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

   1 import os
   2 import re
   3 import subprocess
   4 import sys
   5 import time
   6
   7
   8 from .common import AudioConversionError, PostProcessor
   9
  10 from ..utils import (
  11     compat_subprocess_get_DEVNULL,
  12     encodeArgument,
  13     encodeFilename,
  14     is_outdated_version,
  15     PostProcessingError,
  16     prepend_extension,
  17     shell_quote,
  18     subtitles_filename,
  19 )
  20
  21
  22 def get_version(executable):
  23     """ Returns the version of the specified executable,
  24     or False if the executable is not present """
  25     try:
  26         out, err = subprocess.Popen(
  27             [executable, '-version'],
  28             stdout=subprocess.PIPE, stderr=subprocess.STDOUT).communicate()
  29     except OSError:
  30         return False
  31     firstline = out.partition(b'\n')[0].decode('ascii', 'ignore')
  32     m = re.search(r'version\s+([0-9._-a-zA-Z]+)', firstline)
  33     if not m:
  34         return u'present'
  35     else:
  36         return m.group(1)
  37
  38
  39 class FFmpegPostProcessorError(PostProcessingError):
  40     pass
  41
  42
  43 class FFmpegPostProcessor(PostProcessor):
  44     def __init__(self, downloader=None, deletetempfiles=False):
  45         PostProcessor.__init__(self, downloader)
  46         self._versions = self.get_versions()
  47         self._deletetempfiles = deletetempfiles
  48
  49     def check_version(self):
  50         if not self._executable:
  51             raise FFmpegPostProcessorError(u'ffmpeg or avconv not found. Please install one.')
  52
  53         REQUIRED_VERSION = '1.0'
  54         if is_outdated_version(
  55                 self._versions[self._executable], REQUIRED_VERSION):
  56             warning = u'Your copy of %s is outdated, update %s to version %s or newer if you encounter any errors.' % (
  57                 self._executable, self._executable, REQUIRED_VERSION)
  58             if self._downloader:
  59                 self._downloader.report_warning(warning)
  60
  61     @staticmethod
  62     def get_versions():
  63         programs = ['avprobe', 'avconv', 'ffmpeg', 'ffprobe']
  64         return dict((program, get_version(program)) for program in programs)
  65
  66     @property
  67     def _executable(self):
  68         if self._downloader.params.get('prefer_ffmpeg', False):
  69             prefs = ('ffmpeg', 'avconv')
  70         else:
  71             prefs = ('avconv', 'ffmpeg')
  72         for p in prefs:
  73             if self._versions[p]:
  74                 return p
  75         return None
  76
  77     @property
  78     def _probe_executable(self):
  79         if self._downloader.params.get('prefer_ffmpeg', False):
  80             prefs = ('ffprobe', 'avprobe')
  81         else:
  82             prefs = ('avprobe', 'ffprobe')
  83         for p in prefs:
  84             if self._versions[p]:
  85                 return p
  86         return None
  87
  88     def _uses_avconv(self):
  89         return self._executable == 'avconv'
  90
  91     def run_ffmpeg_multiple_files(self, input_paths, out_path, opts):
  92         self.check_version()
  93
  94         files_cmd = []
  95         for path in input_paths:
  96             files_cmd.extend(['-i', encodeFilename(path, True)])
  97         cmd = ([self._executable, '-y'] + files_cmd
  98                + [encodeArgument(o) for o in opts] +
  99                [encodeFilename(self._ffmpeg_filename_argument(out_path), True)])
 100
 101         if self._downloader.params.get('verbose', False):
 102             self._downloader.to_screen(u'[debug] ffmpeg command line: %s' % shell_quote(cmd))
 103         p = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
 104         stdout, stderr = p.communicate()
 105         if p.returncode != 0:
 106             stderr = stderr.decode('utf-8', 'replace')
 107             msg = stderr.strip().split('\n')[-1]
 108             raise FFmpegPostProcessorError(msg)
 109         if self._deletetempfiles:
 110             for ipath in input_paths:
 111                 os.remove(ipath)
 112
 113     def run_ffmpeg(self, path, out_path, opts):
 114         self.run_ffmpeg_multiple_files([path], out_path, opts)
 115
 116     def _ffmpeg_filename_argument(self, fn):
 117         # ffmpeg broke --, see https://ffmpeg.org/trac/ffmpeg/ticket/2127 for details
 118         if fn.startswith(u'-'):
 119             return u'./' + fn
 120         return fn
 121
 122
 123 class FFmpegExtractAudioPP(FFmpegPostProcessor):
 124     def __init__(self, downloader=None, preferredcodec=None, preferredquality=None, nopostoverwrites=False):
 125         FFmpegPostProcessor.__init__(self, downloader)
 126         if preferredcodec is None:
 127             preferredcodec = 'best'
 128         self._preferredcodec = preferredcodec
 129         self._preferredquality = preferredquality
 130         self._nopostoverwrites = nopostoverwrites
 131
 132     def get_audio_codec(self, path):
 133
 134         if not self._probe_executable:
 135             raise PostProcessingError(u'ffprobe or avprobe not found. Please install one.')
