]> Raphaƫl G. Git Repositories - youtubedl/blob - youtube_dl/postprocessor/ffmpeg.py
debian/control: Update list of extractors/supported sites.
[youtubedl] / youtube_dl / postprocessor / ffmpeg.py
1 from __future__ import unicode_literals
2
3 import io
4 import os
5 import subprocess
6 import time
7 import re
8
9
10 from .common import AudioConversionError, PostProcessor
11
12 from ..utils import (
13 encodeArgument,
14 encodeFilename,
15 get_exe_version,
16 is_outdated_version,
17 PostProcessingError,
18 prepend_extension,
19 shell_quote,
20 subtitles_filename,
21 dfxp2srt,
22 ISO639Utils,
23 replace_extension,
24 )
25
26
27 EXT_TO_OUT_FORMATS = {
28 'aac': 'adts',
29 'flac': 'flac',
30 'm4a': 'ipod',
31 'mka': 'matroska',
32 'mkv': 'matroska',
33 'mpg': 'mpeg',
34 'ogv': 'ogg',
35 'ts': 'mpegts',
36 'wma': 'asf',
37 'wmv': 'asf',
38 }
39 ACODECS = {
40 'mp3': 'libmp3lame',
41 'aac': 'aac',
42 'flac': 'flac',
43 'm4a': 'aac',
44 'opus': 'libopus',
45 'vorbis': 'libvorbis',
46 'wav': None,
47 }
48
49
50 class FFmpegPostProcessorError(PostProcessingError):
51 pass
52
53
54 class FFmpegPostProcessor(PostProcessor):
55 def __init__(self, downloader=None):
56 PostProcessor.__init__(self, downloader)
57 self._determine_executables()
58
59 def check_version(self):
60 if not self.available:
61 raise FFmpegPostProcessorError('ffmpeg or avconv not found. Please install one.')
62
63 required_version = '10-0' if self.basename == 'avconv' else '1.0'
64 if is_outdated_version(
65 self._versions[self.basename], required_version):
66 warning = 'Your copy of %s is outdated, update %s to version %s or newer if you encounter any errors.' % (
67 self.basename, self.basename, required_version)
68 if self._downloader:
69 self._downloader.report_warning(warning)
70
71 @staticmethod
72 def get_versions(downloader=None):
73 return FFmpegPostProcessor(downloader)._versions
74
75 def _determine_executables(self):
76 programs = ['avprobe', 'avconv', 'ffmpeg', 'ffprobe']
77 prefer_ffmpeg = True
78
79 def get_ffmpeg_version(path):
80 ver = get_exe_version(path, args=['-version'])
81 if ver:
82 regexs = [
83 r'(?:\d+:)?([0-9.]+)-[0-9]+ubuntu[0-9.]+$', # Ubuntu, see [1]
84 r'n([0-9.]+)$', # Arch Linux
85 # 1. http://www.ducea.com/2006/06/17/ubuntu-package-version-naming-explanation/
86 ]
87 for regex in regexs:
88 mobj = re.match(regex, ver)
89 if mobj:
90 ver = mobj.group(1)
91 return ver
92
93 self.basename = None
94 self.probe_basename = None
95
96 self._paths = None
97 self._versions = None
98 if self._downloader:
99 prefer_ffmpeg = self._downloader.params.get('prefer_ffmpeg', True)
100 location = self._downloader.params.get('ffmpeg_location')
101 if location is not None:
102 if not os.path.exists(location):
103 self._downloader.report_warning(
104 'ffmpeg-location %s does not exist! '
105 'Continuing without avconv/ffmpeg.' % (location))
106 self._versions = {}
107 return
108 elif not os.path.isdir(location):
109 basename = os.path.splitext(os.path.basename(location))[0]
110 if basename not in programs:
111 self._downloader.report_warning(
112 'Cannot identify executable %s, its basename should be one of %s. '
113 'Continuing without avconv/ffmpeg.' %
114 (location, ', '.join(programs)))
115 self._versions = {}
116 return None
117 location = os.path.dirname(os.path.abspath(location))
118 if basename in ('ffmpeg', 'ffprobe'):
119 prefer_ffmpeg = True
120
121 self._paths = dict(
122 (p, os.path.join(location, p)) for p in programs)
123 self._versions = dict(
124 (p, get_ffmpeg_version(self._paths[p])) for p in programs)
125 if self._versions is None:
126 self._versions = dict(
127 (p, get_ffmpeg_version(p)) for p in programs)
128 self._paths = dict((p, p) for p in programs)
129
130 if prefer_ffmpeg is False:
131 prefs = ('avconv', 'ffmpeg')
132 else:
133 prefs = ('ffmpeg', 'avconv')
134 for p in prefs:
135 if self._versions[p]:
