1 from __future__
import unicode_literals
10 from .common
import AudioConversionError
, PostProcessor
12 from ..compat
import (
13 compat_subprocess_get_DEVNULL
,
30 EXT_TO_OUT_FORMATS
= {
48 'vorbis': 'libvorbis',
53 class FFmpegPostProcessorError(PostProcessingError
):
57 class FFmpegPostProcessor(PostProcessor
):
58 def __init__(self
, downloader
=None):
59 PostProcessor
.__init
__(self
, downloader
)
60 self
._determine
_executables
()
62 def check_version(self
):
63 if not self
.available
:
64 raise FFmpegPostProcessorError('ffmpeg or avconv not found. Please install one.')
66 required_version
= '10-0' if self
.basename
== 'avconv' else '1.0'
67 if is_outdated_version(
68 self
._versions
[self
.basename
], required_version
):
69 warning
= 'Your copy of %s is outdated, update %s to version %s or newer if you encounter any errors.' % (
70 self
.basename
, self
.basename
, required_version
)
72 self
._downloader
.report_warning(warning
)
75 def get_versions(downloader
=None):
76 return FFmpegPostProcessor(downloader
)._versions
78 def _determine_executables(self
):
79 programs
= ['avprobe', 'avconv', 'ffmpeg', 'ffprobe']
82 def get_ffmpeg_version(path
):
83 ver
= get_exe_version(path
, args
=['-version'])
86 r
'(?:\d+:)?([0-9.]+)-[0-9]+ubuntu[0-9.]+$', # Ubuntu, see [1]
87 r
'n([0-9.]+)$', # Arch Linux
88 # 1. http://www.ducea.com/2006/06/17/ubuntu-package-version-naming-explanation/
91 mobj
= re
.match(regex
, ver
)
97 self
.probe_basename
= None
100 self
._versions
= None
102 prefer_ffmpeg
= self
._downloader
.params
.get('prefer_ffmpeg', True)
103 location
= self
._downloader
.params
.get('ffmpeg_location')
104 if location
is not None:
105 if not os
.path
.exists(location
):
106 self
._downloader
.report_warning(
107 'ffmpeg-location %s does not exist! '
108 'Continuing without avconv/ffmpeg.' % (location
))
111 elif not os
.path
.isdir(location
):
112 basename
= os
.path
.splitext(os
.path
.basename(location
))[0]
113 if basename
not in programs
:
114 self
._downloader
.report_warning(
115 'Cannot identify executable %s, its basename should be one of %s. '
116 'Continuing without avconv/ffmpeg.' %
117 (location
, ', '.join(programs
)))
120 location
= os
.path
.dirname(os
.path
.abspath(location
))
121 if basename
in ('ffmpeg', 'ffprobe'):
125 (p
, os
.path
.join(location
, p
)) for p
in programs
)
126 self
._versions
= dict(
127 (p
, get_ffmpeg_version(self
._paths
[p
])) for p
in programs
)
128 if self
._versions
is None:
129 self
._versions
= dict(
130 (p
, get_ffmpeg_version(p
)) for p
in programs
)
131 self
._paths
= dict((p
, p
) for p
in programs
)
133 if prefer_ffmpeg
is False:
134 prefs
= ('avconv', 'ffmpeg')
136 prefs
= ('ffmpeg', 'avconv')
138 if self
._versions
[p
]:
142 if prefer_ffmpeg
is False:
143 prefs
= ('avprobe', 'ffprobe')
145 prefs
= ('ffprobe', 'avprobe')
147 if self
._versions
[p
]:
148 self
.probe_basename
= p
153 return self
.basename
is not None
156 def executable(self
):
157 return self
._paths
[self
.basename
]
160 def probe_available(self
):
161 return self
.probe_basename
is not None
164 def probe_executable(self
):
165 return self
._paths
[self
.probe_basename
]
167 def get_audio_codec(self
, path
):
168 if not self
.probe_available
:
169 raise PostProcessingError('ffprobe or avprobe not found. Please install one.')
