]>
Commit | Line | Data |
---|---|---|
235b3ba4 PH |
1 | #!/usr/bin/env python |
2 | # -*- coding: utf-8 -*- | |
3 | ||
a4bc4336 PH |
4 | from __future__ import unicode_literals |
5 | ||
235b3ba4 | 6 | __license__ = 'Public Domain' |
235b3ba4 | 7 | |
0d94f247 | 8 | import codecs |
8f563f32 | 9 | import io |
235b3ba4 | 10 | import os |
0f818663 | 11 | import random |
235b3ba4 | 12 | import sys |
235b3ba4 | 13 | |
c496ca96 | 14 | |
2daabe49 PH |
15 | from .options import ( |
16 | parseOpts, | |
17 | ) | |
8c25f81b | 18 | from .compat import ( |
4644ac55 | 19 | compat_expanduser, |
e68301af | 20 | compat_getpass, |
a4fd0415 | 21 | compat_print, |
e07e9313 | 22 | workaround_optparse_bug9161, |
8c25f81b PH |
23 | ) |
24 | from .utils import ( | |
a4fd0415 | 25 | DateRange, |
acd69589 | 26 | DEFAULT_OUTTMPL, |
a4fd0415 | 27 | decodeOption, |
a4fd0415 | 28 | DownloadError, |
a4fd0415 | 29 | MaxDownloadsReached, |
a4fd0415 | 30 | preferredencoding, |
62e609ab | 31 | read_batch_urls, |
a4fd0415 | 32 | SameFileError, |
e3946f98 | 33 | setproctitle, |
a4fd0415 PH |
34 | std_headers, |
35 | write_string, | |
a4fd0415 | 36 | ) |
d5ed35b6 | 37 | from .update import update_self |
92a86f4c | 38 | from .downloader import ( |
a4fd0415 PH |
39 | FileDownloader, |
40 | ) | |
0824c28c | 41 | from .extractor import gen_extractors |
8222d8de | 42 | from .YoutubeDL import YoutubeDL |
a4fd0415 | 43 | |
235b3ba4 | 44 | |
b8ad4f02 | 45 | def _real_main(argv=None): |
0d94f247 PH |
46 | # Compatibility fixes for Windows |
47 | if sys.platform == 'win32': | |
48 | # https://github.com/rg3/youtube-dl/issues/820 | |
49 | codecs.register(lambda name: codecs.lookup('utf-8') if name == 'cp65001' else None) | |
50 | ||
e07e9313 PH |
51 | workaround_optparse_bug9161() |
52 | ||
a4bc4336 | 53 | setproctitle('youtube-dl') |
e3946f98 | 54 | |
b8ad4f02 | 55 | parser, opts, args = parseOpts(argv) |
59ae15a5 | 56 | |
59ae15a5 PH |
57 | # Set user agent |
58 | if opts.user_agent is not None: | |
59 | std_headers['User-Agent'] = opts.user_agent | |
1865ed31 | 60 | |
28535652 BH |
61 | # Set referer |
62 | if opts.referer is not None: | |
63 | std_headers['Referer'] = opts.referer | |
59ae15a5 | 64 | |
410afb20 AA |
65 | # Custom HTTP headers |
66 | if opts.headers is not None: | |
67 | for h in opts.headers: | |
68 | if h.find(':', 1) < 0: | |
8bcc8756 | 69 | parser.error('wrong header formatting, it should be key:value, not "%s"' % h) |
410afb20 AA |
70 | key, value = h.split(':', 2) |
71 | if opts.verbose: | |
8bcc8756 | 72 | write_string('[debug] Adding header from command line option %s:%s\n' % (key, value)) |
410afb20 AA |
73 | std_headers[key] = value |
74 | ||
59ae15a5 PH |
75 | # Dump user agent |
76 | if opts.dump_user_agent: | |
93eb15c5 | 77 | compat_print(std_headers['User-Agent']) |
59ae15a5 PH |
78 | sys.exit(0) |
79 | ||
80 | # Batch file verification | |
62e609ab | 81 | batch_urls = [] |
59ae15a5 PH |
82 | if opts.batchfile is not None: |
83 | try: | |
84 | if opts.batchfile == '-': | |
85 | batchfd = sys.stdin | |
