.to_screen('[ffmpeg] Not converting video file %s - already is in target format %s' % (path, self._preferedformat))
 301             return True, information
 302         self._downloader.to_screen('[' + 'ffmpeg' + '] Converting video from %s to %s, Destination: ' % (information['ext'], self._preferedformat) + outpath)
 303         self.run_ffmpeg(path, outpath, [])
 304         information['filepath'] = outpath
 305         information['format'] = self._preferedformat
 306         information['ext'] = self._preferedformat
 307         return False, information
 308
 309
 310 class FFmpegEmbedSubtitlePP(FFmpegPostProcessor):
 311     # See http://www.loc.gov/standards/iso639-2/ISO-639-2_utf-8.txt
 312     _lang_map = {
 313         'aa': 'aar',
 314         'ab': 'abk',
 315         'ae': 'ave',
 316         'af': 'afr',
 317         'ak': 'aka',
 318         'am': 'amh',
 319         'an': 'arg',
 320         'ar': 'ara',
 321         'as': 'asm',
 322         'av': 'ava',
 323         'ay': 'aym',
 324         'az': 'aze',
 325         'ba': 'bak',
 326         'be': 'bel',
 327         'bg': 'bul',
 328         'bh': 'bih',
 329         'bi': 'bis',
 330         'bm': 'bam',
 331         'bn': 'ben',
 332         'bo': 'bod',
 333         'br': 'bre',
 334         'bs': 'bos',
 335         'ca': 'cat',
 336         'ce': 'che',
 337         'ch': 'cha',
 338         'co': 'cos',
 339         'cr': 'cre',
 340         'cs': 'ces',
 341         'cu': 'chu',
 342         'cv': 'chv',
 343         'cy': 'cym',
 344         'da': 'dan',
 345         'de': 'deu',
 346         'dv': 'div',
 347         'dz': 'dzo',
 348         'ee': 'ewe',
 349         'el': 'ell',
 350         'en': 'eng',
 351         'eo': 'epo',
 352         'es': 'spa',
 353         'et': 'est',
 354         'eu': 'eus',
 355         'fa': 'fas',
 356         'ff': 'ful',
 357         'fi': 'fin',
 358         'fj': 'fij',
 359         'fo': 'fao',
 360         'fr': 'fra',
 361         'fy': 'fry',
 362         'ga': 'gle',
 363         'gd': 'gla',
 364         'gl': 'glg',
 365         'gn': 'grn',
 366         'gu': 'guj',
 367         'gv': 'glv',
 368         'ha': 'hau',
 369         'he': 'heb',
 370         'hi': 'hin',
 371         'ho': 'hmo',
 372         'hr': 'hrv',
 373         'ht': 'hat',
 374         'hu': 'hun',
 375         'hy': 'hye',
 376         'hz': 'her',
 377         'ia': 'ina',
 378         'id': 'ind',
 379         'ie': 'ile',
 380         'ig': 'ibo',
 381         'ii': 'iii',
 382         'ik': 'ipk',
 383         'io': 'ido',
 384         'is': 'isl',
 385         'it': 'ita',
 386         'iu': 'iku',
 387         'ja': 'jpn',
 388         'jv': 'jav',
 389         'ka': 'kat',
 390         'kg': 'kon',
 391         'ki': 'kik',
 392         'kj': 'kua',
 393         'kk': 'kaz',
 394         'kl': 'kal',
 395         'km': 'khm',
 396         'kn': 'kan',
 397         'ko': 'kor',
 398         'kr': 'kau',
 399         'ks': 'kas',
 400         'ku': 'kur',
 401         'kv': 'kom',
 402         'kw': 'cor',
 403         'ky': 'kir',
 404         'la': 'lat',
 405         'lb': 'ltz',
 406         'lg': 'lug',
 407         'li': 'lim',
 408         'ln': 'lin',
 409         'lo': 'lao',
 410         'lt': 'lit',
 411         'lu': 'lub',
 412         'lv': 'lav',
 413         'mg': 'mlg',
 414         'mh': 'mah',
 415         'mi': 'mri',
 416         'mk': 'mkd',
 417         'ml': 'mal',
 418         'mn': 'mon',
 419         'mr': 'mar',
 420         'ms': 'msa',
 421         'mt': 'mlt',
 422         'my': 'mya',
 423         'na': 'nau',
 424         'nb': 'nob',
 425         'nd': 'nde',
 426         'ne': 'nep',
 427         'ng': 'ndo',
 428         'nl': 'nld',
 429         'nn': 'nno',
 430         'no': 'nor',
 431         'nr': 'nbl',