 136         try:
 137             cmd = [
 138                 self._probe_executable,
 139                 '-show_streams',
 140                 encodeFilename(self._ffmpeg_filename_argument(path), True)]
 141             handle = subprocess.Popen(cmd, stderr=compat_subprocess_get_DEVNULL(), stdout=subprocess.PIPE)
 142             output = handle.communicate()[0]
 143             if handle.wait() != 0:
 144                 return None
 145         except (IOError, OSError):
 146             return None
 147         audio_codec = None
 148         for line in output.decode('ascii', 'ignore').split('\n'):
 149             if line.startswith('codec_name='):
 150                 audio_codec = line.split('=')[1].strip()
 151             elif line.strip() == 'codec_type=audio' and audio_codec is not None:
 152                 return audio_codec
 153         return None
 154
 155     def run_ffmpeg(self, path, out_path, codec, more_opts):
 156         if codec is None:
 157             acodec_opts = []
 158         else:
 159             acodec_opts = ['-acodec', codec]
 160         opts = ['-vn'] + acodec_opts + more_opts
 161         try:
 162             FFmpegPostProcessor.run_ffmpeg(self, path, out_path, opts)
 163         except FFmpegPostProcessorError as err:
 164             raise AudioConversionError(err.msg)
 165
 166     def run(self, information):
 167         path = information['filepath']
 168
 169         filecodec = self.get_audio_codec(path)
 170         if filecodec is None:
 171             raise PostProcessingError(u'WARNING: unable to obtain file audio codec with ffprobe')
 172
 173         uses_avconv = self._uses_avconv()
 174         more_opts = []
 175         if self._preferredcodec == 'best' or self._preferredcodec == filecodec or (self._preferredcodec == 'm4a' and filecodec == 'aac'):
 176             if filecodec == 'aac' and self._preferredcodec in ['m4a', 'best']:
 177                 # Lossless, but in another container
 178                 acodec = 'copy'
 179                 extension = 'm4a'
 180                 more_opts = ['-bsf:a' if uses_avconv else '-absf', 'aac_adtstoasc']
 181             elif filecodec in ['aac', 'mp3', 'vorbis', 'opus']:
 182                 # Lossless if possible
 183                 acodec = 'copy'
 184                 extension = filecodec
 185                 if filecodec == 'aac':
 186                     more_opts = ['-f', 'adts']
 187                 if filecodec == 'vorbis':
 188                     extension = 'ogg'
 189             else:
 190                 # MP3 otherwise.
 191                 acodec = 'libmp3lame'
 192                 extension = 'mp3'
 193                 more_opts = []
 194                 if self._preferredquality is not None:
 195                     if int(self._preferredquality) < 10:
 196                         more_opts += ['-q:a' if uses_avconv else '-aq', self._preferredquality]
 197                     else:
 198                         more_opts += ['-b:a' if uses_avconv else '-ab', self._preferredquality + 'k']
 199         else:
 200             # We convert the audio (lossy)
 201             acodec = {'mp3': 'libmp3lame', 'aac': 'aac', 'm4a': 'aac', 'opus': 'opus', 'vorbis': 'libvorbis', 'wav': None}[self._preferredcodec]
 202             extension = self._preferredcodec
 203             more_opts = []
 204             if self._preferredquality is not None:
 205                 # The opus codec doesn't support the -aq option
 206                 if int(self._preferredquality) < 10 and extension != 'opus':
 207                     more_opts += ['-q:a' if uses_avconv else '-aq', self._preferredquality]
 208                 else:
 209                     more_opts += ['-b:a' if uses_avconv else '-ab', self._preferredquality + 'k']
 210             if self._preferredcodec == 'aac':
 211                 more_opts += ['-f', 'adts']
 212             if self._preferredcodec == 'm4a':
 213                 more_opts += ['-bsf:a' if uses_avconv else '-absf', 'aac_adtstoasc']
 214             if self._preferredcodec == 'vorbis':
 215                 extension = 'ogg'
 216             if self._preferredcodec == 'wav':
 217                 extension = 'wav'
 218                 more_opts += ['-f', 'wav']
 219
 220         prefix, sep, ext = path.rpartition(u'.') # not os.path.splitext, since the latter does not work on unicode in all setups
 221         new_path = prefix + sep + extension
 222
 223         # If we download foo.mp3 and convert it to... foo.mp3, then don't delete foo.mp3, silly.