172 encodeFilename(self
.probe_executable
, True),
173 encodeArgument('-show_streams'),
174 encodeFilename(self
._ffmpeg
_filename
_argument
(path
), True)]
175 if self
._downloader
.params
.get('verbose', False):
176 self
._downloader
.to_screen('[debug] %s command line: %s' % (self
.basename
, shell_quote(cmd
)))
177 handle
= subprocess
.Popen(cmd
, stderr
=compat_subprocess_get_DEVNULL(), stdout
=subprocess
.PIPE
, stdin
=subprocess
.PIPE
)
178 output
= handle
.communicate()[0]
179 if handle
.wait() != 0:
181 except (IOError, OSError):
184 for line
in output
.decode('ascii', 'ignore').split('\n'):
185 if line
.startswith('codec_name='):
186 audio_codec
= line
.split('=')[1].strip()
187 elif line
.strip() == 'codec_type=audio' and audio_codec
is not None:
191 def run_ffmpeg_multiple_files(self
, input_paths
, out_path
, opts
):
195 os
.stat(encodeFilename(path
)).st_mtime
for path
in input_paths
)
197 opts
+= self
._configuration
_args
()
200 for path
in input_paths
:
202 encodeArgument('-i'),
203 encodeFilename(self
._ffmpeg
_filename
_argument
(path
), True)
205 cmd
= ([encodeFilename(self
.executable
, True), encodeArgument('-y')] +
207 [encodeArgument(o
) for o
in opts
] +
208 [encodeFilename(self
._ffmpeg
_filename
_argument
(out_path
), True)])
210 if self
._downloader
.params
.get('verbose', False):
211 self
._downloader
.to_screen('[debug] ffmpeg command line: %s' % shell_quote(cmd
))
212 p
= subprocess
.Popen(cmd
, stdout
=subprocess
.PIPE
, stderr
=subprocess
.PIPE
, stdin
=subprocess
.PIPE
)
213 stdout
, stderr
= p
.communicate()
214 if p
.returncode
!= 0:
215 stderr
= stderr
.decode('utf-8', 'replace')
216 msg
= stderr
.strip().split('\n')[-1]
217 raise FFmpegPostProcessorError(msg
)
218 self
.try_utime(out_path
, oldest_mtime
, oldest_mtime
)
220 def run_ffmpeg(self
, path
, out_path
, opts
):
221 self
.run_ffmpeg_multiple_files([path
], out_path
, opts
)
223 def _ffmpeg_filename_argument(self
, fn
):
224 # Always use 'file:' because the filename may contain ':' (ffmpeg
225 # interprets that as a protocol) or can start with '-' (-- is broken in
226 # ffmpeg, see https://ffmpeg.org/trac/ffmpeg/ticket/2127 for details)
227 # Also leave '-' intact in order not to break streaming to stdout.
228 return 'file:' + fn
if fn
!= '-' else fn
231 class FFmpegExtractAudioPP(FFmpegPostProcessor
):
232 def __init__(self
, downloader
=None, preferredcodec
=None, preferredquality
=None, nopostoverwrites
=False):
233 FFmpegPostProcessor
.__init
__(self
, downloader
)
234 if preferredcodec
is None:
235 preferredcodec
= 'best'
236 self
._preferredcodec
= preferredcodec
237 self
._preferredquality
= preferredquality
238 self
._nopostoverwrites
= nopostoverwrites
240 def run_ffmpeg(self
, path
, out_path
, codec
, more_opts
):
244 acodec_opts
= ['-acodec', codec
]
245 opts
= ['-vn'] + acodec_opts
+ more_opts
247 FFmpegPostProcessor
.run_ffmpeg(self
, path
, out_path
, opts
)
248 except FFmpegPostProcessorError
as err
:
249 raise AudioConversionError(err
.msg
)
251 def run(self
, information
):
252 path
= information
['filepath']
254 filecodec
= self
.get_audio_codec(path
)
255 if filecodec
is None:
256 raise PostProcessingError('WARNING: unable to obtain file audio codec with ffprobe')
259 if self
._preferredcodec
== 'best' or self
._preferredcodec
== filecodec
or (self
._preferredcodec
== 'm4a' and filecodec
== 'aac'):
260 if filecodec
== 'aac' and self
._preferredcodec
in ['m4a', 'best']:
261 # Lossless, but in another container
264 more_opts
= ['-bsf:a', 'aac_adtstoasc']
265 elif filecodec
in ['aac', 'flac', 'mp3', 'vorbis', 'opus']:
266 # Lossless if possible
268 extension
= filecodec
269 if filecodec
== 'aac':
270 more_opts
= ['-f', 'adts']
