You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

453 lines
18 KiB

11 years ago
11 years ago
11 years ago
11 years ago
11 years ago
11 years ago
11 years ago
11 years ago
11 years ago
10 years ago
  1. #!/usr/bin/env python
  2. # coding: utf-8
  3. from __future__ import unicode_literals
  4. __license__ = 'Public Domain'
  5. import codecs
  6. import io
  7. import os
  8. import random
  9. import sys
  10. from .options import (
  11. parseOpts,
  12. )
  13. from .compat import (
  14. compat_expanduser,
  15. compat_getpass,
  16. compat_shlex_split,
  17. workaround_optparse_bug9161,
  18. )
  19. from .utils import (
  20. DateRange,
  21. decodeOption,
  22. DEFAULT_OUTTMPL,
  23. DownloadError,
  24. match_filter_func,
  25. MaxDownloadsReached,
  26. preferredencoding,
  27. read_batch_urls,
  28. SameFileError,
  29. setproctitle,
  30. std_headers,
  31. write_string,
  32. render_table,
  33. )
  34. from .update import update_self
  35. from .downloader import (
  36. FileDownloader,
  37. )
  38. from .extractor import gen_extractors, list_extractors
  39. from .extractor.adobepass import MSO_INFO
  40. from .YoutubeDL import YoutubeDL
  41. def _real_main(argv=None):
  42. # Compatibility fixes for Windows
  43. if sys.platform == 'win32':
  44. # https://github.com/rg3/youtube-dl/issues/820
  45. codecs.register(lambda name: codecs.lookup('utf-8') if name == 'cp65001' else None)
  46. workaround_optparse_bug9161()
  47. setproctitle('youtube-dl')
  48. parser, opts, args = parseOpts(argv)
  49. # Set user agent
  50. if opts.user_agent is not None:
  51. std_headers['User-Agent'] = opts.user_agent
  52. # Set referer
  53. if opts.referer is not None:
  54. std_headers['Referer'] = opts.referer
  55. # Custom HTTP headers
  56. if opts.headers is not None:
  57. for h in opts.headers:
  58. if ':' not in h:
  59. parser.error('wrong header formatting, it should be key:value, not "%s"' % h)
  60. key, value = h.split(':', 1)
  61. if opts.verbose:
  62. write_string('[debug] Adding header from command line option %s:%s\n' % (key, value))
  63. std_headers[key] = value
  64. # Dump user agent
  65. if opts.dump_user_agent:
  66. write_string(std_headers['User-Agent'] + '\n', out=sys.stdout)
  67. sys.exit(0)
  68. # Batch file verification
  69. batch_urls = []
  70. if opts.batchfile is not None:
  71. try:
  72. if opts.batchfile == '-':
  73. batchfd = sys.stdin
  74. else:
  75. batchfd = io.open(
  76. compat_expanduser(opts.batchfile),
  77. 'r', encoding='utf-8', errors='ignore')
  78. batch_urls = read_batch_urls(batchfd)
  79. if opts.verbose:
  80. write_string('[debug] Batch file urls: ' + repr(batch_urls) + '\n')
  81. except IOError:
  82. sys.exit('ERROR: batch file could not be read')
  83. all_urls = batch_urls + args
  84. all_urls = [url.strip() for url in all_urls]
  85. _enc = preferredencoding()
  86. all_urls = [url.decode(_enc, 'ignore') if isinstance(url, bytes) else url for url in all_urls]
  87. if opts.list_extractors:
  88. for ie in list_extractors(opts.age_limit):
  89. write_string(ie.IE_NAME + (' (CURRENTLY BROKEN)' if not ie._WORKING else '') + '\n', out=sys.stdout)
  90. matchedUrls = [url for url in all_urls if ie.suitable(url)]
  91. for mu in matchedUrls:
  92. write_string(' ' + mu + '\n', out=sys.stdout)
  93. sys.exit(0)
  94. if opts.list_extractor_descriptions:
  95. for ie in list_extractors(opts.age_limit):
  96. if not ie._WORKING:
  97. continue
  98. desc = getattr(ie, 'IE_DESC', ie.IE_NAME)
  99. if desc is False:
  100. continue
  101. if hasattr(ie, 'SEARCH_KEY'):
  102. _SEARCHES = ('cute kittens', 'slithering pythons', 'falling cat', 'angry poodle', 'purple fish', 'running tortoise', 'sleeping bunny', 'burping cow')
  103. _COUNTS = ('', '5', '10', 'all')