136 self.basename = p
137 break
138
139 if prefer_ffmpeg is False:
140 prefs = ('avprobe', 'ffprobe')
141 else:
142 prefs = ('ffprobe', 'avprobe')
143 for p in prefs:
144 if self._versions[p]:
145 self.probe_basename = p
146 break
147
148 @property
149 def available(self):
150 return self.basename is not None
151
152 @property
153 def executable(self):
154 return self._paths[self.basename]
155
156 @property
157 def probe_available(self):
158 return self.probe_basename is not None
159
160 @property
161 def probe_executable(self):
162 return self._paths[self.probe_basename]
163
164 def get_audio_codec(self, path):
165 if not self.probe_available and not self.available:
166 raise PostProcessingError('ffprobe/avprobe and ffmpeg/avconv not found. Please install one.')
167 try:
168 if self.probe_available:
169 cmd = [
170 encodeFilename(self.probe_executable, True),
171 encodeArgument('-show_streams')]
172 else:
173 cmd = [
174 encodeFilename(self.executable, True),
175 encodeArgument('-i')]
176 cmd.append(encodeFilename(self._ffmpeg_filename_argument(path), True))
177 if self._downloader.params.get('verbose', False):
178 self._downloader.to_screen(
179 '[debug] %s command line: %s' % (self.basename, shell_quote(cmd)))
180 handle = subprocess.Popen(
181 cmd, stderr=subprocess.PIPE,
182 stdout=subprocess.PIPE, stdin=subprocess.PIPE)
183 stdout_data, stderr_data = handle.communicate()
184 expected_ret = 0 if self.probe_available else 1
185 if handle.wait() != expected_ret:
186 return None
187 except (IOError, OSError):
188 return None
189 output = (stdout_data if self.probe_available else stderr_data).decode('ascii', 'ignore')
190 if self.probe_available:
191 audio_codec = None
192 for line in output.split('\n'):
193 if line.startswith('codec_name='):
194 audio_codec = line.split('=')[1].strip()
195 elif line.strip() == 'codec_type=audio' and audio_codec is not None:
196 return audio_codec
197 else:
198 # Stream #FILE_INDEX:STREAM_INDEX[STREAM_ID](LANGUAGE): CODEC_TYPE: CODEC_NAME
199 mobj = re.search(
200 r'Stream\s*#\d+:\d+(?:\[0x[0-9a-f]+\])?(?:\([a-z]{3}\))?:\s*Audio:\s*([0-9a-z]+)',
201 output)
202 if mobj:
203 return mobj.group(1)
204 return None
205
206 def run_ffmpeg_multiple_files(self, input_paths, out_path, opts):
207 self.check_version()
208
209 oldest_mtime = min(
210 os.stat(encodeFilename(path)).st_mtime for path in input_paths)
211
212 opts += self._configuration_args()
213
214 files_cmd = []
215 for path in input_paths:
216 files_cmd.extend([
217 encodeArgument('-i'),
218 encodeFilename(self._ffmpeg_filename_argument(path), True)
219 ])
220 cmd = [encodeFilename(self.executable, True), encodeArgument('-y')]
221 # avconv does not have repeat option
222 if self.basename == 'ffmpeg':
223 cmd += [encodeArgument('-loglevel'), encodeArgument('repeat+info')]
224 cmd += (files_cmd
225 + [encodeArgument(o) for o in opts]
226 + [encodeFilename(self._ffmpeg_filename_argument(out_path), True)])
227
228 if self._downloader.params.get('verbose', False):
229 self._downloader.to_screen('[debug] ffmpeg command line: %s' % shell_quote(cmd))
230 p = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE, stdin=subprocess.PIPE)
231 stdout, stderr = p.communicate()
232 if p.returncode != 0:
233 stderr = stderr.decode('utf-8', 'replace')
234 msg = stderr.strip().split('\n')[-1]
235 raise FFmpegPostProcessorError(msg)
236 self.try_utime(out_path, oldest_mtime, oldest_mtime)
237
238 def run_ffmpeg(self, path, out_path, opts):
239 self.run_ffmpeg_multiple_files([path], out_path, opts)
240
241 def _ffmpeg_filename_argument(self, fn):
242 # Always use 'file:' because the filename may contain ':' (ffmpeg
243 # interprets that as a protocol) or can start with '-' (-- is broken in
244 # ffmpeg, see https://ffmpeg.org/trac/ffmpeg/ticket/2127 for details)
245 # Also leave '-' intact in order not to break streaming to stdout.
