]> jfr.im git - yt-dlp.git/blame - yt_dlp/__init__.py
Native SponsorBlock implementation and related improvements (#360)
[yt-dlp.git] / yt_dlp / __init__.py
CommitLineData
cc52de43 1#!/usr/bin/env python3
dcdb292f 2# coding: utf-8
235b3ba4 3
a4bc4336
PH
4from __future__ import unicode_literals
5
235b3ba4 6__license__ = 'Public Domain'
235b3ba4 7
0d94f247 8import codecs
8f563f32 9import io
e9f4ccd1 10import itertools
235b3ba4 11import os
0f818663 12import random
43820c03 13import re
235b3ba4 14import sys
235b3ba4 15
2daabe49
PH
16from .options import (
17 parseOpts,
18)
8c25f81b 19from .compat import (
e68301af 20 compat_getpass,
e9f4ccd1 21 compat_shlex_quote,
e07e9313 22 workaround_optparse_bug9161,
8c25f81b 23)
982ee69a 24from .cookies import SUPPORTED_BROWSERS
8c25f81b 25from .utils import (
a4fd0415
PH
26 DateRange,
27 decodeOption,
a4fd0415 28 DownloadError,
76a264ac 29 error_to_compat_str,
8b0d7497 30 ExistingVideoReached,
590bc6f6 31 expand_path,
347de493 32 match_filter_func,
a4fd0415 33 MaxDownloadsReached,
a4fd0415 34 preferredencoding,
62e609ab 35 read_batch_urls,
8b0d7497 36 RejectedVideoReached,
df692c5a 37 render_table,
a4fd0415 38 SameFileError,
e3946f98 39 setproctitle,
a4fd0415
PH
40 std_headers,
41 write_string,
a4fd0415 42)
c19bc311 43from .update import run_update
92a86f4c 44from .downloader import (
a4fd0415
PH
45 FileDownloader,
46)
2bad0e5d 47from .extractor import gen_extractors, list_extractors
eb8a4433 48from .extractor.common import InfoExtractor
1b6712ab 49from .extractor.adobepass import MSO_INFO
e9f4ccd1 50from .postprocessor import (
81a23040 51 FFmpegExtractAudioPP,
52 FFmpegSubtitlesConvertorPP,
53 FFmpegThumbnailsConvertorPP,
54 FFmpegVideoConvertorPP,
55 FFmpegVideoRemuxerPP,
e9f4ccd1 56 MetadataFromFieldPP,
57 MetadataParserPP,
81a23040 58)
8222d8de 59from .YoutubeDL import YoutubeDL
a4fd0415 60
235b3ba4 61
b8ad4f02 62def _real_main(argv=None):
0d94f247
PH
63 # Compatibility fixes for Windows
64 if sys.platform == 'win32':
067aa17e 65 # https://github.com/ytdl-org/youtube-dl/issues/820
0d94f247
PH
66 codecs.register(lambda name: codecs.lookup('utf-8') if name == 'cp65001' else None)
67
e07e9313
PH
68 workaround_optparse_bug9161()
69
7a5c1cfe 70 setproctitle('yt-dlp')
e3946f98 71
b8ad4f02 72 parser, opts, args = parseOpts(argv)
0d1bb027 73 warnings = []
59ae15a5 74
59ae15a5
PH
75 # Set user agent
76 if opts.user_agent is not None:
77 std_headers['User-Agent'] = opts.user_agent
1865ed31 78
28535652
BH
79 # Set referer
80 if opts.referer is not None:
81 std_headers['Referer'] = opts.referer
59ae15a5 82
410afb20 83 # Custom HTTP headers
45016689 84 std_headers.update(opts.headers)
410afb20 85
59ae15a5
PH
86 # Dump user agent
87 if opts.dump_user_agent:
7b0d1c28 88 write_string(std_headers['User-Agent'] + '\n', out=sys.stdout)
59ae15a5
PH
89 sys.exit(0)
90
91 # Batch file verification
62e609ab 92 batch_urls = []
59ae15a5
PH
93 if opts.batchfile is not None:
94 try:
95 if opts.batchfile == '-':
96 batchfd = sys.stdin
97 else:
e2eca6f6 98 batchfd = io.open(
590bc6f6 99 expand_path(opts.batchfile),
e2eca6f6 100 'r', encoding='utf-8', errors='ignore')
62e609ab 101 batch_urls = read_batch_urls(batchfd)
05afc96b 102 if opts.verbose:
a4bc4336 103 write_string('[debug] Batch file urls: ' + repr(batch_urls) + '\n')
59ae15a5 104 except IOError:
826dcff9 105 sys.exit('ERROR: batch file %s could not be read' % opts.batchfile)
b2fc1c4f 106 all_urls = batch_urls + [url.strip() for url in args] # batch_urls are already striped in read_batch_urls
c774b3c6 107 _enc = preferredencoding()
41292a38 108 all_urls = [url.decode(_enc, 'ignore') if isinstance(url, bytes) else url for url in all_urls]
59ae15a5 109
59ae15a5 110 if opts.list_extractors:
05900629 111 for ie in list_extractors(opts.age_limit):
251ae04e 112 write_string(ie.IE_NAME + (' (CURRENTLY BROKEN)' if not ie.working() else '') + '\n', out=sys.stdout)
1a2c3c0f 113 matchedUrls = [url for url in all_urls if ie.suitable(url)]
59ae15a5 114 for mu in matchedUrls:
7b0d1c28 115 write_string(' ' + mu + '\n', out=sys.stdout)
59ae15a5 116 sys.exit(0)
0f818663 117 if opts.list_extractor_descriptions:
05900629 118 for ie in list_extractors(opts.age_limit):
251ae04e 119 if not ie.working():
0f818663
PH
120 continue
121 desc = getattr(ie, 'IE_DESC', ie.IE_NAME)
15870e90
PH
122 if desc is False:
123 continue
0f818663 124 if hasattr(ie, 'SEARCH_KEY'):
50a0f6df 125 _SEARCHES = ('cute kittens', 'slithering pythons', 'falling cat', 'angry poodle', 'purple fish', 'running tortoise', 'sleeping bunny', 'burping cow')
