Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/postprocessor/ffmpeg.py

   1 import os
   2 import subprocess
   3 import sys
   4 import time
   5
   6
   7 from .common import AudioConversionError, PostProcessor
   8
   9 from ..utils import (
  10     check_executable,
  11     compat_subprocess_get_DEVNULL,
  12     encodeArgument,
  13     encodeFilename,
  14     PostProcessingError,
  15     prepend_extension,
  16     shell_quote,
  17     subtitles_filename,
  18 )
  19
  20
  21 class FFmpegPostProcessorError(PostProcessingError):
  22     pass
  23
  24
  25 class FFmpegPostProcessor(PostProcessor):
  26     def __init__(self, downloader=None, deletetempfiles=False):
  27         PostProcessor.__init__(self, downloader)
  28         self._exes = self.detect_executables()
  29         self._deletetempfiles = deletetempfiles
  30
  31     @staticmethod
  32     def detect_executables():
  33         programs = ['avprobe', 'avconv', 'ffmpeg', 'ffprobe']
  34         return dict((program, check_executable(program, ['-version'])) for program in programs)
  35
  36     def _get_executable(self):
  37         if self._downloader.params.get('prefer_ffmpeg', False):
  38             return self._exes['ffmpeg'] or self._exes['avconv']
  39         else:
  40             return self._exes['avconv'] or self._exes['ffmpeg']
  41
  42     def _uses_avconv(self):
  43         return self._get_executable() == self._exes['avconv']
  44
  45     def run_ffmpeg_multiple_files(self, input_paths, out_path, opts):
  46         if not self._get_executable():
  47             raise FFmpegPostProcessorError(u'ffmpeg or avconv not found. Please install one.')
  48
  49         files_cmd = []
  50         for path in input_paths:
  51             files_cmd.extend(['-i', encodeFilename(path, True)])
  52         cmd = ([self._get_executable(), '-y'] + files_cmd
  53                + [encodeArgument(o) for o in opts] +
  54                [encodeFilename(self._ffmpeg_filename_argument(out_path), True)])
  55
  56         if self._downloader.params.get('verbose', False):
  57             self._downloader.to_screen(u'[debug] ffmpeg command line: %s' % shell_quote(cmd))
  58         p = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
  59         stdout, stderr = p.communicate()
  60         if p.returncode != 0:
  61             stderr = stderr.decode('utf-8', 'replace')
  62             msg = stderr.strip().split('\n')[-1]
  63             raise FFmpegPostProcessorError(msg)
  64         if self._deletetempfiles:
  65             for ipath in input_paths:
  66                 os.remove(ipath)
  67
  68     def run_ffmpeg(self, path, out_path, opts):
  69         self.run_ffmpeg_multiple_files([path], out_path, opts)
  70
  71     def _ffmpeg_filename_argument(self, fn):
  72         # ffmpeg broke --, see https://ffmpeg.org/trac/ffmpeg/ticket/2127 for details
  73         if fn.startswith(u'-'):
  74             return u'./' + fn
  75         return fn
  76
  77
  78 class FFmpegExtractAudioPP(FFmpegPostProcessor):
  79     def __init__(self, downloader=None, preferredcodec=None, preferredquality=None, nopostoverwrites=False):
  80         FFmpegPostProcessor.__init__(self, downloader)
  81         if preferredcodec is None:
  82             preferredcodec = 'best'
  83         self._preferredcodec = preferredcodec
  84         self._preferredquality = preferredquality
  85         self._nopostoverwrites = nopostoverwrites
  86
  87     def get_audio_codec(self, path):
  88         if not self._exes['ffprobe'] and not self._exes['avprobe']:
  89             raise PostProcessingError(u'ffprobe or avprobe not found. Please install one.')