 224         if new_path == path:
 225             self._nopostoverwrites = True
 226
 227         try:
 228             if self._nopostoverwrites and os.path.exists(encodeFilename(new_path)):
 229                 self._downloader.to_screen(u'[youtube] Post-process file %s exists, skipping' % new_path)
 230             else:
 231                 self._downloader.to_screen(u'[' + self._executable + '] Destination: ' + new_path)
 232                 self.run_ffmpeg(path, new_path, acodec, more_opts)
 233         except:
 234             etype,e,tb = sys.exc_info()
 235             if isinstance(e, AudioConversionError):
 236                 msg = u'audio conversion failed: ' + e.msg
 237             else:
 238                 msg = u'error running ' + self._executable
 239             raise PostProcessingError(msg)
 240
 241         # Try to update the date time for extracted audio file.
 242         if information.get('filetime') is not None:
 243             try:
 244                 os.utime(encodeFilename(new_path), (time.time(), information['filetime']))
 245             except:
 246                 self._downloader.report_warning(u'Cannot update utime of audio file')
 247
 248         information['filepath'] = new_path
 249         return self._nopostoverwrites,information
 250
 251
 252 class FFmpegVideoConvertor(FFmpegPostProcessor):
 253     def __init__(self, downloader=None,preferedformat=None):
 254         super(FFmpegVideoConvertor, self).__init__(downloader)
 255         self._preferedformat=preferedformat
 256
 257     def run(self, information):
 258         path = information['filepath']
 259         prefix, sep, ext = path.rpartition(u'.')
 260         outpath = prefix + sep + self._preferedformat
 261         if information['ext'] == self._preferedformat:
 262             self._downloader.to_screen(u'[ffmpeg] Not converting video file %s - already is in target format %s' % (path, self._preferedformat))
 263             return True,information
 264         self._downloader.to_screen(u'['+'ffmpeg'+'] Converting video from %s to %s, Destination: ' % (information['ext'], self._preferedformat) +outpath)
 265         self.run_ffmpeg(path, outpath, [])
 266         information['filepath'] = outpath
 267         information['format'] = self._preferedformat
 268         information['ext'] = self._preferedformat
 269         return False,information
 270
 271
 272 class FFmpegEmbedSubtitlePP(FFmpegPostProcessor):
 273     # See http://www.loc.gov/standards/iso639-2/ISO-639-2_utf-8.txt
 274     _lang_map = {
 275         'aa': 'aar',
 276         'ab': 'abk',
 277         'ae': 'ave',
 278         'af': 'afr',
 279         'ak': 'aka',
 280         'am': 'amh',
 281         'an': 'arg',
 282         'ar': 'ara',
 283         'as': 'asm',
 284         'av': 'ava',
 285         'ay': 'aym',
 286         'az': 'aze',
 287         'ba': 'bak',
 288         'be': 'bel',
 289         'bg': 'bul',
 290         'bh': 'bih',
 291         'bi': 'bis',
 292         'bm': 'bam',
 293         'bn': 'ben',
 294         'bo': 'bod',
 295         'br': 'bre',
 296         'bs': 'bos',
 297         'ca': 'cat',
 298         'ce': 'che',
 299         'ch': 'cha',
 300         'co': 'cos',
 301         'cr': 'cre',
 302         'cs': 'ces',
 303         'cu': 'chu',
 304         'cv': 'chv',
 305         'cy': 'cym',
 306         'da': 'dan',
 307         'de': 'deu',
 308         'dv': 'div',
 309         'dz': 'dzo',
 310         'ee': 'ewe',
 311         'el': 'ell',
 312         'en': 'eng',
 313         'eo': 'epo',
 314         'es': 'spa',
 315         'et': 'est',
 316         'eu': 'eus',
 317         'fa': 'fas',
 318         'ff': 'ful',
 319         'fi': 'fin',
 320         'fj': 'fij',
 321         'fo': 'fao',
 322         'fr': 'fra',
 323         'fy': 'fry',
 324         'ga': 'gle',