a4bc4336
PH
126 _COUNTS = ('', '5', '10', 'all')
127 desc += ' (Example: "%s%s:%s" )' % (ie.SEARCH_KEY, random.choice(_COUNTS), random.choice(_SEARCHES))
7b0d1c28 128 write_string(desc + '\n', out=sys.stdout)
0f818663 129 sys.exit(0)
87148bb7 130 if opts.ap_list_mso:
1b6712ab 131 table = [[mso_id, mso_info['name']] for mso_id, mso_info in MSO_INFO.items()]
797c636b 132 write_string('Supported TV Providers:\n' + render_table(['mso', 'mso name'], table) + '\n', out=sys.stdout)
1b6712ab 133 sys.exit(0)
0f818663 134
59ae15a5
PH
135 # Conflicting, missing and erroneous options
136 if opts.usenetrc and (opts.username is not None or opts.password is not None):
a4bc4336 137 parser.error('using .netrc conflicts with giving username/password')
59ae15a5 138 if opts.password is not None and opts.username is None:
a4bc4336 139 parser.error('account username missing\n')
1b6712ab
RA
140 if opts.ap_password is not None and opts.ap_username is None:
141 parser.error('TV Provider account username missing\n')
1a241a2d
S
142 if opts.autonumber_size is not None:
143 if opts.autonumber_size <= 0:
144 parser.error('auto number size must be positive')
145 if opts.autonumber_start is not None:
146 if opts.autonumber_start < 0:
147 parser.error('auto number start must be positive or 0')
59ae15a5 148 if opts.username is not None and opts.password is None:
a4bc4336 149 opts.password = compat_getpass('Type account password and press [Return]: ')
1b6712ab
RA
150 if opts.ap_username is not None and opts.ap_password is None:
151 opts.ap_password = compat_getpass('Type TV provider account password and press [Return]: ')
59ae15a5
PH
152 if opts.ratelimit is not None:
153 numeric_limit = FileDownloader.parse_bytes(opts.ratelimit)
154 if numeric_limit is None:
a4bc4336 155 parser.error('invalid rate limit specified')
59ae15a5 156 opts.ratelimit = numeric_limit
51d9739f 157 if opts.throttledratelimit is not None:
158 numeric_limit = FileDownloader.parse_bytes(opts.throttledratelimit)
159 if numeric_limit is None:
160 parser.error('invalid rate limit specified')
161 opts.throttledratelimit = numeric_limit
9e982f9e
JC
162 if opts.min_filesize is not None:
163 numeric_limit = FileDownloader.parse_bytes(opts.min_filesize)
164 if numeric_limit is None:
a4bc4336 165 parser.error('invalid min_filesize specified')
9e982f9e
JC
166 opts.min_filesize = numeric_limit
167 if opts.max_filesize is not None:
168 numeric_limit = FileDownloader.parse_bytes(opts.max_filesize)
169 if numeric_limit is None:
a4bc4336 170 parser.error('invalid max_filesize specified')
9e982f9e 171 opts.max_filesize = numeric_limit
065bc354 172 if opts.sleep_interval is not None:
173 if opts.sleep_interval < 0:
1ad6b891
S
174 parser.error('sleep interval must be positive or 0')
175 if opts.max_sleep_interval is not None:
176 if opts.max_sleep_interval < 0:
177 parser.error('max sleep interval must be positive or 0')
050afa60
JZ
178 if opts.sleep_interval is None:
179 parser.error('min sleep interval must be specified, use --min-sleep-interval')
1ad6b891
S
180 if opts.max_sleep_interval < opts.sleep_interval:
181 parser.error('max sleep interval must be greater than or equal to min sleep interval')
182 else:
183 opts.max_sleep_interval = opts.sleep_interval
1cf376f5 184 if opts.sleep_interval_subtitles is not None:
185 if opts.sleep_interval_subtitles < 0:
186 parser.error('subtitles sleep interval must be positive or 0')
187 if opts.sleep_interval_requests is not None:
188 if opts.sleep_interval_requests < 0:
189 parser.error('requests sleep interval must be positive or 0')
797c636b 190 if opts.ap_mso and opts.ap_mso not in MSO_INFO:
87148bb7 191 parser.error('Unsupported TV Provider, use --ap-list-mso to get a list of supported TV Providers')
486fb179 192 if opts.overwrites: # --yes-overwrites implies --no-continue
0c3d0f51 193 opts.continue_dl = False
4cf1e5d2 194 if opts.concurrent_fragment_downloads <= 0:
195 raise ValueError('Concurrent fragments must be positive')
52bb437e 196
62bff2c1 197 def parse_retries(retries, name=''):
52bb437e
S
198 if retries in ('inf', 'infinite'):
199 parsed_retries = float('inf')
baeaeffc
PH
200 else:
201 try:
52bb437e 202 parsed_retries = int(retries)
baeaeffc 203 except (TypeError, ValueError):
62bff2c1 204 parser.error('invalid %sretry count specified' % name)
52bb437e
S
205 return parsed_retries
206 if opts.retries is not None:
207 opts.retries = parse_retries(opts.retries)