  90         try:
  91             cmd = [
  92                 self._exes['avprobe'] or self._exes['ffprobe'],
  93                 '-show_streams',
  94                 encodeFilename(self._ffmpeg_filename_argument(path), True)]
  95             handle = subprocess.Popen(cmd, stderr=compat_subprocess_get_DEVNULL(), stdout=subprocess.PIPE)
  96             output = handle.communicate()[0]
  97             if handle.wait() != 0:
  98                 return None
  99         except (IOError, OSError):
 100             return None
 101         audio_codec = None
 102         for line in output.decode('ascii', 'ignore').split('\n'):
 103             if line.startswith('codec_name='):
 104                 audio_codec = line.split('=')[1].strip()
 105             elif line.strip() == 'codec_type=audio' and audio_codec is not None:
 106                 return audio_codec
 107         return None
 108
 109     def run_ffmpeg(self, path, out_path, codec, more_opts):
 110         if codec is None:
 111             acodec_opts = []
 112         else:
 113             acodec_opts = ['-acodec', codec]
 114         opts = ['-vn'] + acodec_opts + more_opts
 115         try:
 116             FFmpegPostProcessor.run_ffmpeg(self, path, out_path, opts)
 117         except FFmpegPostProcessorError as err:
 118             raise AudioConversionError(err.msg)
 119
 120     def run(self, information):
 121         path = information['filepath']
 122
 123         filecodec = self.get_audio_codec(path)
 124         if filecodec is None:
 125             raise PostProcessingError(u'WARNING: unable to obtain file audio codec with ffprobe')
 126
 127         uses_avconv = self._uses_avconv()
 128         more_opts = []
 129         if self._preferredcodec == 'best' or self._preferredcodec == filecodec or (self._preferredcodec == 'm4a' and filecodec == 'aac'):
 130             if filecodec == 'aac' and self._preferredcodec in ['m4a', 'best']:
 131                 # Lossless, but in another container
 132                 acodec = 'copy'
 133                 extension = 'm4a'
 134                 more_opts = ['-bsf:a' if uses_avconv else '-absf', 'aac_adtstoasc']
 135             elif filecodec in ['aac', 'mp3', 'vorbis', 'opus']:
 136                 # Lossless if possible
 137                 acodec = 'copy'
 138                 extension = filecodec
 139                 if filecodec == 'aac':
 140                     more_opts = ['-f', 'adts']
 141                 if filecodec == 'vorbis':
 142                     extension = 'ogg'
 143             else:
 144                 # MP3 otherwise.
 145                 acodec = 'libmp3lame'
 146                 extension = 'mp3'
 147                 more_opts = []
 148                 if self._preferredquality is not None:
 149                     if int(self._preferredquality) < 10:
 150                         more_opts += ['-q:a' if uses_avconv else '-aq', self._preferredquality]
 151                     else:
 152                         more_opts += ['-b:a' if uses_avconv else '-ab', self._preferredquality + 'k']
 153         else:
 154             # We convert the audio (lossy)
 155             acodec = {'mp3': 'libmp3lame', 'aac': 'aac', 'm4a': 'aac', 'opus': 'opus', 'vorbis': 'libvorbis', 'wav': None}[self._preferredcodec]
 156             extension = self._preferredcodec
 157             more_opts = []
 158             if self._preferredquality is not None:
 159                 # The opus codec doesn't support the -aq option
 160                 if int(self._preferredquality) < 10 and extension != 'opus':
 161                     more_opts += ['-q:a' if uses_avconv else '-aq', self._preferredquality]
 162                 else:
 163                     more_opts += ['-b:a' if uses_avconv else '-ab', self._preferredquality + 'k']
 164             if self._preferredcodec == 'aac':
 165                 more_opts += ['-f', 'adts']
 166             if self._preferredcodec == 'm4a':
 167                 more_opts += ['-bsf:a' if uses_avconv else '-absf', 'aac_adtstoasc']
 168             if self._preferredcodec == 'vorbis':
 169                 extension = 'ogg'
 170             if self._preferredcodec == 'wav':
 171                 extension = 'wav'
 172                 more_opts += ['-f', 'wav']
 173
 174         prefix, sep, ext = path.rpartition(u'.') # not os.path.splitext, since the latter does not work on unicode in all setups
 175         new_path = prefix + sep + extension
 176
 177         # If we download foo.mp3 and convert it to... foo.mp3, then don't delete foo.mp3, silly.