246 return 'file:' + fn if fn != '-' else fn
247
248
249 class FFmpegExtractAudioPP(FFmpegPostProcessor):
250 def __init__(self, downloader=None, preferredcodec=None, preferredquality=None, nopostoverwrites=False):
251 FFmpegPostProcessor.__init__(self, downloader)
252 if preferredcodec is None:
253 preferredcodec = 'best'
254 self._preferredcodec = preferredcodec
255 self._preferredquality = preferredquality
256 self._nopostoverwrites = nopostoverwrites
257
258 def run_ffmpeg(self, path, out_path, codec, more_opts):
259 if codec is None:
260 acodec_opts = []
261 else:
262 acodec_opts = ['-acodec', codec]
263 opts = ['-vn'] + acodec_opts + more_opts
264 try:
265 FFmpegPostProcessor.run_ffmpeg(self, path, out_path, opts)
266 except FFmpegPostProcessorError as err:
267 raise AudioConversionError(err.msg)
268
269 def run(self, information):
270 path = information['filepath']
271
272 filecodec = self.get_audio_codec(path)
273 if filecodec is None:
274 raise PostProcessingError('WARNING: unable to obtain file audio codec with ffprobe')
275
276 more_opts = []
277 if self._preferredcodec == 'best' or self._preferredcodec == filecodec or (self._preferredcodec == 'm4a' and filecodec == 'aac'):
278 if filecodec == 'aac' and self._preferredcodec in ['m4a', 'best']:
279 # Lossless, but in another container
280 acodec = 'copy'
281 extension = 'm4a'
282 more_opts = ['-bsf:a', 'aac_adtstoasc']
283 elif filecodec in ['aac', 'flac', 'mp3', 'vorbis', 'opus']:
284 # Lossless if possible
285 acodec = 'copy'
286 extension = filecodec
287 if filecodec == 'aac':
288 more_opts = ['-f', 'adts']
289 if filecodec == 'vorbis':
290 extension = 'ogg'
291 else:
292 # MP3 otherwise.
293 acodec = 'libmp3lame'
294 extension = 'mp3'
295 more_opts = []
296 if self._preferredquality is not None:
297 if int(self._preferredquality) < 10:
298 more_opts += ['-q:a', self._preferredquality]
299 else:
300 more_opts += ['-b:a', self._preferredquality + 'k']
301 else:
302 # We convert the audio (lossy if codec is lossy)
303 acodec = ACODECS[self._preferredcodec]
304 extension = self._preferredcodec
305 more_opts = []
306 if self._preferredquality is not None:
307 # The opus codec doesn't support the -aq option
308 if int(self._preferredquality) < 10 and extension != 'opus':
309 more_opts += ['-q:a', self._preferredquality]
310 else:
311 more_opts += ['-b:a', self._preferredquality + 'k']
312 if self._preferredcodec == 'aac':
313 more_opts += ['-f', 'adts']
314 if self._preferredcodec == 'm4a':
315 more_opts += ['-bsf:a', 'aac_adtstoasc']
316 if self._preferredcodec == 'vorbis':
317 extension = 'ogg'
318 if self._preferredcodec == 'wav':
319 extension = 'wav'
320 more_opts += ['-f', 'wav']
321
322 prefix, sep, ext = path.rpartition('.') # not os.path.splitext, since the latter does not work on unicode in all setups
323 new_path = prefix + sep + extension
324
325 information['filepath'] = new_path
326 information['ext'] = extension
327
328 # If we download foo.mp3 and convert it to... foo.mp3, then don't delete foo.mp3, silly.