208 if opts.fragment_retries is not None:
62bff2c1 209 opts.fragment_retries = parse_retries(opts.fragment_retries, 'fragment ')
210 if opts.extractor_retries is not None:
211 opts.extractor_retries = parse_retries(opts.extractor_retries, 'extractor ')
59ae15a5
PH
212 if opts.buffersize is not None:
213 numeric_buffersize = FileDownloader.parse_bytes(opts.buffersize)
214 if numeric_buffersize is None:
a4bc4336 215 parser.error('invalid buffer size specified')
59ae15a5 216 opts.buffersize = numeric_buffersize
ba515388
S
217 if opts.http_chunk_size is not None:
218 numeric_chunksize = FileDownloader.parse_bytes(opts.http_chunk_size)
219 if not numeric_chunksize:
220 parser.error('invalid http chunk size specified')
221 opts.http_chunk_size = numeric_chunksize
a19fd00c 222 if opts.playliststart <= 0:
a4bc4336 223 raise ValueError('Playlist start must be positive')
a19fd00c 224 if opts.playlistend not in (-1, None) and opts.playlistend < opts.playliststart:
a4bc4336 225 raise ValueError('Playlist end must be greater than playlist start')
59ae15a5 226 if opts.extractaudio:
81a23040 227 if opts.audioformat not in ['best'] + list(FFmpegExtractAudioPP.SUPPORTED_EXTS):
a4bc4336 228 parser.error('invalid audio format specified')
59ae15a5
PH
229 if opts.audioquality:
230 opts.audioquality = opts.audioquality.strip('k').strip('K')
231 if not opts.audioquality.isdigit():
a4bc4336 232 parser.error('invalid audio quality specified')
7851b379 233 if opts.recodevideo is not None:
81a23040 234 opts.recodevideo = opts.recodevideo.replace(' ', '')
235 if not re.match(FFmpegVideoConvertorPP.FORMAT_RE, opts.recodevideo):
236 parser.error('invalid video remux format specified')
17912249 237 if opts.remuxvideo is not None:
df692c5a 238 opts.remuxvideo = opts.remuxvideo.replace(' ', '')
81a23040 239 if not re.match(FFmpegVideoRemuxerPP.FORMAT_RE, opts.remuxvideo):
17912249 240 parser.error('invalid video remux format specified')
e9fade72 241 if opts.convertsubtitles is not None:
81a23040 242 if opts.convertsubtitles not in FFmpegSubtitlesConvertorPP.SUPPORTED_EXTS:
e9fade72 243 parser.error('invalid subtitle format specified')
8fa43c73 244 if opts.convertthumbnails is not None:
81a23040 245 if opts.convertthumbnails not in FFmpegThumbnailsConvertorPP.SUPPORTED_EXTS:
8fa43c73 246 parser.error('invalid thumbnail format specified')
bd1a281e 247
982ee69a
MB
248 if opts.cookiesfrombrowser is not None:
249 opts.cookiesfrombrowser = [
250 part.strip() or None for part in opts.cookiesfrombrowser.split(':', 1)]
251 if opts.cookiesfrombrowser[0] not in SUPPORTED_BROWSERS:
252 parser.error('unsupported browser specified for cookies')
253
bd558525
JMF
254 if opts.date is not None:
255 date = DateRange.day(opts.date)
256 else:
257 date = DateRange(opts.dateafter, opts.datebefore)
59ae15a5 258
31654882 259 compat_opts = opts.compat_opts
53ed7066 260
261 def _unused_compat_opt(name):
262 if name not in compat_opts:
263 return False
264 compat_opts.discard(name)
265 compat_opts.update(['*%s' % name])
266 return True
267
e4f02757 268 def set_default_compat(compat_name, opt_name, default=True, remove_compat=True):
53ed7066 269 attr = getattr(opts, opt_name)
270 if compat_name in compat_opts:
271 if attr is None:
272 setattr(opts, opt_name, not default)
273 return True
274 else:
275 if remove_compat:
276 _unused_compat_opt(compat_name)
277 return False
278 elif attr is None:
279 setattr(opts, opt_name, default)
280 return None
281
282 set_default_compat('abort-on-error', 'ignoreerrors')
283 set_default_compat('no-playlist-metafiles', 'allow_playlist_files')
e4f02757 284 set_default_compat('no-clean-infojson', 'clean_infojson')
53ed7066 285 if 'format-sort' in compat_opts:
286 opts.format_sort.extend(InfoExtractor.FormatSort.ytdl_default)
287 _video_multistreams_set = set_default_compat('multistreams', 'allow_multiple_video_streams', False, remove_compat=False)
288 _audio_multistreams_set = set_default_compat('multistreams', 'allow_multiple_audio_streams', False, remove_compat=False)
289 if _video_multistreams_set is False and _audio_multistreams_set is False:
290 _unused_compat_opt('multistreams')
291 outtmpl_default = opts.outtmpl.get('default')
292 if 'filename' in compat_opts:
293 if outtmpl_default is None:
9d65e7bd 294 outtmpl_default = '%(title)s-%(id)s.%(ext)s'
53ed7066 295 opts.outtmpl.update({'default': outtmpl_default})
296 else:
297 _unused_compat_opt('filename')
298
76a264ac 299 def validate_outtmpl(tmpl, msg):
300 err = YoutubeDL.validate_outtmpl(tmpl)