86 | else: | |
62e609ab PH |
87 | batchfd = io.open(opts.batchfile, 'r', encoding='utf-8', errors='ignore') |
88 | batch_urls = read_batch_urls(batchfd) | |
05afc96b | 89 | if opts.verbose: |
a4bc4336 | 90 | write_string('[debug] Batch file urls: ' + repr(batch_urls) + '\n') |
59ae15a5 | 91 | except IOError: |
a4bc4336 | 92 | sys.exit('ERROR: batch file could not be read') |
62e609ab | 93 | all_urls = batch_urls + args |
59ae15a5 | 94 | all_urls = [url.strip() for url in all_urls] |
c774b3c6 | 95 | _enc = preferredencoding() |
41292a38 | 96 | all_urls = [url.decode(_enc, 'ignore') if isinstance(url, bytes) else url for url in all_urls] |
59ae15a5 | 97 | |
59ae15a5 PH |
98 | extractors = gen_extractors() |
99 | ||
100 | if opts.list_extractors: | |
7dba9cd0 | 101 | for ie in sorted(extractors, key=lambda ie: ie.IE_NAME.lower()): |
93eb15c5 | 102 | compat_print(ie.IE_NAME + (' (CURRENTLY BROKEN)' if not ie._WORKING else '')) |
1a2c3c0f | 103 | matchedUrls = [url for url in all_urls if ie.suitable(url)] |
59ae15a5 | 104 | for mu in matchedUrls: |
a4bc4336 | 105 | compat_print(' ' + mu) |
59ae15a5 | 106 | sys.exit(0) |
0f818663 PH |
107 | if opts.list_extractor_descriptions: |
108 | for ie in sorted(extractors, key=lambda ie: ie.IE_NAME.lower()): | |
109 | if not ie._WORKING: | |
110 | continue | |
111 | desc = getattr(ie, 'IE_DESC', ie.IE_NAME) | |
15870e90 PH |
112 | if desc is False: |
113 | continue | |
0f818663 | 114 | if hasattr(ie, 'SEARCH_KEY'): |
a4bc4336 PH |
115 | _SEARCHES = ('cute kittens', 'slithering pythons', 'falling cat', 'angry poodle', 'purple fish', 'running tortoise', 'sleeping bunny') |
116 | _COUNTS = ('', '5', '10', 'all') | |
117 | desc += ' (Example: "%s%s:%s" )' % (ie.SEARCH_KEY, random.choice(_COUNTS), random.choice(_SEARCHES)) | |
0f818663 PH |
118 | compat_print(desc) |
119 | sys.exit(0) | |
120 | ||
59ae15a5 PH |
121 | # Conflicting, missing and erroneous options |
122 | if opts.usenetrc and (opts.username is not None or opts.password is not None): | |
a4bc4336 | 123 | parser.error('using .netrc conflicts with giving username/password') |
59ae15a5 | 124 | if opts.password is not None and opts.username is None: |
a4bc4336 | 125 | parser.error('account username missing\n') |
59ae15a5 | 126 | if opts.outtmpl is not None and (opts.usetitle or opts.autonumber or opts.useid): |
a4bc4336 | 127 | parser.error('using output template conflicts with using title, video ID or auto number') |
59ae15a5 | 128 | if opts.usetitle and opts.useid: |
a4bc4336 | 129 | parser.error('using title conflicts with using video ID') |
59ae15a5 | 130 | if opts.username is not None and opts.password is None: |
a4bc4336 | 131 | opts.password = compat_getpass('Type account password and press [Return]: ') |
59ae15a5 PH |
132 | if opts.ratelimit is not None: |
133 | numeric_limit = FileDownloader.parse_bytes(opts.ratelimit) | |
134 | if numeric_limit is None: | |
a4bc4336 | 135 | parser.error('invalid rate limit specified') |
59ae15a5 | 136 | opts.ratelimit = numeric_limit |
9e982f9e JC |