 432         'nv': 'nav',
 433         'ny': 'nya',
 434         'oc': 'oci',
 435         'oj': 'oji',
 436         'om': 'orm',
 437         'or': 'ori',
 438         'os': 'oss',
 439         'pa': 'pan',
 440         'pi': 'pli',
 441         'pl': 'pol',
 442         'ps': 'pus',
 443         'pt': 'por',
 444         'qu': 'que',
 445         'rm': 'roh',
 446         'rn': 'run',
 447         'ro': 'ron',
 448         'ru': 'rus',
 449         'rw': 'kin',
 450         'sa': 'san',
 451         'sc': 'srd',
 452         'sd': 'snd',
 453         'se': 'sme',
 454         'sg': 'sag',
 455         'si': 'sin',
 456         'sk': 'slk',
 457         'sl': 'slv',
 458         'sm': 'smo',
 459         'sn': 'sna',
 460         'so': 'som',
 461         'sq': 'sqi',
 462         'sr': 'srp',
 463         'ss': 'ssw',
 464         'st': 'sot',
 465         'su': 'sun',
 466         'sv': 'swe',
 467         'sw': 'swa',
 468         'ta': 'tam',
 469         'te': 'tel',
 470         'tg': 'tgk',
 471         'th': 'tha',
 472         'ti': 'tir',
 473         'tk': 'tuk',
 474         'tl': 'tgl',
 475         'tn': 'tsn',
 476         'to': 'ton',
 477         'tr': 'tur',
 478         'ts': 'tso',
 479         'tt': 'tat',
 480         'tw': 'twi',
 481         'ty': 'tah',
 482         'ug': 'uig',
 483         'uk': 'ukr',
 484         'ur': 'urd',
 485         'uz': 'uzb',
 486         've': 'ven',
 487         'vi': 'vie',
 488         'vo': 'vol',
 489         'wa': 'wln',
 490         'wo': 'wol',
 491         'xh': 'xho',
 492         'yi': 'yid',
 493         'yo': 'yor',
 494         'za': 'zha',
 495         'zh': 'zho',
 496         'zu': 'zul',
 497     }
 498
 499     @classmethod
 500     def _conver_lang_code(cls, code):
 501         """Convert language code from ISO 639-1 to ISO 639-2/T"""
 502         return cls._lang_map.get(code[:2])
 503
 504     def run(self, information):
 505         if information['ext'] != 'mp4':
 506             self._downloader.to_screen('[ffmpeg] Subtitles can only be embedded in mp4 files')
 507             return True, information
 508         subtitles = information.get('requested_subtitles')
 509         if not subtitles:
 510             self._downloader.to_screen('[ffmpeg] There aren\'t any subtitles to embed')
 511             return True, information
 512
 513         sub_langs = list(subtitles.keys())
 514         filename = information['filepath']
 515         input_files = [filename] + [subtitles_filename(filename, lang, sub_info['ext']) for lang, sub_info in subtitles.items()]
 516
 517         opts = [
 518             '-map', '0',
 519             '-c', 'copy',
 520             # Don't copy the existing subtitles, we may be running the
 521             # postprocessor a second time
 522             '-map', '-0:s',
 523             '-c:s', 'mov_text',
 524         ]
 525         for (i, lang) in enumerate(sub_langs):
 526             opts.extend(['-map', '%d:0' % (i + 1)])
 527             lang_code = self._conver_lang_code(lang)
 528             if lang_code is not None:
 529                 opts.extend(['-metadata:s:s:%d' % i, 'language=%s' % lang_code])
 530
 531         temp_filename = prepend_extension(filename, 'temp')
 532         self._downloader.to_screen('[ffmpeg] Embedding subtitles in \'%s\'' % filename)
 533         self.run_ffmpeg_multiple_files(input_files, temp_filename, opts)
 534         os.remove(encodeFilename(filename))
 535         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 536
 537         return True, information
 538
 539
 540 class FFmpegMetadataPP(FFmpegPostProcessor):
 541     def run(self, info):
 542         metadata = {}
 543         if info.get('title') is not None:
 544             metadata['title'] = info['title']