301 if err:
302 parser.error('invalid %s %r: %s' % (msg, tmpl, error_to_compat_str(err)))
303
304 for k, tmpl in opts.outtmpl.items():
305 validate_outtmpl(tmpl, '%s output template' % k)
ad3dc496 306 opts.forceprint = opts.forceprint or []
307 for tmpl in opts.forceprint or []:
76a264ac 308 validate_outtmpl(tmpl, 'print template')
7a340e0d 309 validate_outtmpl(opts.sponsorblock_chapter_title, 'SponsorBlock chapter title')
76a264ac 310
de3ef3ed
PH
311 if opts.extractaudio and not opts.keepvideo and opts.format is None:
312 opts.format = 'bestaudio/best'
313
de6000d9 314 if outtmpl_default is not None and not os.path.splitext(outtmpl_default)[1] and opts.extractaudio:
a4bc4336
PH
315 parser.error('Cannot download a video and extract audio into the same'
316 ' file! Use "{0}.%(ext)s" instead of "{0}" as the output'
de6000d9 317 ' template'.format(outtmpl_default))
0202b52a 318
eb8a4433 319 for f in opts.format_sort:
320 if re.match(InfoExtractor.FormatSort.regex, f) is None:
321 parser.error('invalid format sort string "%s" specified' % f)
29c7a63d 322
e9f4ccd1 323 def metadataparser_actions(f):
324 if isinstance(f, str):
325 cmd = '--parse-metadata %s' % compat_shlex_quote(f)
326 try:
327 actions = [MetadataFromFieldPP.to_action(f)]
328 except Exception as err:
329 parser.error(f'{cmd} is invalid; {err}')
330 else:
331 cmd = '--replace-in-metadata %s' % ' '.join(map(compat_shlex_quote, f))
332 actions = ((MetadataParserPP.Actions.REPLACE, x, *f[1:]) for x in f[0].split(','))
333
334 for action in actions:
335 try:
336 MetadataParserPP.validate_action(*action)
337 except Exception as err:
338 parser.error(f'{cmd} is invalid; {err}')
339 yield action
340
341 if opts.parse_metadata is None:
342 opts.parse_metadata = []
5bfa4862 343 if opts.metafromtitle is not None:
e9f4ccd1 344 opts.parse_metadata.append('title:%s' % opts.metafromtitle)
345 opts.parse_metadata = list(itertools.chain(*map(metadataparser_actions, opts.parse_metadata)))
5bfa4862 346
d2a1fad9 347 any_getting = opts.forceprint or opts.geturl or opts.gettitle or opts.getid or opts.getthumbnail or opts.getdescription or opts.getfilename or opts.getformat or opts.getduration or opts.dumpjson or opts.dump_single_json
c0bdf32a 348 any_printing = opts.print_json
590bc6f6 349 download_archive_fn = expand_path(opts.download_archive) if opts.download_archive is not None else opts.download_archive
525ef922 350
f0884c8b 351 # If JSON is not printed anywhere, but comments are requested, save it to file
352 printing_json = opts.dumpjson or opts.print_json or opts.dump_single_json
353 if opts.getcomments and not printing_json:
354 opts.writeinfojson = True
355
7a340e0d
NA
356 if opts.no_sponsorblock:
357 opts.sponsorblock_mark = set()
358 opts.sponsorblock_remove = set()
359 sponsorblock_query = opts.sponsorblock_mark | opts.sponsorblock_remove
360
361 if (opts.addmetadata or opts.sponsorblock_mark) and opts.addchapters is None:
362 opts.addchapters = True
363 opts.remove_chapters = opts.remove_chapters or []
364
63ad4d43 365 def report_conflict(arg1, arg2):
0d1bb027 366 warnings.append('%s is ignored since %s was given' % (arg2, arg1))
72755351 367
7a340e0d
NA
368 if (opts.remove_chapters or sponsorblock_query) and opts.sponskrub is not False:
369 if opts.sponskrub:
370 if opts.remove_chapters:
371 report_conflict('--remove-chapters', '--sponskrub')
372 if opts.sponsorblock_mark:
373 report_conflict('--sponsorblock-mark', '--sponskrub')
374 if opts.sponsorblock_remove:
375 report_conflict('--sponsorblock-remove', '--sponskrub')
376 opts.sponskrub = False
72755351 377 if opts.sponskrub_cut and opts.split_chapters and opts.sponskrub is not False:
378 report_conflict('--split-chapter', '--sponskrub-cut')
379 opts.sponskrub_cut = False
380
7a340e0d
NA
381 if opts.remuxvideo and opts.recodevideo:
382 report_conflict('--recode-video', '--remux-video')
383 opts.remuxvideo = False
384
63ad4d43 385 if opts.allow_unplayable_formats:
386 if opts.extractaudio:
387 report_conflict('--allow-unplayable-formats', '--extract-audio')
388 opts.extractaudio = False
389 if opts.remuxvideo:
390 report_conflict('--allow-unplayable-formats', '--remux-video')
391 opts.remuxvideo = False
392 if opts.recodevideo:
393 report_conflict('--allow-unplayable-formats', '--recode-video')
394 opts.recodevideo = False
395 if opts.addmetadata:
396 report_conflict('--allow-unplayable-formats', '--add-metadata')
397 opts.addmetadata = False
398 if opts.embedsubtitles:
399 report_conflict('--allow-unplayable-formats', '--embed-subs')
400 opts.embedsubtitles = False
401 if opts.embedthumbnail:
402 report_conflict('--allow-unplayable-formats', '--embed-thumbnail')
403 opts.embedthumbnail = False
404 if opts.xattrs:
405 report_conflict('--allow-unplayable-formats', '--xattrs')
406 opts.xattrs = False
407 if opts.fixup and opts.fixup.lower() not in ('never', 'ignore'):
408 report_conflict('--allow-unplayable-formats', '--fixup')
409 opts.fixup = 'never'
7a340e0d
NA
410 if opts.remove_chapters:
411 report_conflict('--allow-unplayable-formats', '--remove-chapters')
412 opts.remove_chapters = []
413 if opts.sponsorblock_remove:
414 report_conflict('--allow-unplayable-formats', '--sponsorblock-remove')
415 opts.sponsorblock_remove = set()
63ad4d43 416 if opts.sponskrub:
417 report_conflict('--allow-unplayable-formats', '--sponskrub')
418 opts.sponskrub = False
419
4f026faf
PH
420 # PostProcessors
421 postprocessors = []
7a340e0d
NA
422 if sponsorblock_query:
423 postprocessors.append({
424 'key': 'SponsorBlock',
425 'categories': sponsorblock_query,
426 'api': opts.sponsorblock_api,
427 # Run this immediately after extraction is complete
428 'when': 'pre_process'
429 })
e9f4ccd1 430 if opts.parse_metadata:
e7db87f7 431 postprocessors.append({
e9f4ccd1 432 'key': 'MetadataParser',
433 'actions': opts.parse_metadata,
56d868db 434 # Run this immediately after extraction is complete
435 'when': 'pre_process'
436 })
437 if opts.convertsubtitles:
438 postprocessors.append({
439 'key': 'FFmpegSubtitlesConvertor',
440 'format': opts.convertsubtitles,
441 # Run this before the actual video download
442 'when': 'before_dl'
e7db87f7 443 })
8fa43c73 444 if opts.convertthumbnails:
445 postprocessors.append({
446 'key': 'FFmpegThumbnailsConvertor',
447 'format': opts.convertthumbnails,
448 # Run this before the actual video download
449 'when': 'before_dl'
450 })
5520aa2d 451 # Must be after all other before_dl
452 if opts.exec_before_dl_cmd:
453 postprocessors.append({
ad3dc496 454 'key': 'Exec',
5520aa2d 455 'exec_cmd': opts.exec_before_dl_cmd,
456 'when': 'before_dl'
457 })
4f026faf
PH
458 if opts.extractaudio:
459 postprocessors.append({
460 'key': 'FFmpegExtractAudio',
461 'preferredcodec': opts.audioformat,
462 'preferredquality': opts.audioquality,
463 'nopostoverwrites': opts.nopostoverwrites,
464 })
efe87a10
FS
465 if opts.remuxvideo:
466 postprocessors.append({
467 'key': 'FFmpegVideoRemuxer',
468 'preferedformat': opts.remuxvideo,
469 })
4f026faf
PH
470 if opts.recodevideo:
471 postprocessors.append({
472 'key': 'FFmpegVideoConvertor',
473 'preferedformat': opts.recodevideo,
474 })
7a340e0d 475 # If ModifyChapters is going to remove chapters, subtitles must already be in the container.
4f026faf 476 if opts.embedsubtitles:
b51d2ae3 477 already_have_subtitle = opts.writesubtitles and 'no-keep-subs' not in compat_opts
4f026faf
PH
478 postprocessors.append({
479 'key': 'FFmpegEmbedSubtitle',
56d868db 480 # already_have_subtitle = True prevents the file from being deleted after embedding
cffab0ee 481 'already_have_subtitle': already_have_subtitle
4f026faf 482 })
b51d2ae3 483 if not opts.writeautomaticsub and 'no-keep-subs' not in compat_opts:
cffab0ee 484 opts.writesubtitles = True
485 # --all-sub automatically sets --write-sub if --write-auto-sub is not given
486 # this was the old behaviour if only --all-sub was given.
487 if opts.allsubtitles and not opts.writeautomaticsub:
488 opts.writesubtitles = True
7a340e0d
NA
489 # ModifyChapters must run before FFmpegMetadataPP
490 remove_chapters_patterns = []
491 for regex in opts.remove_chapters:
492 try:
493 remove_chapters_patterns.append(re.compile(regex))
494 except re.error as err:
495 parser.error(f'invalid --remove-chapters regex {regex!r} - {err}')
496 if opts.remove_chapters or sponsorblock_query:
497 postprocessors.append({
498 'key': 'ModifyChapters',
499 'remove_chapters_patterns': remove_chapters_patterns,
500 'remove_sponsor_segments': opts.sponsorblock_remove,
501 'sponsorblock_chapter_title': opts.sponsorblock_chapter_title,
502 'force_keyframes': opts.force_keyframes_at_cuts
503 })
504 # FFmpegMetadataPP should be run after FFmpegVideoConvertorPP and
505 # FFmpegExtractAudioPP as containers before conversion may not support
506 # metadata (3gp, webm, etc.)
507 # By default ffmpeg preserves metadata applicable for both
508 # source and target containers. From this point the container won't change,
509 # so metadata can be added here.
