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