137 | if opts.min_filesize is not None: |
138 | numeric_limit = FileDownloader.parse_bytes(opts.min_filesize) | |
139 | if numeric_limit is None: | |
a4bc4336 | 140 | parser.error('invalid min_filesize specified') |
9e982f9e JC |
141 | opts.min_filesize = numeric_limit |
142 | if opts.max_filesize is not None: | |
143 | numeric_limit = FileDownloader.parse_bytes(opts.max_filesize) | |
144 | if numeric_limit is None: | |
a4bc4336 | 145 | parser.error('invalid max_filesize specified') |
9e982f9e | 146 | opts.max_filesize = numeric_limit |
59ae15a5 PH |
147 | if opts.retries is not None: |
148 | try: | |
149 | opts.retries = int(opts.retries) | |
dca08720 | 150 | except (TypeError, ValueError): |
a4bc4336 | 151 | parser.error('invalid retry count specified') |
59ae15a5 PH |
152 | if opts.buffersize is not None: |
153 | numeric_buffersize = FileDownloader.parse_bytes(opts.buffersize) | |
154 | if numeric_buffersize is None: | |
a4bc4336 | 155 | parser.error('invalid buffer size specified') |
59ae15a5 | 156 | opts.buffersize = numeric_buffersize |
a19fd00c | 157 | if opts.playliststart <= 0: |
a4bc4336 | 158 | raise ValueError('Playlist start must be positive') |
a19fd00c | 159 | if opts.playlistend not in (-1, None) and opts.playlistend < opts.playliststart: |
a4bc4336 | 160 | raise ValueError('Playlist end must be greater than playlist start') |
59ae15a5 | 161 | if opts.extractaudio: |
510e6f6d | 162 | if opts.audioformat not in ['best', 'aac', 'mp3', 'm4a', 'opus', 'vorbis', 'wav']: |
a4bc4336 | 163 | parser.error('invalid audio format specified') |
59ae15a5 PH |
164 | if opts.audioquality: |
165 | opts.audioquality = opts.audioquality.strip('k').strip('K') | |
166 | if not opts.audioquality.isdigit(): | |
a4bc4336 | 167 | parser.error('invalid audio quality specified') |
7851b379 | 168 | if opts.recodevideo is not None: |
b7d73595 | 169 | if opts.recodevideo not in ['mp4', 'flv', 'webm', 'ogg', 'mkv']: |
a4bc4336 | 170 | parser.error('invalid video recode format specified') |
bd558525 JMF |
171 | if opts.date is not None: |
172 | date = DateRange.day(opts.date) | |
173 | else: | |
174 | date = DateRange(opts.dateafter, opts.datebefore) | |
59ae15a5 | 175 | |
de3ef3ed PH |
176 | # Do not download videos when there are audio-only formats |
177 | if opts.extractaudio and not opts.keepvideo and opts.format is None: | |
178 | opts.format = 'bestaudio/best' | |
179 | ||
0b7f3118 JMF |
180 | # --all-sub automatically sets --write-sub if --write-auto-sub is not given |
181 | # this was the old behaviour if only --all-sub was given. | |
b74e86f4 | 182 | if opts.allsubtitles and not opts.writeautomaticsub: |
0b7f3118 JMF |
183 | opts.writesubtitles = True |
184 | ||
5cb9c312 PH |
185 | if sys.version_info < (3,): |
186 | # In Python 2, sys.argv is a bytestring (also note http://bugs.python.org/issue2128 for Windows systems) | |
0be41ec2 PH |
187 | if opts.outtmpl is not None: |
188 | opts.outtmpl = opts.outtmpl.decode(preferredencoding()) | |
5f6a1245 | 189 | outtmpl = ((opts.outtmpl is not None and opts.outtmpl) |
8bcc8756 JW |