510 if opts.addmetadata or opts.addchapters:
511 postprocessors.append({
512 'key': 'FFmpegMetadata',
513 'add_chapters': opts.addchapters,
514 'add_metadata': opts.addmetadata,
515 })
f4e4be19 516 # This should be above EmbedThumbnail since sponskrub removes the thumbnail attachment
517 # but must be below EmbedSubtitle and FFmpegMetadata
518 # See https://github.com/yt-dlp/yt-dlp/issues/204 , https://github.com/faissaloo/SponSkrub/issues/29
a9e7f546 519 # If opts.sponskrub is None, sponskrub is used, but it silently fails if the executable can't be found
520 if opts.sponskrub is not False:
521 postprocessors.append({
522 'key': 'SponSkrub',
523 'path': opts.sponskrub_path,
524 'args': opts.sponskrub_args,
525 'cut': opts.sponskrub_cut,
526 'force': opts.sponskrub_force,
527 'ignoreerror': opts.sponskrub is None,
528 })
f4e4be19 529 if opts.embedthumbnail:
530 already_have_thumbnail = opts.writethumbnail or opts.write_all_thumbnails
531 postprocessors.append({
532 'key': 'EmbedThumbnail',
56d868db 533 # already_have_thumbnail = True prevents the file from being deleted after embedding
f4e4be19 534 'already_have_thumbnail': already_have_thumbnail
535 })
536 if not already_have_thumbnail:
537 opts.writethumbnail = True
72755351 538 if opts.split_chapters:
7a340e0d
NA
539 postprocessors.append({
540 'key': 'FFmpegSplitChapters',
541 'force_keyframes': opts.force_keyframes_at_cuts,
542 })
72755351 543 # XAttrMetadataPP should be run after post-processors that may change file contents
544 if opts.xattrs:
545 postprocessors.append({'key': 'XAttrMetadata'})
ad3dc496 546 # Exec must be the last PP
4f026faf
PH
547 if opts.exec_cmd:
548 postprocessors.append({
ad3dc496 549 'key': 'Exec',
4f026faf 550 'exec_cmd': opts.exec_cmd,
56d868db 551 # Run this only after the files have been moved to their final locations
552 'when': 'after_move'
4f026faf 553 })
1b77b347 554
06869367 555 def report_args_compat(arg, name):
0d1bb027 556 warnings.append('%s given without specifying name. The arguments will be given to all %s' % (arg, name))
557
b8f6bbe6 558 if 'default' in opts.external_downloader_args:
0d1bb027 559 report_args_compat('--downloader-args', 'external downloaders')
b8f6bbe6 560
45016689 561 if 'default-compat' in opts.postprocessor_args and 'default' not in opts.postprocessor_args:
06869367 562 report_args_compat('--post-processor-args', 'post-processors')
45016689 563 opts.postprocessor_args.setdefault('sponskrub', [])
564 opts.postprocessor_args['default'] = opts.postprocessor_args['default-compat']
1b77b347 565
df692c5a 566 final_ext = (
81a23040 567 opts.recodevideo if opts.recodevideo in FFmpegVideoConvertorPP.SUPPORTED_EXTS
568 else opts.remuxvideo if opts.remuxvideo in FFmpegVideoRemuxerPP.SUPPORTED_EXTS
569 else opts.audioformat if (opts.extractaudio and opts.audioformat != 'best')
570 else None)
f6d7624f 571
347de493
PH
572 match_filter = (
573 None if opts.match_filter is None
574 else match_filter_func(opts.match_filter))
4f026faf 575
bdde425c 576 ydl_opts = {
59ae15a5
PH
577 'usenetrc': opts.usenetrc,
578 'username': opts.username,
579 'password': opts.password,
83317f69 580 'twofactor': opts.twofactor,
c6c19746 581 'videopassword': opts.videopassword,
797c636b 582 'ap_mso': opts.ap_mso,
1b6712ab
RA
583 'ap_username': opts.ap_username,
584 'ap_password': opts.ap_password,
c0bdf32a 585 'quiet': (opts.quiet or any_getting or any_printing),
ad8915b7 586 'no_warnings': opts.no_warnings,
59ae15a5
PH
587 'forceurl': opts.geturl,
588 'forcetitle': opts.gettitle,
1a2adf3f 589 'forceid': opts.getid,
59ae15a5
PH
590 'forcethumbnail': opts.getthumbnail,
591 'forcedescription': opts.getdescription,
525ef922 592 'forceduration': opts.getduration,
59ae15a5
PH
593 'forcefilename': opts.getfilename,
594 'forceformat': opts.getformat,
d2a1fad9 595 'forceprint': opts.forceprint,
c0bdf32a 596 'forcejson': opts.dumpjson or opts.print_json,
63e0be34 597 'dump_single_json': opts.dump_single_json,
2d30509f 598 'force_write_download_archive': opts.force_write_download_archive,
b7b04c78 599 'simulate': (any_getting or None) if opts.simulate is None else opts.simulate,
1bdeb7be 600 'skip_download': opts.skip_download,
59ae15a5 601 'format': opts.format,
63ad4d43 602 'allow_unplayable_formats': opts.allow_unplayable_formats,
b7da73eb 603 'ignore_no_formats_error': opts.ignore_no_formats_error,
eb8a4433 604 'format_sort': opts.format_sort,