271 if filecodec
== 'vorbis':
275 acodec
= 'libmp3lame'
278 if self
._preferredquality
is not None:
279 if int(self
._preferredquality
) < 10:
280 more_opts
+= ['-q:a', self
._preferredquality
]
282 more_opts
+= ['-b:a', self
._preferredquality
+ 'k']
284 # We convert the audio (lossy if codec is lossy)
285 acodec
= ACODECS
[self
._preferredcodec
]
286 extension
= self
._preferredcodec
288 if self
._preferredquality
is not None:
289 # The opus codec doesn't support the -aq option
290 if int(self
._preferredquality
) < 10 and extension
!= 'opus':
291 more_opts
+= ['-q:a', self
._preferredquality
]
293 more_opts
+= ['-b:a', self
._preferredquality
+ 'k']
294 if self
._preferredcodec
== 'aac':
295 more_opts
+= ['-f', 'adts']
296 if self
._preferredcodec
== 'm4a':
297 more_opts
+= ['-bsf:a', 'aac_adtstoasc']
298 if self
._preferredcodec
== 'vorbis':
300 if self
._preferredcodec
== 'wav':
302 more_opts
+= ['-f', 'wav']
304 prefix
, sep
, ext
= path
.rpartition('.') # not os.path.splitext, since the latter does not work on unicode in all setups
305 new_path
= prefix
+ sep
+ extension
307 information
['filepath'] = new_path
308 information
['ext'] = extension
310 # If we download foo.mp3 and convert it to... foo.mp3, then don't delete foo.mp3, silly.
311 if (new_path
== path
or
312 (self
._nopostoverwrites
and os
.path
.exists(encodeFilename(new_path
)))):
313 self
._downloader
.to_screen('[ffmpeg] Post-process file %s exists, skipping' % new_path
)
314 return [], information
317 self
._downloader
.to_screen('[ffmpeg] Destination: ' + new_path
)
318 self
.run_ffmpeg(path
, new_path
, acodec
, more_opts
)
319 except AudioConversionError
as e
:
320 raise PostProcessingError(
321 'audio conversion failed: ' + e
.msg
)
323 raise PostProcessingError('error running ' + self
.basename
)
325 # Try to update the date time for extracted audio file.
326 if information
.get('filetime') is not None:
328 new_path
, time
.time(), information
['filetime'],
329 errnote
='Cannot update utime of audio file')
331 return [path
], information
334 class FFmpegVideoConvertorPP(FFmpegPostProcessor
):
335 def __init__(self
, downloader
=None, preferedformat
=None):
336 super(FFmpegVideoConvertorPP
, self
).__init
__(downloader
)
337 self
._preferedformat
= preferedformat
339 def run(self
, information
):
340 path
= information
['filepath']
341 if information
['ext'] == self
._preferedformat
:
342 self
._downloader
.to_screen('[ffmpeg] Not converting video file %s - already is in target format %s' % (path
, self
._preferedformat
))
343 return [], information
345 if self
._preferedformat
== 'avi':
346 options
.extend(['-c:v', 'libxvid', '-vtag', 'XVID'])
347 prefix
, sep
, ext
= path
.rpartition('.')
348 outpath
= prefix
+ sep
+ self
._preferedformat
349 self
._downloader
.to_screen('[' + 'ffmpeg' + '] Converting video from %s to %s, Destination: ' % (information
['ext'], self
._preferedformat
) + outpath
)
350 self
.run_ffmpeg(path
, outpath
, options
)
351 information
['filepath'] = outpath
352 information
['format'] = self
._preferedformat
353 information
['ext'] = self
._preferedformat
354 return [path
], information
357 class FFmpegEmbedSubtitlePP(FFmpegPostProcessor
):
358 def run(self
, information
):
359 if information
['ext'] not in ('mp4', 'webm', 'mkv'):
360 self
._downloader
.to_screen('[ffmpeg] Subtitles can only be embedded in mp4, webm or mkv files')
361 return [], information
362 subtitles
= information
.get('requested_subtitles')
364 self
._downloader
.to_screen('[ffmpeg] There aren\'t any subtitles to embed')
365 return [], information
367 filename
= information
['filepath']
369 ext
= information
['ext']
372 webm_vtt_warn
= False
374 for lang
, sub_info