 178         if new_path == path:
 179             self._nopostoverwrites = True
 180
 181         try:
 182             if self._nopostoverwrites and os.path.exists(encodeFilename(new_path)):
 183                 self._downloader.to_screen(u'[youtube] Post-process file %s exists, skipping' % new_path)
 184             else:
 185                 self._downloader.to_screen(u'[' + self._get_executable() + '] Destination: ' + new_path)
 186                 self.run_ffmpeg(path, new_path, acodec, more_opts)
 187         except:
 188             etype,e,tb = sys.exc_info()
 189             if isinstance(e, AudioConversionError):
 190                 msg = u'audio conversion failed: ' + e.msg
 191             else:
 192                 msg = u'error running ' + self._get_executable()
 193             raise PostProcessingError(msg)
 194
 195         # Try to update the date time for extracted audio file.
 196         if information.get('filetime') is not None:
 197             try:
 198                 os.utime(encodeFilename(new_path), (time.time(), information['filetime']))
 199             except:
 200                 self._downloader.report_warning(u'Cannot update utime of audio file')
 201
 202         information['filepath'] = new_path
 203         return self._nopostoverwrites,information
 204
 205
 206 class FFmpegVideoConvertor(FFmpegPostProcessor):
 207     def __init__(self, downloader=None,preferedformat=None):
 208         super(FFmpegVideoConvertor, self).__init__(downloader)
 209         self._preferedformat=preferedformat
 210
 211     def run(self, information):
 212         path = information['filepath']
 213         prefix, sep, ext = path.rpartition(u'.')
 214         outpath = prefix + sep + self._preferedformat
 215         if information['ext'] == self._preferedformat:
 216             self._downloader.to_screen(u'[ffmpeg] Not converting video file %s - already is in target format %s' % (path, self._preferedformat))
 217             return True,information
 218         self._downloader.to_screen(u'['+'ffmpeg'+'] Converting video from %s to %s, Destination: ' % (information['ext'], self._preferedformat) +outpath)
 219         self.run_ffmpeg(path, outpath, [])
 220         information['filepath'] = outpath
 221         information['format'] = self._preferedformat
 222         information['ext'] = self._preferedformat
 223         return False,information
 224
 225
 226 class FFmpegEmbedSubtitlePP(FFmpegPostProcessor):
 227     # See http://www.loc.gov/standards/iso639-2/ISO-639-2_utf-8.txt
 228     _lang_map = {
 229         'aa': 'aar',
 230         'ab': 'abk',
 231         'ae': 'ave',
 232         'af': 'afr',
 233         'ak': 'aka',
 234         'am': 'amh',
 235         'an': 'arg',
 236         'ar': 'ara',
 237         'as': 'asm',
 238         'av': 'ava',
 239         'ay': 'aym',
 240         'az': 'aze',
 241         'ba': 'bak',
 242         'be': 'bel',
 243         'bg': 'bul',
 244         'bh': 'bih',
 245         'bi': 'bis',
 246         'bm': 'bam',
 247         'bn': 'ben',
 248         'bo': 'bod',
 249         'br': 'bre',
 250         'bs': 'bos',
 251         'ca': 'cat',
 252         'ce': 'che',
 253         'ch': 'cha',
 254         'co': 'cos',
 255         'cr': 'cre',
 256         'cs': 'ces',
 257         'cu': 'chu',
 258         'cv': 'chv',
 259         'cy': 'cym',
 260         'da': 'dan',
 261         'de': 'deu',
 262         'dv': 'div',
 263         'dz': 'dzo',
 264         'ee': 'ewe',
 265         'el': 'ell',
 266         'en': 'eng',
 267         'eo': 'epo',
 268         'es': 'spa',
 269         'et': 'est',
 270         'eu': 'eus',
 271         'fa': 'fas',
 272         'ff': 'ful',
 273         'fi': 'fin',
 274         'fj': 'fij',
 275         'fo': 'fao',
 276         'fr': 'fra',
 277         'fy': 'fry',