 325         'gd': 'gla',
 326         'gl': 'glg',
 327         'gn': 'grn',
 328         'gu': 'guj',
 329         'gv': 'glv',
 330         'ha': 'hau',
 331         'he': 'heb',
 332         'hi': 'hin',
 333         'ho': 'hmo',
 334         'hr': 'hrv',
 335         'ht': 'hat',
 336         'hu': 'hun',
 337         'hy': 'hye',
 338         'hz': 'her',
 339         'ia': 'ina',
 340         'id': 'ind',
 341         'ie': 'ile',
 342         'ig': 'ibo',
 343         'ii': 'iii',
 344         'ik': 'ipk',
 345         'io': 'ido',
 346         'is': 'isl',
 347         'it': 'ita',
 348         'iu': 'iku',
 349         'ja': 'jpn',
 350         'jv': 'jav',
 351         'ka': 'kat',
 352         'kg': 'kon',
 353         'ki': 'kik',
 354         'kj': 'kua',
 355         'kk': 'kaz',
 356         'kl': 'kal',
 357         'km': 'khm',
 358         'kn': 'kan',
 359         'ko': 'kor',
 360         'kr': 'kau',
 361         'ks': 'kas',
 362         'ku': 'kur',
 363         'kv': 'kom',
 364         'kw': 'cor',
 365         'ky': 'kir',
 366         'la': 'lat',
 367         'lb': 'ltz',
 368         'lg': 'lug',
 369         'li': 'lim',
 370         'ln': 'lin',
 371         'lo': 'lao',
 372         'lt': 'lit',
 373         'lu': 'lub',
 374         'lv': 'lav',
 375         'mg': 'mlg',
 376         'mh': 'mah',
 377         'mi': 'mri',
 378         'mk': 'mkd',
 379         'ml': 'mal',
 380         'mn': 'mon',
 381         'mr': 'mar',
 382         'ms': 'msa',
 383         'mt': 'mlt',
 384         'my': 'mya',
 385         'na': 'nau',
 386         'nb': 'nob',
 387         'nd': 'nde',
 388         'ne': 'nep',
 389         'ng': 'ndo',
 390         'nl': 'nld',
 391         'nn': 'nno',
 392         'no': 'nor',
 393         'nr': 'nbl',
 394         'nv': 'nav',
 395         'ny': 'nya',
 396         'oc': 'oci',
 397         'oj': 'oji',
 398         'om': 'orm',
 399         'or': 'ori',
 400         'os': 'oss',
 401         'pa': 'pan',
 402         'pi': 'pli',
 403         'pl': 'pol',
 404         'ps': 'pus',
 405         'pt': 'por',
 406         'qu': 'que',
 407         'rm': 'roh',
 408         'rn': 'run',
 409         'ro': 'ron',
 410         'ru': 'rus',
 411         'rw': 'kin',
 412         'sa': 'san',
 413         'sc': 'srd',
 414         'sd': 'snd',
 415         'se': 'sme',
 416         'sg': 'sag',
 417         'si': 'sin',
 418         'sk': 'slk',
 419         'sl': 'slv',
 420         'sm': 'smo',
 421         'sn': 'sna',
 422         'so': 'som',
 423         'sq': 'sqi',
 424         'sr': 'srp',
 425         'ss': 'ssw',
 426         'st': 'sot',
 427         'su': 'sun',
 428         'sv': 'swe',
 429         'sw': 'swa',
 430         'ta': 'tam',
 431         'te': 'tel',
 432         'tg': 'tgk',
 433         'th': 'tha',
 434         'ti': 'tir',
 435         'tk': 'tuk',
 436         'tl': 'tgl',
 437         'tn': 'tsn',
 438         'to': 'ton',
 439         'tr': 'tur',
 440         'ts': 'tso',
 441         'tt': 'tat',
 442         'tw': 'twi',
 443         'ty': 'tah',
 444         'ug': 'uig',
 445         'uk': 'ukr',
 446         'ur': 'urd',
 447         'uz': 'uzb',
 448         've': 'ven',
 449         'vi': 'vie',
 450         'vo': 'vol',
 451         'wa': 'wln',
 452         'wo': 'wol',
 453         'xh': 'xho',
 454         'yi': 'yid',
 455         'yo': 'yor',
 456         'za': 'zha',
 457         'zh': 'zho',
 458         'zu': 'zul',
 459     }
 460
 461     def __init__(self, downloader=None, subtitlesformat='srt'):
 462         super(FFmpegEmbedSubtitlePP, self).__init__(downloader)
 463         self._subformat = subtitlesformat
 464
 465     @classmethod
 466     def _conver_lang_code(cls, code):
 467         """Convert language code from ISO 639-1 to ISO 639-2/T"""
 468         return cls._lang_map.get(code[:2])