  104. desc += ' (Example: "%s%s:%s" )' % (ie.SEARCH_KEY, random.choice(_COUNTS), random.choice(_SEARCHES))
  105. write_string(desc + '\n', out=sys.stdout)
  106. sys.exit(0)
  107. if opts.ap_list_mso:
  108. table = [[mso_id, mso_info['name']] for mso_id, mso_info in MSO_INFO.items()]
  109. write_string('Supported TV Providers:\n' + render_table(['mso', 'mso name'], table) + '\n', out=sys.stdout)
  110. sys.exit(0)
  111. # Conflicting, missing and erroneous options
  112. if opts.usenetrc and (opts.username is not None or opts.password is not None):
  113. parser.error('using .netrc conflicts with giving username/password')
  114. if opts.password is not None and opts.username is None:
  115. parser.error('account username missing\n')
  116. if opts.ap_password is not None and opts.ap_username is None:
  117. parser.error('TV Provider account username missing\n')
  118. if opts.outtmpl is not None and (opts.usetitle or opts.autonumber or opts.useid):
  119. parser.error('using output template conflicts with using title, video ID or auto number')
  120. if opts.usetitle and opts.useid:
  121. parser.error('using title conflicts with using video ID')
  122. if opts.username is not None and opts.password is None:
  123. opts.password = compat_getpass('Type account password and press [Return]: ')
  124. if opts.ap_username is not None and opts.ap_password is None:
  125. opts.ap_password = compat_getpass('Type TV provider account password and press [Return]: ')
  126. if opts.ratelimit is not None:
  127. numeric_limit = FileDownloader.parse_bytes(opts.ratelimit)
  128. if numeric_limit is None:
  129. parser.error('invalid rate limit specified')
  130. opts.ratelimit = numeric_limit
  131. if opts.min_filesize is not None:
  132. numeric_limit = FileDownloader.parse_bytes(opts.min_filesize)
  133. if numeric_limit is None:
  134. parser.error('invalid min_filesize specified')
  135. opts.min_filesize = numeric_limit
  136. if opts.max_filesize is not None:
  137. numeric_limit = FileDownloader.parse_bytes(opts.max_filesize)
  138. if numeric_limit is None:
  139. parser.error('invalid max_filesize specified')
  140. opts.max_filesize = numeric_limit
  141. if opts.sleep_interval is not None:
  142. if opts.sleep_interval < 0:
  143. parser.error('sleep interval must be positive or 0')
  144. if opts.max_sleep_interval is not None:
  145. if opts.max_sleep_interval < 0:
  146. parser.error('max sleep interval must be positive or 0')
  147. if opts.max_sleep_interval < opts.sleep_interval:
  148. parser.error('max sleep interval must be greater than or equal to min sleep interval')
  149. else:
  150. opts.max_sleep_interval = opts.sleep_interval
  151. if opts.ap_mso and opts.ap_mso not in MSO_INFO:
  152. parser.error('Unsupported TV Provider, use --ap-list-mso to get a list of supported TV Providers')
  153. def parse_retries(retries):
  154. if retries in ('inf', 'infinite'):
  155. parsed_retries = float('inf')
  156. else:
  157. try:
  158. parsed_retries = int(retries)
  159. except (TypeError, ValueError):
  160. parser.error('invalid retry count specified')
  161. return parsed_retries
  162. if opts.retries is not None:
  163. opts.retries = parse_retries(opts.retries)
  164. if opts.fragment_retries is not None:
  165. opts.fragment_retries = parse_retries(opts.fragment_retries)
  166. if opts.buffersize is not None:
  167. numeric_buffersize = FileDownloader.parse_bytes(opts.buffersize)
  168. if numeric_buffersize is None:
  169. parser.error('invalid buffer size specified')
  170. opts.buffersize = numeric_buffersize
  171. if opts.playliststart <= 0:
  172. raise ValueError('Playlist start must be positive')
  173. if opts.playlistend not in (-1, None) and opts.playlistend < opts.playliststart:
  174. raise ValueError('Playlist end must be greater than playlist start')
  175. if opts.extractaudio:
  176. if opts.audioformat not in ['best', 'aac', 'mp3', 'm4a', 'opus', 'vorbis', 'wav']:
  177. parser.error('invalid audio format specified')
  178. if opts.audioquality:
  179. opts.audioquality = opts.audioquality.strip('k').strip('K')
  180. if not opts.audioquality.isdigit():
  181. parser.error('invalid audio quality specified')
  182. if opts.recodevideo is not None:
  183. if opts.recodevideo not in ['mp4', 'flv', 'webm', 'ogg', 'mkv', 'avi']:
  184. parser.error('invalid video recode format specified')
  185. if opts.convertsubtitles is not None:
  186. if opts.convertsubtitles not in ['srt', 'vtt', 'ass']:
  187. parser.error('invalid subtitle format specified')
  188. if opts.date is not None:
  189. date = DateRange.day(opts.date)
  190. else:
  191. date = DateRange(opts.dateafter, opts.datebefore)
  192. # Do not download videos when there are audio-only formats
  193. if opts.extractaudio and not opts.keepvideo and opts.format is None:
  194. opts.format = 'bestaudio/best'
  195. # --all-sub automatically sets --write-sub if --write-auto-sub is not given
  196. # this was the old behaviour if only --all-sub was given.
  197. if opts.allsubtitles and not opts.writeautomaticsub:
  198. opts.writesubtitles = True
  199. outtmpl = ((opts.outtmpl is not None and opts.outtmpl) or
  200. (opts.format == '-1' and opts.usetitle and '%(title)s-%(id)s-%(format)s.%(ext)s') or
  201. (opts.format == '-1' and '%(id)s-%(format)s.%(ext)s') or
  202. (opts.usetitle and opts.autonumber and '%(autonumber)s-%(title)s-%(id)s.%(ext)s') or
  203. (opts.usetitle and '%(title)s-%(id)s.%(ext)s') or
  204. (opts.useid and '%(id)s.%(ext)s') or
  205. (opts.autonumber and '%(autonumber)s-%(id)s.%(ext)s') or
  206. DEFAULT_OUTTMPL)
  207. if not os.path.splitext(outtmpl)[1] and opts.extractaudio:
  208. parser.error('Cannot download a video and extract audio into the same'
  209. ' file! Use "{0}.%(ext)s" instead of "{0}" as the output'
  210. ' template'.format(outtmpl))
  211. any_getting = 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
  212. any_printing = opts.print_json
  213. download_archive_fn = compat_expanduser(opts.download_archive) if opts.download_archive is not None else opts.download_archive
  214. # PostProcessors
  215. postprocessors = []
  216. # Add the metadata pp first, the other pps will copy it
  217. if opts.metafromtitle:
  218. postprocessors.append({
  219. 'key': 'MetadataFromTitle',
  220. 'titleformat': opts.metafromtitle
  221. })
  222. if opts.addmetadata:
  223. postprocessors.append({'key': 'FFmpegMetadata'})
  224. if opts.extractaudio:
  225. postprocessors.append({
  226. 'key': 'FFmpegExtractAudio',
  227. 'preferredcodec': opts.audioformat,
  228. 'preferredquality': opts.audioquality,
  229. 'nopostoverwrites': opts.nopostoverwrites,
  230. })
  231. if opts.recodevideo:
  232. postprocessors.append({
  233. 'key': 'FFmpegVideoConvertor',
  234. 'preferedformat': opts.recodevideo,
  235. })
  236. if opts.convertsubtitles:
  237. postprocessors.append({
  238. 'key': 'FFmpegSubtitlesConvertor',
  239. 'format': opts.convertsubtitles,
  240. })
  241. if opts.embedsubtitles:
  242. postprocessors.append({
  243. 'key': 'FFmpegEmbedSubtitle',
  244. })
  245. if opts.embedthumbnail:
  246. already_have_thumbnail = opts.writethumbnail or opts.write_all_thumbnails
  247. postprocessors.append({
  248. 'key': 'EmbedThumbnail',
  249. 'already_have_thumbnail': already_have_thumbnail
  250. })
  251. if not already_have_thumbnail:
  252. opts.writethumbnail = True
  253. # XAttrMetadataPP should be run after post-processors that may change file
  254. # contents
  255. if opts.xattrs:
  256. postprocessors.append({'key': 'XAttrMetadata'})
  257. # Please keep ExecAfterDownload towards the bottom as it allows the user to modify the final file in any way.