605 'format_sort_force': opts.format_sort_force,
909d24dd 606 'allow_multiple_video_streams': opts.allow_multiple_video_streams,
607 'allow_multiple_audio_streams': opts.allow_multiple_audio_streams,
e8e73840 608 'check_formats': opts.check_formats,
59ae15a5 609 'listformats': opts.listformats,
76d321f6 610 'listformats_table': opts.listformats_table,
486fb179 611 'outtmpl': opts.outtmpl,
a820dc72 612 'outtmpl_na_placeholder': opts.outtmpl_na_placeholder,
0202b52a 613 'paths': opts.paths,
213c31ae 614 'autonumber_size': opts.autonumber_size,
acbb2374 615 'autonumber_start': opts.autonumber_start,
59ae15a5 616 'restrictfilenames': opts.restrictfilenames,
c2934512 617 'windowsfilenames': opts.windowsfilenames,
59ae15a5 618 'ignoreerrors': opts.ignoreerrors,
d22dec74 619 'force_generic_extractor': opts.force_generic_extractor,
59ae15a5 620 'ratelimit': opts.ratelimit,
51d9739f 621 'throttledratelimit': opts.throttledratelimit,
0c3d0f51 622 'overwrites': opts.overwrites,
52bb437e
S
623 'retries': opts.retries,
624 'fragment_retries': opts.fragment_retries,
62bff2c1 625 'extractor_retries': opts.extractor_retries,
9603b660 626 'skip_unavailable_fragments': opts.skip_unavailable_fragments,
0eee52f3 627 'keep_fragments': opts.keep_fragments,
4cf1e5d2 628 'concurrent_fragment_downloads': opts.concurrent_fragment_downloads,
59ae15a5
PH
629 'buffersize': opts.buffersize,
630 'noresizebuffer': opts.noresizebuffer,
ba515388 631 'http_chunk_size': opts.http_chunk_size,
59ae15a5
PH
632 'continuedl': opts.continue_dl,
633 'noprogress': opts.noprogress,
5717d91a 634 'progress_with_newline': opts.progress_with_newline,
59ae15a5
PH
635 'playliststart': opts.playliststart,
636 'playlistend': opts.playlistend,
ff815fe6 637 'playlistreverse': opts.playlist_reverse,
75822ca7 638 'playlistrandom': opts.playlist_random,
47192f92 639 'noplaylist': opts.noplaylist,
de6000d9 640 'logtostderr': outtmpl_default == '-',
59ae15a5
PH
641 'consoletitle': opts.consoletitle,
642 'nopart': opts.nopart,
643 'updatetime': opts.updatetime,
644 'writedescription': opts.writedescription,
1fb07d10 645 'writeannotations': opts.writeannotations,
f0884c8b 646 'writeinfojson': opts.writeinfojson,
1ea24129 647 'allow_playlist_files': opts.allow_playlist_files,
75d43ca0 648 'clean_infojson': opts.clean_infojson,
06167fbb 649 'getcomments': opts.getcomments,
11d9224e 650 'writethumbnail': opts.writethumbnail,
ec82d85a 651 'write_all_thumbnails': opts.write_all_thumbnails,
732044af 652 'writelink': opts.writelink,
653 'writeurllink': opts.writeurllink,
654 'writewebloclink': opts.writewebloclink,
655 'writedesktoplink': opts.writedesktoplink,
59ae15a5 656 'writesubtitles': opts.writesubtitles,
b004821f 657 'writeautomaticsub': opts.writeautomaticsub,
ae608b80 658 'allsubtitles': opts.allsubtitles,
2a4093ea 659 'listsubtitles': opts.listsubtitles,
9e62bc44 660 'subtitlesformat': opts.subtitlesformat,
d6e203b3 661 'subtitleslangs': opts.subtitleslangs,
8271226a
PH
662 'matchtitle': decodeOption(opts.matchtitle),
663 'rejecttitle': decodeOption(opts.rejecttitle),
59ae15a5
PH
664 'max_downloads': opts.max_downloads,
665 'prefer_free_formats': opts.prefer_free_formats,
bdc3fd2f 666 'trim_file_name': opts.trim_file_name,
59ae15a5 667 'verbose': opts.verbose,
855703e5 668 'dump_intermediate_pages': opts.dump_intermediate_pages,
d41e6efc 669 'write_pages': opts.write_pages,
8d5d3a5d 670 'test': opts.test,
7851b379 671 'keepvideo': opts.keepvideo,
9e982f9e 672 'min_filesize': opts.min_filesize,
bd558525 673 'max_filesize': opts.max_filesize,
5fe18bdb
PH
674 'min_views': opts.min_views,
675 'max_views': opts.max_views,
11d9224e 676 'daterange': date,
7f747732 677 'cachedir': opts.cachedir,
f8061589 678 'youtube_print_sig_code': opts.youtube_print_sig_code,
8dbe9899 679 'age_limit': opts.age_limit,
17093b83 680 'download_archive': download_archive_fn,
ea6e0c2b 681 'break_on_existing': opts.break_on_existing,
8b0d7497 682 'break_on_reject': opts.break_on_reject,
26e2805c 683 'skip_playlist_after_errors': opts.skip_playlist_after_errors,
dca08720 684 'cookiefile': opts.cookiefile,
982ee69a 685 'cookiesfrombrowser': opts.cookiesfrombrowser,
dca08720 686 'nocheckcertificate': opts.no_check_certificate,
7e8c0af0 687 'prefer_insecure': opts.prefer_insecure,
c2e52508 688 'proxy': opts.proxy,
6ad14cab 689 'socket_timeout': opts.socket_timeout,
0783b09b 690 'bidi_workaround': opts.bidi_workaround,