in subtitles
.items():
375 sub_ext
= sub_info
['ext']
376 if ext
!= 'webm' or ext
== 'webm' and sub_ext
== 'vtt':
377 sub_langs
.append(lang
)
378 sub_filenames
.append(subtitles_filename(filename
, lang
, sub_ext
))
380 if not webm_vtt_warn
and ext
== 'webm' and sub_ext
!= 'vtt':
382 self
._downloader
.to_screen('[ffmpeg] Only WebVTT subtitles can be embedded in webm files')
385 return [], information
387 input_files
= [filename
] + sub_filenames
392 # Don't copy the existing subtitles, we may be running the
393 # postprocessor a second time
396 if information
['ext'] == 'mp4':
397 opts
+= ['-c:s', 'mov_text']
398 for (i
, lang
) in enumerate(sub_langs
):
399 opts
.extend(['-map', '%d:0' % (i
+ 1)])
400 lang_code
= ISO639Utils
.short2long(lang
) or lang
401 opts
.extend(['-metadata:s:s:%d' % i
, 'language=%s' % lang_code
])
403 temp_filename
= prepend_extension(filename
, 'temp')
404 self
._downloader
.to_screen('[ffmpeg] Embedding subtitles in \'%s\'' % filename
)
405 self
.run_ffmpeg_multiple_files(input_files
, temp_filename
, opts
)
406 os
.remove(encodeFilename(filename
))
407 os
.rename(encodeFilename(temp_filename
), encodeFilename(filename
))
409 return sub_filenames
, information
412 class FFmpegMetadataPP(FFmpegPostProcessor
):
416 def add(meta_list
, info_list
=None):
418 info_list
= meta_list
419 if not isinstance(meta_list
, (list, tuple)):
420 meta_list
= (meta_list
,)
421 if not isinstance(info_list
, (list, tuple)):
422 info_list
= (info_list
,)
423 for info_f
in info_list
:
424 if info
.get(info_f
) is not None:
425 for meta_f
in meta_list
:
426 metadata
[meta_f
] = info
[info_f
]
429 add('title', ('track', 'title'))
430 add('date', 'upload_date')
431 add(('description', 'comment'), 'description')
432 add('purl', 'webpage_url')
433 add('track', 'track_number')
434 add('artist', ('artist', 'creator', 'uploader', 'uploader_id'))
438 add('disc', 'disc_number')
441 self
._downloader
.to_screen('[ffmpeg] There isn\'t any metadata to add')
444 filename
= info
['filepath']
445 temp_filename
= prepend_extension(filename
, 'temp')
446 in_filenames
= [filename
]
449 if info
['ext'] == 'm4a':
450 options
.extend(['-vn', '-acodec', 'copy'])
452 options
.extend(['-c', 'copy'])
454 for (name
, value
) in metadata
.items():
455 options
.extend(['-metadata', '%s=%s' % (name
, value
)])
457 chapters
= info
.get('chapters', [])
459 metadata_filename
= replace_extension(filename
, 'meta')
460 with io
.open(metadata_filename
, 'wt', encoding
='utf-8') as f
:
461 def ffmpeg_escape(text
):
462 return re
.sub(r
'(=|;|#|\\|\n)', r
'\\\1', text
)
464 metadata_file_content
= ';FFMETADATA1\n'
465 for chapter
in chapters
:
466 metadata_file_content
+= '[CHAPTER]\nTIMEBASE=1/1000\n'
467 metadata_file_content
+= 'START=%d\n' % (chapter
['start_time'] * 1000)
468 metadata_file_content
+= 'END=%d\n' % (chapter
['end_time'] * 1000)
469 chapter_title
= chapter
.get('title')
471 metadata_file_content
+= 'title=%s\n' % ffmpeg_escape(chapter_title
)
472 f
.write(metadata_file_content
)
473 in_filenames
.append(metadata_filename
)
474 options
.extend(['-map_metadata', '1'])
476 self
._downloader
.to_screen('[ffmpeg] Adding metadata to \'%s\'' % filename
)
477 self
.run_ffmpeg_multiple_files(in_filenames
, temp_filename
, options
)
479 os
.remove(metadata_filename
)
480 os
.remove(encodeFilename(filename
))
481 os
.rename(encodeFilename(temp_filename
), encodeFilename(filename
))
485 class FFmpegMergerPP(FFmpegPostProcessor
):
487 filename
= info
['filepath']
488 temp_filename
= prepend_extension(filename
, 'temp')
489 args
= ['-c', 'copy', '-map', '0:v:0', '-map', '1:a:0']
490 self
._downloader
.to_screen('[ffmpeg] Merging formats into "%s"' % filename
)
491 self
