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  43             warning = 'Your copy of %s is outdated, update %s to version %s or newer if you encounter any errors.'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