 278         'ga': 'gle',
 279         'gd': 'gla',
 280         'gl': 'glg',
 281         'gn': 'grn',
 282         'gu': 'guj',
 283         'gv': 'glv',
 284         'ha': 'hau',
 285         'he': 'heb',
 286         'hi': 'hin',
 287         'ho': 'hmo',
 288         'hr': 'hrv',
 289         'ht': 'hat',
 290         'hu': 'hun',
 291         'hy': 'hye',
 292         'hz': 'her',
 293         'ia': 'ina',
 294         'id': 'ind',
 295         'ie': 'ile',
 296         'ig': 'ibo',
 297         'ii': 'iii',
 298         'ik': 'ipk',
 299         'io': 'ido',
 300         'is': 'isl',
 301         'it': 'ita',
 302         'iu': 'iku',
 303         'ja': 'jpn',
 304         'jv': 'jav',
 305         'ka': 'kat',
 306         'kg': 'kon',
 307         'ki': 'kik',
 308         'kj': 'kua',
 309         'kk': 'kaz',
 310         'kl': 'kal',
 311         'km': 'khm',
 312         'kn': 'kan',
 313         'ko': 'kor',
 314         'kr': 'kau',
 315         'ks': 'kas',
 316         'ku': 'kur',
 317         'kv': 'kom',
 318         'kw': 'cor',
 319         'ky': 'kir',
 320         'la': 'lat',
 321         'lb': 'ltz',
 322         'lg': 'lug',
 323         'li': 'lim',
 324         'ln': 'lin',
 325         'lo': 'lao',
 326         'lt': 'lit',
 327         'lu': 'lub',
 328         'lv': 'lav',
 329         'mg': 'mlg',
 330         'mh': 'mah',
 331         'mi': 'mri',
 332         'mk': 'mkd',
 333         'ml': 'mal',
 334         'mn': 'mon',
 335         'mr': 'mar',
 336         'ms': 'msa',
 337         'mt': 'mlt',
 338         'my': 'mya',
 339         'na': 'nau',
 340         'nb': 'nob',
 341         'nd': 'nde',
 342         'ne': 'nep',
 343         'ng': 'ndo',
 344         'nl': 'nld',
 345         'nn': 'nno',
 346         'no': 'nor',
 347         'nr': 'nbl',
 348         'nv': 'nav',
 349         'ny': 'nya',
 350         'oc': 'oci',
 351         'oj': 'oji',
 352         'om': 'orm',
 353         'or': 'ori',
 354         'os': 'oss',
 355         'pa': 'pan',
 356         'pi': 'pli',
 357         'pl': 'pol',
 358         'ps': 'pus',
 359         'pt': 'por',
 360         'qu': 'que',
 361         'rm': 'roh',
 362         'rn': 'run',
 363         'ro': 'ron',
 364         'ru': 'rus',
 365         'rw': 'kin',
 366         'sa': 'san',
 367         'sc': 'srd',
 368         'sd': 'snd',
 369         'se': 'sme',
 370         'sg': 'sag',
 371         'si': 'sin',
 372         'sk': 'slk',
 373         'sl': 'slv',
 374         'sm': 'smo',
 375         'sn': 'sna',
 376         'so': 'som',
 377         'sq': 'sqi',
 378         'sr': 'srp',
 379         'ss': 'ssw',
 380         'st': 'sot',
 381         'su': 'sun',
 382         'sv': 'swe',
 383         'sw': 'swa',
 384         'ta': 'tam',
 385         'te': 'tel',
 386         'tg': 'tgk',
 387         'th': 'tha',
 388         'ti': 'tir',
 389         'tk': 'tuk',
 390         'tl': 'tgl',
 391         'tn': 'tsn',
 392         'to': 'ton',
 393         'tr': 'tur',
 394         'ts': 'tso',
 395         'tt': 'tat',
 396         'tw': 'twi',
 397         'ty': 'tah',
 398         'ug': 'uig',
 399         'uk': 'ukr',
 400         'ur': 'urd',
 401         'uz': 'uzb',
 402         've': 'ven',
 403         'vi': 'vie',
 404         'vo': 'vol',
 405         'wa': 'wln',
 406         'wo': 'wol',
 407         'xh': 'xho',
 408         'yi': 'yid',
 409         'yo': 'yor',
 410         'za': 'zha',
 411         'zh': 'zho',
 412         'zu': 'zul',
 413     }
 414
 415     def __init__(self, downloader=None, subtitlesformat='srt'):
 416         super(FFmpegEmbedSubtitlePP, self).__init__(downloader)
 417         self._subformat = subtitlesformat
 418
 419     @classmethod
 420     def _conver_lang_code(cls, code):
 421         """Convert language code from ISO 639-1 to ISO 639-2/T"""