 469
 470     def run(self, information):
 471         if information['ext'] != u'mp4':
 472             self._downloader.to_screen(u'[ffmpeg] Subtitles can only be embedded in mp4 files')
 473             return True, information
 474         if not information.get('subtitles'):
 475             self._downloader.to_screen(u'[ffmpeg] There aren\'t any subtitles to embed')
 476             return True, information
 477
 478         sub_langs = [key for key in information['subtitles']]
 479         filename = information['filepath']
 480         input_files = [filename] + [subtitles_filename(filename, lang, self._subformat) for lang in sub_langs]
 481
 482         opts = ['-map', '0:0', '-map', '0:1', '-c:v', 'copy', '-c:a', 'copy']
 483         for (i, lang) in enumerate(sub_langs):
 484             opts.extend(['-map', '%d:0' % (i+1), '-c:s:%d' % i, 'mov_text'])
 485             lang_code = self._conver_lang_code(lang)
 486             if lang_code is not None:
 487                 opts.extend(['-metadata:s:s:%d' % i, 'language=%s' % lang_code])
 488         opts.extend(['-f', 'mp4'])
 489
 490         temp_filename = filename + u'.temp'
 491         self._downloader.to_screen(u'[ffmpeg] Embedding subtitles in \'%s\'' % filename)
 492         self.run_ffmpeg_multiple_files(input_files, temp_filename, opts)
 493         os.remove(encodeFilename(filename))
 494         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 495
 496         return True, information
 497
 498
 499 class FFmpegMetadataPP(FFmpegPostProcessor):
 500     def run(self, info):
 501         metadata = {}
 502         if info.get('title') is not None:
 503             metadata['title'] = info['title']
 504         if info.get('upload_date') is not None:
 505             metadata['date'] = info['upload_date']
 506         if info.get('uploader') is not None:
 507             metadata['artist'] = info['uploader']
 508         elif info.get('uploader_id') is not None:
 509             metadata['artist'] = info['uploader_id']
 510
 511         if not metadata:
 512             self._downloader.to_screen(u'[ffmpeg] There isn\'t any metadata to add')
 513             return True, info
 514
 515         filename = info['filepath']
 516         temp_filename = prepend_extension(filename, 'temp')
 517
 518         if info['ext'] == u'm4a':
 519             options = ['-vn', '-acodec', 'copy']
 520         else:
 521             options = ['-c', 'copy']
 522
 523         for (name, value) in metadata.items():
 524             options.extend(['-metadata', '%s=%s' % (name, value)])
 525
 526         self._downloader.to_screen(u'[ffmpeg] Adding metadata to \'%s\'' % filename)
 527         self.run_ffmpeg(filename, temp_filename, options)
 528         os.remove(encodeFilename(filename))
 529         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 530         return True, info
 531
 532
 533 class FFmpegMergerPP(FFmpegPostProcessor):
 534     def run(self, info):
 535         filename = info['filepath']
 536         args = ['-c', 'copy', '-map', '0:v:0', '-map', '1:a:0', '-shortest']
 537         self._downloader.to_screen(u'[ffmpeg] Merging formats into "%s"' % filename)
 538         self.run_ffmpeg_multiple_files(info['__files_to_merge'], filename, args)
 539         return True, info
 540
 541
 542 class FFmpegAudioFixPP(FFmpegPostProcessor):
 543     def run(self, info):
 544         filename = info['filepath']
 545         temp_filename = prepend_extension(filename, 'temp')
 546
 547         options = ['-vn', '-acodec', 'copy']
 548         self._downloader.to_screen(u'[ffmpeg] Fixing audio file "%s"' % filename)
 549         self.run_ffmpeg(filename, temp_filename, options)
 550
 551         os.remove(encodeFilename(filename))
 552         os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 553
 554         return True, info