.run_ffmpeg_multiple_files(info
['__files_to_merge'], temp_filename
, args
)
492 os
.rename(encodeFilename(temp_filename
), encodeFilename(filename
))
493 return info
['__files_to_merge'], info
496 # TODO: figure out merge-capable ffmpeg version
497 if self
.basename
!= 'avconv':
500 required_version
= '10-0'
501 if is_outdated_version(
502 self
._versions
[self
.basename
], required_version
):
503 warning
= ('Your copy of %s is outdated and unable to properly mux separate video and audio files, '
504 'youtube-dl will download single file media. '
505 'Update %s to version %s or newer to fix this.') % (
506 self
.basename
, self
.basename
, required_version
)
508 self
._downloader
.report_warning(warning
)
513 class FFmpegFixupStretchedPP(FFmpegPostProcessor
):
515 stretched_ratio
= info
.get('stretched_ratio')
516 if stretched_ratio
is None or stretched_ratio
== 1:
519 filename
= info
['filepath']
520 temp_filename
= prepend_extension(filename
, 'temp')
522 options
= ['-c', 'copy', '-aspect', '%f' % stretched_ratio
]
523 self
._downloader
.to_screen('[ffmpeg] Fixing aspect ratio in "%s"' % filename
)
524 self
.run_ffmpeg(filename
, temp_filename
, options
)
526 os
.remove(encodeFilename(filename
))
527 os
.rename(encodeFilename(temp_filename
), encodeFilename(filename
))
532 class FFmpegFixupM4aPP(FFmpegPostProcessor
):
534 if info
.get('container') != 'm4a_dash':
537 filename
= info
['filepath']
538 temp_filename
= prepend_extension(filename
, 'temp')
540 options
= ['-c', 'copy', '-f', 'mp4']
541 self
._downloader
.to_screen('[ffmpeg] Correcting container in "%s"' % filename
)
542 self
.run_ffmpeg(filename
, temp_filename
, options
)
544 os
.remove(encodeFilename(filename
))
545 os
.rename(encodeFilename(temp_filename
), encodeFilename(filename
))
550 class FFmpegFixupM3u8PP(FFmpegPostProcessor
):
552 filename
= info
['filepath']
553 if self
.get_audio_codec(filename
) == 'aac':
554 temp_filename
= prepend_extension(filename
, 'temp')
556 options
= ['-c', 'copy', '-f', 'mp4', '-bsf:a', 'aac_adtstoasc']
557 self
._downloader
.to_screen('[ffmpeg] Fixing malformed AAC bitstream in "%s"' % filename
)
558 self
.run_ffmpeg(filename
, temp_filename
, options
)
560 os
.remove(encodeFilename(filename
))
561 os
.rename(encodeFilename(temp_filename
), encodeFilename(filename
))
565 class FFmpegSubtitlesConvertorPP(FFmpegPostProcessor
):
566 def __init__(self
, downloader
=None, format
=None):
567 super(FFmpegSubtitlesConvertorPP
, self
).__init
__(downloader
)
571 subs
= info
.get('requested_subtitles')
572 filename
= info
['filepath']
573 new_ext
= self
.format
575 if new_format
== 'vtt':
576 new_format
= 'webvtt'
578 self
._downloader
.to_screen('[ffmpeg] There aren\'t any subtitles to convert')
580 self
._downloader
.to_screen('[ffmpeg] Converting subtitles')
582 for lang
, sub
in subs
.items():
585 self
._downloader
.to_screen(
586 '[ffmpeg] Subtitle file for %s is already in the requested format' % new_ext
)
588 old_file
= subtitles_filename(filename
, lang
, ext
)
589 sub_filenames
.append(old_file
)
590 new_file
= subtitles_filename(filename
, lang
, new_ext
)
592 if ext
in ('dfxp', 'ttml', 'tt'):
593 self
._downloader
.report_warning(
594 'You have requested to convert dfxp (TTML) subtitles into another format, '
595 'which results in style information loss')
598 srt_file
= subtitles_filename(filename
, lang
, 'srt')
600 with open(dfxp_file
, 'rb') as f
:
601 srt_data
= dfxp2srt(f
.read())
603 with io
.open(srt_file
, 'wt', encoding
='utf-8') as f
:
615 sub_filenames
.append(srt_file
)
617 self
.run_ffmpeg(old_file
, new_file
, ['-f', new_format
])
619 with io
.open(new_file
, 'rt', encoding
='utf-8') as f
:
625 return sub_filenames
, info