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