  258. # So if the user is able to remove the file before your postprocessor runs it might cause a few problems.
  259. if opts.exec_cmd:
  260. postprocessors.append({
  261. 'key': 'ExecAfterDownload',
  262. 'exec_cmd': opts.exec_cmd,
  263. })
  264. external_downloader_args = None
  265. if opts.external_downloader_args:
  266. external_downloader_args = compat_shlex_split(opts.external_downloader_args)
  267. postprocessor_args = None
  268. if opts.postprocessor_args:
  269. postprocessor_args = compat_shlex_split(opts.postprocessor_args)
  270. match_filter = (
  271. None if opts.match_filter is None
  272. else match_filter_func(opts.match_filter))
  273. ydl_opts = {
  274. 'usenetrc': opts.usenetrc,
  275. 'username': opts.username,
  276. 'password': opts.password,
  277. 'twofactor': opts.twofactor,
  278. 'videopassword': opts.videopassword,
  279. 'ap_mso': opts.ap_mso,
  280. 'ap_username': opts.ap_username,
  281. 'ap_password': opts.ap_password,
  282. 'quiet': (opts.quiet or any_getting or any_printing),
  283. 'no_warnings': opts.no_warnings,
  284. 'forceurl': opts.geturl,
  285. 'forcetitle': opts.gettitle,
  286. 'forceid': opts.getid,
  287. 'forcethumbnail': opts.getthumbnail,
  288. 'forcedescription': opts.getdescription,
  289. 'forceduration': opts.getduration,
  290. 'forcefilename': opts.getfilename,
  291. 'forceformat': opts.getformat,
  292. 'forcejson': opts.dumpjson or opts.print_json,
  293. 'dump_single_json': opts.dump_single_json,
  294. 'simulate': opts.simulate or any_getting,
  295. 'skip_download': opts.skip_download,
  296. 'format': opts.format,
  297. 'listformats': opts.listformats,
  298. 'outtmpl': outtmpl,
  299. 'autonumber_size': opts.autonumber_size,
  300. 'restrictfilenames': opts.restrictfilenames,
  301. 'ignoreerrors': opts.ignoreerrors,
  302. 'force_generic_extractor': opts.force_generic_extractor,
  303. 'ratelimit': opts.ratelimit,
  304. 'nooverwrites': opts.nooverwrites,
  305. 'retries': opts.retries,
  306. 'fragment_retries': opts.fragment_retries,
  307. 'skip_unavailable_fragments': opts.skip_unavailable_fragments,
  308. 'buffersize': opts.buffersize,
  309. 'noresizebuffer': opts.noresizebuffer,
  310. 'continuedl': opts.continue_dl,
  311. 'noprogress': opts.noprogress,
  312. 'progress_with_newline': opts.progress_with_newline,
  313. 'playliststart': opts.playliststart,
  314. 'playlistend': opts.playlistend,
  315. 'playlistreverse': opts.playlist_reverse,
  316. 'noplaylist': opts.noplaylist,
  317. 'logtostderr': opts.outtmpl == '-',
  318. 'consoletitle': opts.consoletitle,
  319. 'nopart': opts.nopart,
  320. 'updatetime': opts.updatetime,
  321. 'writedescription': opts.writedescription,
  322. 'writeannotations': opts.writeannotations,
  323. 'writeinfojson': opts.writeinfojson,
  324. 'writethumbnail': opts.writethumbnail,
  325. 'write_all_thumbnails': opts.write_all_thumbnails,
  326. 'writesubtitles': opts.writesubtitles,
  327. 'writeautomaticsub': opts.writeautomaticsub,
  328. 'allsubtitles': opts.allsubtitles,
  329. 'listsubtitles': opts.listsubtitles,
  330. 'subtitlesformat': opts.subtitlesformat,
  331. 'subtitleslangs': opts.subtitleslangs,
  332. 'matchtitle': decodeOption(opts.matchtitle),
  333. 'rejecttitle': decodeOption(opts.rejecttitle),
  334. 'max_downloads': opts.max_downloads,
  335. 'prefer_free_formats': opts.prefer_free_formats,
  336. 'verbose': opts.verbose,
