Add wav audio output
This commit is contained in:
		
							parent
							
								
									a5647b79ce
								
							
						
					
					
						commit
						633cf7cbad
					
				
					 1 changed files with 32 additions and 11 deletions
				
			
		| 
						 | 
					@ -3924,6 +3924,9 @@ class PostProcessor(object):
 | 
				
			||||||
		"""
 | 
							"""
 | 
				
			||||||
		return information # by default, do nothing
 | 
							return information # by default, do nothing
 | 
				
			||||||
 | 
					
 | 
				
			||||||
 | 
					class AudioConversionError(BaseException):
 | 
				
			||||||
 | 
						def __init__(self, message):
 | 
				
			||||||
 | 
							self.message = message
 | 
				
			||||||
 | 
					
 | 
				
			||||||
class FFmpegExtractAudioPP(PostProcessor):
 | 
					class FFmpegExtractAudioPP(PostProcessor):
 | 
				
			||||||
 | 
					
 | 
				
			||||||
| 
						 | 
					@ -3955,12 +3958,23 @@ class FFmpegExtractAudioPP(PostProcessor):
 | 
				
			||||||
 | 
					
 | 
				
			||||||
	@staticmethod
 | 
						@staticmethod
 | 
				
			||||||
	def run_ffmpeg(path, out_path, codec, more_opts):
 | 
						def run_ffmpeg(path, out_path, codec, more_opts):
 | 
				
			||||||
 | 
							if codec is None:
 | 
				
			||||||
 | 
								acodec_opts = []
 | 
				
			||||||
 | 
							else:
 | 
				
			||||||
 | 
								acodec_opts = ['-acodec', codec]
 | 
				
			||||||
 | 
							cmd = ['ffmpeg', '-y', '-i', path, '-vn'] + acodec_opts + more_opts + ['--', out_path]
 | 
				
			||||||
		try:
 | 
							try:
 | 
				
			||||||
			cmd = ['ffmpeg', '-y', '-i', path, '-vn', '-acodec', codec] + more_opts + ['--', out_path]
 | 
								p = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
 | 
				
			||||||
			ret = subprocess.call(cmd, stdout=file(os.path.devnull, 'w'), stderr=subprocess.STDOUT)
 | 
								stdout,stderr = p.communicate()
 | 
				
			||||||
			return (ret == 0)
 | 
					 | 
				
			||||||
		except (IOError, OSError):
 | 
							except (IOError, OSError):
 | 
				
			||||||
			return False
 | 
								e = sys.exc_info()[1]
 | 
				
			||||||
 | 
								if isinstance(e, OSError) and e.errno == 2:
 | 
				
			||||||
 | 
									raise AudioConversionError('ffmpeg not found. Please install ffmpeg.')
 | 
				
			||||||
 | 
								else:
 | 
				
			||||||
 | 
									raise e
 | 
				
			||||||
 | 
							if p.returncode != 0:
 | 
				
			||||||
 | 
								msg = stderr.strip().split('\n')[-1]
 | 
				
			||||||
 | 
								raise AudioConversionError(msg)
 | 
				
			||||||
 | 
					
 | 
				
			||||||
	def run(self, information):
 | 
						def run(self, information):
 | 
				
			||||||
		path = information['filepath']
 | 
							path = information['filepath']
 | 
				
			||||||
| 
						 | 
					@ -3994,7 +4008,7 @@ class FFmpegExtractAudioPP(PostProcessor):
 | 
				
			||||||
					more_opts += ['-ab', self._preferredquality]
 | 
										more_opts += ['-ab', self._preferredquality]
 | 
				
			||||||
		else:
 | 
							else:
 | 
				
			||||||
			# We convert the audio (lossy)
 | 
								# We convert the audio (lossy)
 | 
				
			||||||
			acodec = {'mp3': 'libmp3lame', 'aac': 'aac', 'm4a': 'aac', 'vorbis': 'libvorbis'}[self._preferredcodec]
 | 
								acodec = {'mp3': 'libmp3lame', 'aac': 'aac', 'm4a': 'aac', 'vorbis': 'libvorbis', 'wav': None}[self._preferredcodec]
 | 
				
			||||||
			extension = self._preferredcodec
 | 
								extension = self._preferredcodec
 | 
				
			||||||
			more_opts = []
 | 
								more_opts = []
 | 
				
			||||||
			if self._preferredquality is not None:
 | 
								if self._preferredquality is not None:
 | 
				
			||||||
| 
						 | 
					@ -4005,14 +4019,21 @@ class FFmpegExtractAudioPP(PostProcessor):
 | 
				
			||||||
				more_opts += ['-absf', 'aac_adtstoasc']
 | 
									more_opts += ['-absf', 'aac_adtstoasc']
 | 
				