190 | or (opts.format == '-1' and opts.usetitle and '%(title)s-%(id)s-%(format)s.%(ext)s') |
191 | or (opts.format == '-1' and '%(id)s-%(format)s.%(ext)s') | |
192 | or (opts.usetitle and opts.autonumber and '%(autonumber)s-%(title)s-%(id)s.%(ext)s') | |
193 | or (opts.usetitle and '%(title)s-%(id)s.%(ext)s') | |
194 | or (opts.useid and '%(id)s.%(ext)s') | |
195 | or (opts.autonumber and '%(autonumber)s-%(id)s.%(ext)s') | |
196 | or DEFAULT_OUTTMPL) | |
dca02c80 | 197 | if not os.path.splitext(outtmpl)[1] and opts.extractaudio: |
a4bc4336 PH |
198 | parser.error('Cannot download a video and extract audio into the same' |
199 | ' file! Use "{0}.%(ext)s" instead of "{0}" as the output' | |
200 | ' template'.format(outtmpl)) | |
29c7a63d | 201 | |
63e0be34 | 202 | any_printing = 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 |
4644ac55 | 203 | download_archive_fn = compat_expanduser(opts.download_archive) if opts.download_archive is not None else opts.download_archive |
525ef922 | 204 | |
4f026faf PH |
205 | # PostProcessors |
206 | postprocessors = [] | |
207 | # Add the metadata pp first, the other pps will copy it | |
208 | if opts.addmetadata: | |
209 | postprocessors.append({'key': 'FFmpegMetadata'}) | |
210 | if opts.extractaudio: | |
211 | postprocessors.append({ | |
212 | 'key': 'FFmpegExtractAudio', | |
213 | 'preferredcodec': opts.audioformat, | |
214 | 'preferredquality': opts.audioquality, | |
215 | 'nopostoverwrites': opts.nopostoverwrites, | |
216 | }) | |
217 | if opts.recodevideo: | |
218 | postprocessors.append({ | |
219 | 'key': 'FFmpegVideoConvertor', | |
220 | 'preferedformat': opts.recodevideo, | |
221 | }) | |
222 | if opts.embedsubtitles: | |
223 | postprocessors.append({ | |
224 | 'key': 'FFmpegEmbedSubtitle', | |
225 | 'subtitlesformat': opts.subtitlesformat, | |
226 | }) | |
227 | if opts.xattrs: | |
228 | postprocessors.append({'key': 'XAttrMetadata'}) | |
229 | if opts.embedthumbnail: | |
230 | if not opts.addmetadata: | |
231 | postprocessors.append({'key': 'FFmpegAudioFix'}) | |
232 | postprocessors.append({'key': 'AtomicParsley'}) | |
233 | # Please keep ExecAfterDownload towards the bottom as it allows the user to modify the final file in any way. | |
234 | # So if the user is able to remove the file before your postprocessor runs it might cause a few problems. | |
235 | if opts.exec_cmd: | |
236 | postprocessors.append({ | |
237 | 'key': 'ExecAfterDownload', | |
238 | 'verboseOutput': opts.verbose, | |
239 | 'exec_cmd': opts.exec_cmd, | |
240 | }) | |
241 | ||
bdde425c | 242 | ydl_opts = { |
59ae15a5 PH |
243 | 'usenetrc': opts.usenetrc, |
244 | 'username': opts.username, | |
245 | 'password': opts.password, | |
83317f69 | 246 | 'twofactor': opts.twofactor, |
c6c19746 | 247 | 'videopassword': opts.videopassword, |
525ef922 | 248 | 'quiet': (opts.quiet or any_printing), |
ad8915b7 | 249 | 'no_warnings': opts.no_warnings, |
59ae15a5 PH |
250 | 'forceurl': opts.geturl, |
251 | 'forcetitle': opts.gettitle, | |
1a2adf3f | 252 | 'forceid': opts.getid, |