329 if (new_path == path
330 or (self._nopostoverwrites and os.path.exists(encodeFilename(new_path)))):
331 self._downloader.to_screen('[ffmpeg] Post-process file %s exists, skipping' % new_path)
332 return [], information
333
334 try:
335 self._downloader.to_screen('[ffmpeg] Destination: ' + new_path)
336 self.run_ffmpeg(path, new_path, acodec, more_opts)
337 except AudioConversionError as e:
338 raise PostProcessingError(
339 'audio conversion failed: ' + e.msg)
340 except Exception:
341 raise PostProcessingError('error running ' + self.basename)
342
343 # Try to update the date time for extracted audio file.
344 if information.get('filetime') is not None:
345 self.try_utime(
346 new_path, time.time(), information['filetime'],
347 errnote='Cannot update utime of audio file')
348
349 return [path], information
350
351
352 class FFmpegVideoConvertorPP(FFmpegPostProcessor):
353 def __init__(self, downloader=None, preferedformat=None):
354 super(FFmpegVideoConvertorPP, self).__init__(downloader)
355 self._preferedformat = preferedformat
356
357 def run(self, information):
358 path = information['filepath']
359 if information['ext'] == self._preferedformat:
360 self._downloader.to_screen('[ffmpeg] Not converting video file %s - already is in target format %s' % (path, self._preferedformat))
361 return [], information
362 options = []
363 if self._preferedformat == 'avi':
364 options.extend(['-c:v', 'libxvid', '-vtag', 'XVID'])
365 prefix, sep, ext = path.rpartition('.')
366 outpath = prefix + sep + self._preferedformat
367 self._downloader.to_screen('[' + 'ffmpeg' + '] Converting video from %s to %s, Destination: ' % (information['ext'], self._preferedformat) + outpath)
368 self.run_ffmpeg(path, outpath, options)
369 information['filepath'] = outpath
370 information['format'] = self._preferedformat
371 information['ext'] = self._preferedformat
372 return [path], information
373
374
375 class FFmpegEmbedSubtitlePP(FFmpegPostProcessor):
376 def run(self, information):
377 if information['ext'] not in ('mp4', 'webm', 'mkv'):
378 self._downloader.to_screen('[ffmpeg] Subtitles can only be embedded in mp4, webm or mkv files')
379 return [], information
380 subtitles = information.get('requested_subtitles')
381 if not subtitles:
382 self._downloader.to_screen('[ffmpeg] There aren\'t any subtitles to embed')
383 return [], information
384
385 filename = information['filepath']
386
387 ext = information['ext']
388 sub_langs = []
389 sub_filenames = []
390 webm_vtt_warn = False
391
392 for lang, sub_info in subtitles.items():
393 sub_ext = sub_info['ext']
394 if ext != 'webm' or ext == 'webm' and sub_ext == 'vtt':
395 sub_langs.append(lang)
396 sub_filenames.append(subtitles_filename(filename, lang, sub_ext))
397 else:
398 if not webm_vtt_warn and ext == 'webm' and sub_ext != 'vtt':
399 webm_vtt_warn = True
400 self._downloader.to_screen('[ffmpeg] Only WebVTT subtitles can be embedded in webm files')
401
402 if not sub_langs:
403 return [], information
404
405 input_files = [filename] + sub_filenames
406
407 opts = [
408 '-map', '0',
409 '-c', 'copy',
410 # Don't copy the existing subtitles, we may be running the
411 # postprocessor a second time
412 '-map', '-0:s',
413 # Don't copy Apple TV chapters track, bin_data (see #19042, #19024,
414 # https://trac.ffmpeg.org/ticket/6016)
415 '-map', '-0:d',
416 ]
417 if information['ext'] == 'mp4':
418 opts += ['-c:s', 'mov_text']
419 for (i, lang) in enumerate(sub_langs):
420 opts.extend(['-map', '%d:0' % (i + 1)])
421 lang_code = ISO639Utils.short2long(lang) or lang
422 opts.extend(['-metadata:s:s:%d' % i, 'language=%s' % lang_code])
423
424 temp_filename = prepend_extension(filename, 'temp')
425 self._downloader.to_screen('[ffmpeg] Embedding subtitles in \'%s\'' % filename)