			||||||
			if self._preferredcodec == 'vorbis':
 | 
								if self._preferredcodec == 'vorbis':
 | 
				
			||||||
				extension = 'ogg'
 | 
									extension = 'ogg'
 | 
				
			||||||
 | 
								if self._preferredcodec == 'wav':
 | 
				
			||||||
 | 
									extension = 'wav'
 | 
				
			||||||
 | 
									more_opts += ['-f', 'wav']
 | 
				
			||||||
 | 
					
 | 
				
			||||||
		(prefix, ext) = os.path.splitext(path)
 | 
							(prefix, ext) = os.path.splitext(path)
 | 
				
			||||||
		new_path = prefix + '.' + extension
 | 
							new_path = prefix + '.' + extension
 | 
				
			||||||
		self._downloader.to_screen(u'[ffmpeg] Destination: %s' % new_path)
 | 
							self._downloader.to_screen(u'[ffmpeg] Destination: %s' % new_path)
 | 
				
			||||||
		status = self.run_ffmpeg(path, new_path, acodec, more_opts)
 | 
							try:
 | 
				
			||||||
 | 
								self.run_ffmpeg(path, new_path, acodec, more_opts)
 | 
				
			||||||
		if not status:
 | 
							except:
 | 
				
			||||||
			self._downloader.to_stderr(u'WARNING: error running ffmpeg')
 | 
								etype,e,tb = sys.exc_info()
 | 
				
			||||||
 | 
								if isinstance(e, AudioConversionError):
 | 
				
			||||||
 | 
									self._downloader.to_stderr(u'ERROR: audio conversion failed: ' + e.message)
 | 
				
			||||||
 | 
								else:
 | 
				
			||||||
 | 
									self._downloader.to_stderr(u'ERROR: error running ffmpeg')
 | 
				
			||||||
			return None
 | 
								return None
 | 
				
			||||||
 | 
					
 | 
				
			||||||
 		# Try to update the date time for extracted audio file.
 | 
					 		# Try to update the date time for extracted audio file.
 | 
				
			||||||
| 
						 | 
					@ -4251,7 +4272,7 @@ def parseOpts():
 | 
				
			||||||
	postproc.add_option('--extract-audio', action='store_true', dest='extractaudio', default=False,
 | 
						postproc.add_option('--extract-audio', action='store_true', dest='extractaudio', default=False,
 | 
				
			||||||
			help='convert video files to audio-only files (requires ffmpeg and ffprobe)')
 | 
								help='convert video files to audio-only files (requires ffmpeg and ffprobe)')
 | 
				
			||||||
	postproc.add_option('--audio-format', metavar='FORMAT', dest='audioformat', default='best',
 | 
						postproc.add_option('--audio-format', metavar='FORMAT', dest='audioformat', default='best',
 | 
				
			||||||
			help='"best", "aac", "vorbis", "mp3", or "m4a"; best by default')
 | 
								help='"best", "aac", "vorbis", "mp3", "m4a", or "wav"; best by default')
 | 
				
			||||||
	postproc.add_option('--audio-quality', metavar='QUALITY', dest='audioquality', default='128K',
 | 
						postproc.add_option('--audio-quality', metavar='QUALITY', dest='audioquality', default='128K',
 | 
				
			||||||
			help='ffmpeg audio bitrate specification, 128k by default')
 | 
								help='ffmpeg audio bitrate specification, 128k by default')
 | 
				
			||||||
	postproc.add_option('-k', '--keep-video', action='store_true', dest='keepvideo', default=False,
 | 
						postproc.add_option('-k', '--keep-video', action='store_true', dest='keepvideo', default=False,
 | 
				
			||||||
| 
						 | 
					@ -4397,7 +4418,7 @@ def _real_main():
 | 
				
			||||||
	except (TypeError, ValueError), err:
 | 
						except (TypeError, ValueError), err:
 | 
				
			||||||
		parser.error(u'invalid playlist end number specified')
 | 
							parser.error(u'invalid playlist end number specified')
 | 
				
			||||||
	if opts.extractaudio:
 | 
						if opts.extractaudio:
 | 
				
			||||||
		if opts.audioformat not in ['best', 'aac', 'mp3', 'vorbis', 'm4a']:
 | 
							if opts.audioformat not in ['best', 'aac', 'mp3', 'vorbis', 'm4a', 'wav']:
 | 
				
			||||||
			parser.error(u'invalid audio format specified')
 | 
								parser.error(u'invalid audio format specified')
 | 
				
			||||||
 | 
					
 | 
				
			||||||
	# File downloader
 | 
						# File downloader
 | 
				
			||||||
| 
						 | 
					
 | 
				
			||||||
		Loading…
	
	Add table
		Add a link
		
	
		Reference in a new issue