a0ddb8a2 691 'debug_printtraffic': opts.debug_printtraffic,
76b1bd67 692 'prefer_ffmpeg': opts.prefer_ffmpeg,
7b0817e8 693 'include_ads': opts.include_ads,
04b4d394 694 'default_search': opts.default_search,
78895bd3 695 'dynamic_mpd': opts.dynamic_mpd,
5d3a0e79 696 'extractor_args': opts.extractor_args,
4919603f 697 'youtube_include_dash_manifest': opts.youtube_include_dash_manifest,
78895bd3 698 'youtube_include_hls_manifest': opts.youtube_include_hls_manifest,
62fec3b2 699 'encoding': opts.encoding,
057a5206 700 'extract_flat': opts.extract_flat,
d77ab8e2 701 'mark_watched': opts.mark_watched,
34c781a2 702 'merge_output_format': opts.merge_output_format,
df692c5a 703 'final_ext': final_ext,
4f026faf 704 'postprocessors': postprocessors,
6271f1ca 705 'fixup': opts.fixup,
be4a824d 706 'source_address': opts.source_address,
58b1f00d 707 'call_home': opts.call_home,
1cf376f5 708 'sleep_interval_requests': opts.sleep_interval_requests,
5f0d813d 709 'sleep_interval': opts.sleep_interval,
065bc354 710 'max_sleep_interval': opts.max_sleep_interval,
0c9df79e 711 'sleep_interval_subtitles': opts.sleep_interval_subtitles,
222516d9 712 'external_downloader': opts.external_downloader,
cfb56d1a 713 'list_thumbnails': opts.list_thumbnails,
c14e88f0 714 'playlist_items': opts.playlist_items,
881e6a1f 715 'xattr_set_filesize': opts.xattr_set_filesize,
347de493 716 'match_filter': match_filter,
7e5db8c9 717 'no_color': opts.no_color,
73fac4e9 718 'ffmpeg_location': opts.ffmpeg_location,
85729c51 719 'hls_prefer_native': opts.hls_prefer_native,
7d106a65 720 'hls_use_mpegts': opts.hls_use_mpegts,
310c2ed2 721 'hls_split_discontinuity': opts.hls_split_discontinuity,
46ee996e 722 'external_downloader_args': opts.external_downloader_args,
45016689 723 'postprocessor_args': opts.postprocessor_args,
91410c9b 724 'cn_verification_proxy': opts.cn_verification_proxy,
38cce791 725 'geo_verification_proxy': opts.geo_verification_proxy,
0a840f58
S
726 'geo_bypass': opts.geo_bypass,
727 'geo_bypass_country': opts.geo_bypass_country,
5f95927a 728 'geo_bypass_ip_block': opts.geo_bypass_ip_block,
0d1bb027 729 'warnings': warnings,
53ed7066 730 'compat_opts': compat_opts,
731 # just for deprecation check
0d1bb027 732 'autonumber': opts.autonumber or None,
733 'usetitle': opts.usetitle or None,
486fb179 734 'useid': opts.useid or None,
bdde425c 735 }
59ae15a5 736
bdde425c 737 with YoutubeDL(ydl_opts) as ydl:
e5813e53 738 actual_use = len(all_urls) or opts.load_info_filename
bdde425c 739
052421ff
PH
740 # Remove cache dir
741 if opts.rm_cachedir:
a0e07d31 742 ydl.cache.remove()
052421ff 743
e5813e53 744 # Update version
745 if opts.update_self:
746 # If updater returns True, exit. Required for windows
c19bc311 747 if run_update(ydl):
e5813e53 748 if actual_use:
6b027907 749 sys.exit('ERROR: The program must exit for the update to complete')
e5813e53 750 sys.exit()
751
bdde425c 752 # Maybe do nothing
e5813e53 753 if not actual_use:
7d4111ed 754 if opts.update_self or opts.rm_cachedir:
bdde425c 755 sys.exit()
59ae15a5 756
7d4111ed 757 ydl.warn_if_short_id(sys.argv[1:] if argv is None else argv)
adc0ae3c
PH
758 parser.error(
759 'You must provide at least one URL.\n'
7a5c1cfe 760 'Type yt-dlp --help to see a list of all options.')
7d4111ed 761
bdde425c 762 try:
1dcc4c0c 763 if opts.load_info_filename is not None:
590bc6f6 764 retcode = ydl.download_with_info_file(expand_path(opts.load_info_filename))
1dcc4c0c
JMF
765 else:
766 retcode = ydl.download(all_urls)
8b0d7497 767 except (MaxDownloadsReached, ExistingVideoReached, RejectedVideoReached):
768 ydl.to_screen('Aborting remaining downloads')
bdde425c 769 retcode = 101
59ae15a5 770
59ae15a5 771 sys.exit(retcode)
235b3ba4 772
a27b9e8b 773
b8ad4f02 774def main(argv=None):
59ae15a5 775 try:
b8ad4f02 776 _real_main(argv)
59ae15a5
PH
777 except DownloadError:
778 sys.exit(1)
779 except SameFileError:
a4bc4336 780 sys.exit('ERROR: fixed output name but more than one file to download')
59ae15a5 781 except KeyboardInterrupt:
a4bc4336 782 sys.exit('\nERROR: Interrupted by user')
cc3fa8d3 783 except BrokenPipeError:
784 # https://docs.python.org/3/library/signal.html#note-on-sigpipe
785 devnull = os.open(os.devnull, os.O_WRONLY)
786 os.dup2(devnull, sys.stdout.fileno())
787 sys.exit(r'\nERROR: {err}')
2bad0e5d 788
582be358 789
2bad0e5d 790__all__ = ['main', 'YoutubeDL', 'gen_extractors', 'list_extractors']