426 self.run_ffmpeg_multiple_files(input_files, temp_filename, opts)
427 os.remove(encodeFilename(filename))
428 os.rename(encodeFilename(temp_filename), encodeFilename(filename))
429
430 return sub_filenames, information
431
432
433 class FFmpegMetadataPP(FFmpegPostProcessor):
434 def run(self, info):
435 metadata = {}
436
437 def add(meta_list, info_list=None):
438 if not info_list:
439 info_list = meta_list
440 if not isinstance(meta_list, (list, tuple)):
441 meta_list = (meta_list,)
442 if not isinstance(info_list, (list, tuple)):
443 info_list = (info_list,)
444 for info_f in info_list:
445 if info.get(info_f) is not None:
446 for meta_f in meta_list:
447 metadata[meta_f] = info[info_f]
448 break
449
450 add('title', ('track', 'title'))
451 add('date', 'upload_date')
452 add(('description', 'comment'), 'description')
453 add('purl', 'webpage_url')
454 add('track', 'track_number')
455 add('artist', ('artist', 'creator', 'uploader', 'uploader_id'))
456 add('genre')
457 add('album')
458 add('album_artist')
459 add('disc', 'disc_number')
460
461 if not metadata:
462 self._downloader.to_screen('[ffmpeg] There isn\'t any metadata to add')
463 return [], info
464
465 filename = info['filepath']
466 temp_filename = prepend_extension(filename, 'temp')
467 in_filenames = [filename]
468 options = []
469
470 if info['ext'] == 'm4a':
471 options.extend(['-vn', '-acodec', 'copy'])
472 else:
473 options.extend(['-c', 'copy'])
474
475 for (name, value) in metadata.items():
476 options.extend(['-metadata', '%s=%s' % (name, value)])
477
478 chapters = info.get('chapters', [])
479 if chapters:
480 metadata_filename = replace_extension(filename, 'meta')
481 with io.open(metadata_filename, 'wt', encoding='utf-8') as f:
482 def ffmpeg_escape(text):
483 return re.sub(r'(=|;|#|\\|\n)', r'\\\1', text)
484
485 metadata_file_content = ';FFMETADATA1\n'
486 for chapter in chapters:
487 metadata_file_content += '[CHAPTER]\nTIMEBASE=1/1000\n'
488 metadata_file_content += 'START=%d\n' % (chapter['start_time'] * 1000)
489 metadata_file_content += 'END=%d\n' % (chapter['end_time'] * 1000)
490 chapter_title = chapter.get('title')
491 if chapter_title:
492 metadata_file_content += 'title=%s\n' % ffmpeg_escape(chapter_title)
493 f.write(metadata_file_content)
494 in_filenames.append(metadata_filename)
495 options.extend(['-map_metadata', '1'])
496
497 self._downloader.to_screen('[ffmpeg] Adding metadata to \'%s\'' % filename)
498 self.run_ffmpeg_multiple_files(in_filenames, temp_filename, options)
499 if chapters:
500 os.remove(metadata_filename)
501 os.remove(encodeFilename(filename))
502 os.rename(encodeFilename(temp_filename), encodeFilename(filename))
503 return [], info
504
505
506 class FFmpegMergerPP(FFmpegPostProcessor):
507 def run(self, info):
508 filename = info['filepath']
509 temp_filename = prepend_extension(filename, 'temp')
510 args = ['-c', 'copy', '-map', '0:v:0', '-map', '1:a:0']
511 self._downloader.to_screen('[ffmpeg] Merging formats into "%s"' % filename)
512 self.run_ffmpeg_multiple_files(info['__files_to_merge'], temp_filename, args)
513 os.rename(encodeFilename(temp_filename), encodeFilename(filename))
514 return info['__files_to_merge'], info
515
516 def can_merge(self):
517 # TODO: figure out merge-capable ffmpeg version
518 if self.basename != 'avconv':
519 return True
520
521 required_version = '10-0'
522 if is_outdated_version(
523 self._versions[self.basename], required_version):
524 warning = ('Your copy of %s is outdated and unable to properly mux separate video and audio files, '
525 'youtube-dl will download single file media. '
526 'Update %s to version %s or newer to fix this.') % (
527 self.basename, self.basename, required_version)