 545         if info.get('upload_date') is not None:
 546             metadata['date'] = info['upload_date']
 547         if info.get('uploader') is not None:
 548             metadata['artist'] = info['uploader']
 549         elif info.get('uploader_id') is not None:
 550             metadata['artist'] = info['uploader_id']
 551         if info.get('description') is not None:
 552             metadata['description'] = info['description']
 553             metadata['comment'] = info['description']
 554         if info.get('webpage_url') is not None:
 555             metadata['purl'] = info['webpage_url']
 556
 557         if not metadata:
 558             self._downloader.to_screen('[ffmpeg] There isn\'t any metadata to add')
 559             return True, info
 560
 561         filename = info['filepath']
 562         temp_filename = prepend_extension(filename, 'temp')
 563
 564         if info['ext'] == 'm4a':
 565             options = ['-vn', '-acodec', 'copy']
 566         else:
 567             options = ['-c', 'copy']
 568
 569         for (name, value) in metadata.items():
 570             options.extend(['-metadata', '%s=%s' % (name, value)])
 571
 572         self._downloader.to_screen('[ffmpeg] Adding metadata to \'%s\'' % filename)
 573         self.run_ffmpeg(filename, temp_filename, options)
 574         os.remove(encodeFilename(filename))
 575         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 576         return True, info
 577
 578
 579 class FFmpegMergerPP(FFmpegPostProcessor):
 580     def run(self, info):
 581         filename = info['filepath']
 582         args = ['-c', 'copy', '-map', '0:v:0', '-map', '1:a:0']
 583         self._downloader.to_screen('[ffmpeg] Merging formats into "%s"' % filename)
 584         self.run_ffmpeg_multiple_files(info['__files_to_merge'], filename, args)
 585         return True, info
 586
 587
 588 class FFmpegAudioFixPP(FFmpegPostProcessor):
 589     def run(self, info):
 590         filename = info['filepath']
 591         temp_filename = prepend_extension(filename, 'temp')
 592
 593         options = ['-vn', '-acodec', 'copy']
 594         self._downloader.to_screen('[ffmpeg] Fixing audio file "%s"' % filename)
 595         self.run_ffmpeg(filename, temp_filename, options)
 596
 597         os.remove(encodeFilename(filename))
 598         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 599
 600         return True, info
 601
 602
 603 class FFmpegFixupStretchedPP(FFmpegPostProcessor):
 604     def run(self, info):
 605         stretched_ratio = info.get('stretched_ratio')
 606         if stretched_ratio is None or stretched_ratio == 1:
 607             return True, info
 608
 609         filename = info['filepath']
 610         temp_filename = prepend_extension(filename, 'temp')
 611
 612         options = ['-c', 'copy', '-aspect', '%f' % stretched_ratio]
 613         self._downloader.to_screen('[ffmpeg] Fixing aspect ratio in "%s"' % filename)
 614         self.run_ffmpeg(filename, temp_filename, options)
 615
 616         os.remove(encodeFilename(filename))
 617         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 618
 619         return True, info
 620
 621
 622 class FFmpegFixupM4aPP(FFmpegPostProcessor):
 623     def run(self, info):
 624         if info.get('container') != 'm4a_dash':
 625             return True, info
 626
 627         filename = info['filepath']
 628         temp_filename = prepend_extension(filename, 'temp')
 629
 630         options = ['-c', 'copy', '-f', 'mp4']
 631         self._downloader.to_screen('[ffmpeg] Correcting container in "%s"' % filename)
 632         self.run_ffmpeg(filename, temp_filename, options)
 633
 634         os.remove(encodeFilename(filename))
 635         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 636
 637         return True, info