 422         return cls._lang_map.get(code[:2])
 423
 424     def run(self, information):
 425         if information['ext'] != u'mp4':
 426             self._downloader.to_screen(u'[ffmpeg] Subtitles can only be embedded in mp4 files')
 427             return True, information
 428         if not information.get('subtitles'):
 429             self._downloader.to_screen(u'[ffmpeg] There aren\'t any subtitles to embed')
 430             return True, information
 431
 432         sub_langs = [key for key in information['subtitles']]
 433         filename = information['filepath']
 434         input_files = [filename] + [subtitles_filename(filename, lang, self._subformat) for lang in sub_langs]
 435
 436         opts = ['-map', '0:0', '-map', '0:1', '-c:v', 'copy', '-c:a', 'copy']
 437         for (i, lang) in enumerate(sub_langs):
 438             opts.extend(['-map', '%d:0' % (i+1), '-c:s:%d' % i, 'mov_text'])
 439             lang_code = self._conver_lang_code(lang)
 440             if lang_code is not None:
 441                 opts.extend(['-metadata:s:s:%d' % i, 'language=%s' % lang_code])
 442         opts.extend(['-f', 'mp4'])
 443
 444         temp_filename = filename + u'.temp'
 445         self._downloader.to_screen(u'[ffmpeg] Embedding subtitles in \'%s\'' % filename)
 446         self.run_ffmpeg_multiple_files(input_files, temp_filename, opts)
 447         os.remove(encodeFilename(filename))
 448         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 449
 450         return True, information
 451
 452
 453 class FFmpegMetadataPP(FFmpegPostProcessor):
 454     def run(self, info):
 455         metadata = {}
 456         if info.get('title') is not None:
 457             metadata['title'] = info['title']
 458         if info.get('upload_date') is not None:
 459             metadata['date'] = info['upload_date']
 460         if info.get('uploader') is not None:
 461             metadata['artist'] = info['uploader']
 462         elif info.get('uploader_id') is not None:
 463             metadata['artist'] = info['uploader_id']
 464
 465         if not metadata:
 466             self._downloader.to_screen(u'[ffmpeg] There isn\'t any metadata to add')
 467             return True, info
 468
 469         filename = info['filepath']
 470         temp_filename = prepend_extension(filename, 'temp')
 471
 472         if info['ext'] == u'm4a':
 473             options = ['-vn', '-acodec', 'copy']
 474         else:
 475             options = ['-c', 'copy']
 476
 477         for (name, value) in metadata.items():
 478             options.extend(['-metadata', '%s=%s' % (name, value)])
 479
 480         self._downloader.to_screen(u'[ffmpeg] Adding metadata to \'%s\'' % filename)
 481         self.run_ffmpeg(filename, temp_filename, options)
 482         os.remove(encodeFilename(filename))
 483         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 484         return True, info
 485
 486
 487 class FFmpegMergerPP(FFmpegPostProcessor):
 488     def run(self, info):
 489         filename = info['filepath']
 490         args = ['-c', 'copy']
 491         self._downloader.to_screen(u'[ffmpeg] Merging formats into "%s"' % filename)
 492         self.run_ffmpeg_multiple_files(info['__files_to_merge'], filename, args)
 493         return True, info
 494
 495
 496 class FFmpegAudioFixPP(FFmpegPostProcessor):
 497     def run(self, info):
 498         filename = info['filepath']
 499         temp_filename = prepend_extension(filename, 'temp')
 500
 501         options = ['-vn', '-acodec', 'copy']
 502         self._downloader.to_screen(u'[ffmpeg] Fixing audio file "%s"' % filename)
 503         self.run_ffmpeg(filename, temp_filename, options)
 504
 505         os.remove(encodeFilename(filename))
 506         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 507
 508         return True, info