528 if self._downloader:
529 self._downloader.report_warning(warning)
530 return False
531 return True
532
533
534 class FFmpegFixupStretchedPP(FFmpegPostProcessor):
535 def run(self, info):
536 stretched_ratio = info.get('stretched_ratio')
537 if stretched_ratio is None or stretched_ratio == 1:
538 return [], info
539
540 filename = info['filepath']
541 temp_filename = prepend_extension(filename, 'temp')
542
543 options = ['-c', 'copy', '-aspect', '%f' % stretched_ratio]
544 self._downloader.to_screen('[ffmpeg] Fixing aspect ratio in "%s"' % filename)
545 self.run_ffmpeg(filename, temp_filename, options)
546
547 os.remove(encodeFilename(filename))
548 os.rename(encodeFilename(temp_filename), encodeFilename(filename))
549
550 return [], info
551
552
553 class FFmpegFixupM4aPP(FFmpegPostProcessor):
554 def run(self, info):
555 if info.get('container') != 'm4a_dash':
556 return [], info
557
558 filename = info['filepath']
559 temp_filename = prepend_extension(filename, 'temp')
560
561 options = ['-c', 'copy', '-f', 'mp4']
562 self._downloader.to_screen('[ffmpeg] Correcting container in "%s"' % filename)
563 self.run_ffmpeg(filename, temp_filename, options)
564
565 os.remove(encodeFilename(filename))
566 os.rename(encodeFilename(temp_filename), encodeFilename(filename))
567
568 return [], info
569
570
571 class FFmpegFixupM3u8PP(FFmpegPostProcessor):
572 def run(self, info):
573 filename = info['filepath']
574 if self.get_audio_codec(filename) == 'aac':
575 temp_filename = prepend_extension(filename, 'temp')
576
577 options = ['-c', 'copy', '-f', 'mp4', '-bsf:a', 'aac_adtstoasc']
578 self._downloader.to_screen('[ffmpeg] Fixing malformed AAC bitstream in "%s"' % filename)
579 self.run_ffmpeg(filename, temp_filename, options)
580
581 os.remove(encodeFilename(filename))
582 os.rename(encodeFilename(temp_filename), encodeFilename(filename))
583 return [], info
584
585
586 class FFmpegSubtitlesConvertorPP(FFmpegPostProcessor):
587 def __init__(self, downloader=None, format=None):
588 super(FFmpegSubtitlesConvertorPP, self).__init__(downloader)
589 self.format = format
590
591 def run(self, info):
592 subs = info.get('requested_subtitles')
593 filename = info['filepath']
594 new_ext = self.format
595 new_format = new_ext
596 if new_format == 'vtt':
597 new_format = 'webvtt'
598 if subs is None:
599 self._downloader.to_screen('[ffmpeg] There aren\'t any subtitles to convert')
600 return [], info
601 self._downloader.to_screen('[ffmpeg] Converting subtitles')
602 sub_filenames = []
603 for lang, sub in subs.items():
604 ext = sub['ext']
605 if ext == new_ext:
606 self._downloader.to_screen(
607 '[ffmpeg] Subtitle file for %s is already in the requested format' % new_ext)
608 continue
609 old_file = subtitles_filename(filename, lang, ext)
610 sub_filenames.append(old_file)
611 new_file = subtitles_filename(filename, lang, new_ext)
612
613 if ext in ('dfxp', 'ttml', 'tt'):
614 self._downloader.report_warning(
615 'You have requested to convert dfxp (TTML) subtitles into another format, '
616 'which results in style information loss')
617
618 dfxp_file = old_file
619 srt_file = subtitles_filename(filename, lang, 'srt')
620
621 with open(dfxp_file, 'rb') as f:
622 srt_data = dfxp2srt(f.read())
623
624 with io.open(srt_file, 'wt', encoding='utf-8') as f:
625 f.write(srt_data)
626 old_file = srt_file
627
628 subs[lang] = {
629 'ext': 'srt',
630 'data': srt_data
631 }
632
633 if new_ext == 'srt':
634 continue
635 else:
636 sub_filenames.append(srt_file)
637
638 self.run_ffmpeg(old_file, new_file, ['-f', new_format])
639
640 with io.open(new_file, 'rt', encoding='utf-8') as f:
641 subs[lang] = {
642 'ext': new_ext,
643 'data': f.read(),
644 }
645
646 return sub_filenames, info