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  43             warning = 'Your copy of %s is outdated, update %s to version %s or newer if you encounter any errors.'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