  337. 'dump_intermediate_pages': opts.dump_intermediate_pages,
  338. 'write_pages': opts.write_pages,
  339. 'test': opts.test,
  340. 'keepvideo': opts.keepvideo,
  341. 'min_filesize': opts.min_filesize,
  342. 'max_filesize': opts.max_filesize,
  343. 'min_views': opts.min_views,
  344. 'max_views': opts.max_views,
  345. 'daterange': date,
  346. 'cachedir': opts.cachedir,
  347. 'youtube_print_sig_code': opts.youtube_print_sig_code,
  348. 'age_limit': opts.age_limit,
  349. 'download_archive': download_archive_fn,
  350. 'cookiefile': opts.cookiefile,
  351. 'nocheckcertificate': opts.no_check_certificate,
  352. 'prefer_insecure': opts.prefer_insecure,
  353. 'proxy': opts.proxy,
  354. 'socket_timeout': opts.socket_timeout,
  355. 'bidi_workaround': opts.bidi_workaround,
  356. 'debug_printtraffic': opts.debug_printtraffic,
  357. 'prefer_ffmpeg': opts.prefer_ffmpeg,
  358. 'include_ads': opts.include_ads,
  359. 'default_search': opts.default_search,
  360. 'youtube_include_dash_manifest': opts.youtube_include_dash_manifest,
  361. 'encoding': opts.encoding,
  362. 'extract_flat': opts.extract_flat,
  363. 'mark_watched': opts.mark_watched,
  364. 'merge_output_format': opts.merge_output_format,
  365. 'postprocessors': postprocessors,
  366. 'fixup': opts.fixup,
  367. 'source_address': opts.source_address,
  368. 'call_home': opts.call_home,
  369. 'sleep_interval': opts.sleep_interval,
  370. 'max_sleep_interval': opts.max_sleep_interval,
  371. 'external_downloader': opts.external_downloader,
  372. 'list_thumbnails': opts.list_thumbnails,
  373. 'playlist_items': opts.playlist_items,
  374. 'xattr_set_filesize': opts.xattr_set_filesize,
  375. 'match_filter': match_filter,
  376. 'no_color': opts.no_color,
  377. 'ffmpeg_location': opts.ffmpeg_location,
  378. 'hls_prefer_native': opts.hls_prefer_native,
  379. 'hls_use_mpegts': opts.hls_use_mpegts,
  380. 'external_downloader_args': external_downloader_args,
  381. 'postprocessor_args': postprocessor_args,
  382. 'cn_verification_proxy': opts.cn_verification_proxy,
  383. 'geo_verification_proxy': opts.geo_verification_proxy,
  384. }
  385. with YoutubeDL(ydl_opts) as ydl:
  386. # Update version
  387. if opts.update_self:
  388. update_self(ydl.to_screen, opts.verbose, ydl._opener)
  389. # Remove cache dir
  390. if opts.rm_cachedir:
  391. ydl.cache.remove()
  392. # Maybe do nothing
  393. if (len(all_urls) < 1) and (opts.load_info_filename is None):
  394. if opts.update_self or opts.rm_cachedir:
  395. sys.exit()
  396. ydl.warn_if_short_id(sys.argv[1:] if argv is None else argv)
  397. parser.error(
  398. 'You must provide at least one URL.\n'
  399. 'Type youtube-dl --help to see a list of all options.')
  400. try:
  401. if opts.load_info_filename is not None:
  402. retcode = ydl.download_with_info_file(compat_expanduser(opts.load_info_filename))
  403. else:
  404. retcode = ydl.download(all_urls)
  405. except MaxDownloadsReached:
  406. ydl.to_screen('--max-download limit reached, aborting.')
  407. retcode = 101
  408. sys.exit(retcode)
  409. def main(argv=None):
  410. try:
  411. _real_main(argv)
  412. except DownloadError:
  413. sys.exit(1)
  414. except SameFileError:
  415. sys.exit('ERROR: fixed output name but more than one file to download')
  416. except KeyboardInterrupt:
  417. sys.exit('\nERROR: Interrupted by user')
  418. __all__ = ['main', 'YoutubeDL', 'gen_extractors', 'list_extractors']