59ae15a5 PH |
253 | 'forcethumbnail': opts.getthumbnail, |
254 | 'forcedescription': opts.getdescription, | |
525ef922 | 255 | 'forceduration': opts.getduration, |
59ae15a5 PH |
256 | 'forcefilename': opts.getfilename, |
257 | 'forceformat': opts.getformat, | |
9d153818 | 258 | 'forcejson': opts.dumpjson, |
63e0be34 | 259 | 'dump_single_json': opts.dump_single_json, |
1bdeb7be JMF |
260 | 'simulate': opts.simulate or any_printing, |
261 | 'skip_download': opts.skip_download, | |
59ae15a5 PH |
262 | 'format': opts.format, |
263 | 'format_limit': opts.format_limit, | |
264 | 'listformats': opts.listformats, | |
5cb9c312 | 265 | 'outtmpl': outtmpl, |
213c31ae | 266 | 'autonumber_size': opts.autonumber_size, |
59ae15a5 PH |
267 | 'restrictfilenames': opts.restrictfilenames, |
268 | 'ignoreerrors': opts.ignoreerrors, | |
269 | 'ratelimit': opts.ratelimit, | |
270 | 'nooverwrites': opts.nooverwrites, | |
271 | 'retries': opts.retries, | |
272 | 'buffersize': opts.buffersize, | |
273 | 'noresizebuffer': opts.noresizebuffer, | |
274 | 'continuedl': opts.continue_dl, | |
275 | 'noprogress': opts.noprogress, | |
5717d91a | 276 | 'progress_with_newline': opts.progress_with_newline, |
59ae15a5 PH |
277 | 'playliststart': opts.playliststart, |
278 | 'playlistend': opts.playlistend, | |
ff815fe6 | 279 | 'playlistreverse': opts.playlist_reverse, |
47192f92 | 280 | 'noplaylist': opts.noplaylist, |
59ae15a5 PH |
281 | 'logtostderr': opts.outtmpl == '-', |
282 | 'consoletitle': opts.consoletitle, | |
283 | 'nopart': opts.nopart, | |
284 | 'updatetime': opts.updatetime, | |
285 | 'writedescription': opts.writedescription, | |
1fb07d10 | 286 | 'writeannotations': opts.writeannotations, |
59ae15a5 | 287 | 'writeinfojson': opts.writeinfojson, |
11d9224e | 288 | 'writethumbnail': opts.writethumbnail, |
59ae15a5 | 289 | 'writesubtitles': opts.writesubtitles, |
b004821f | 290 | 'writeautomaticsub': opts.writeautomaticsub, |
ae608b80 | 291 | 'allsubtitles': opts.allsubtitles, |
2a4093ea | 292 | 'listsubtitles': opts.listsubtitles, |
9e62bc44 | 293 | 'subtitlesformat': opts.subtitlesformat, |
d6e203b3 | 294 | 'subtitleslangs': opts.subtitleslangs, |
8271226a PH |
295 | 'matchtitle': decodeOption(opts.matchtitle), |
296 | 'rejecttitle': decodeOption(opts.rejecttitle), | |
59ae15a5 PH |
297 | 'max_downloads': opts.max_downloads, |
298 | 'prefer_free_formats': opts.prefer_free_formats, | |
299 | 'verbose': opts.verbose, | |
855703e5 | 300 | 'dump_intermediate_pages': opts.dump_intermediate_pages, |
d41e6efc | 301 | 'write_pages': opts.write_pages, |
8d5d3a5d | 302 | 'test': opts.test, |
7851b379 | 303 | 'keepvideo': opts.keepvideo, |
9e982f9e | 304 | 'min_filesize': opts.min_filesize, |
bd558525 | 305 | 'max_filesize': opts.max_filesize, |
5fe18bdb PH |
306 | 'min_views': opts.min_views, |
307 | 'max_views': opts.max_views, | |
11d9224e | 308 | 'daterange': date, |
7f747732 | 309 | 'cachedir': opts.cachedir, |
f8061589 | 310 | 'youtube_print_sig_code': opts.youtube_print_sig_code, |
8dbe9899 | 311 | 'age_limit': opts.age_limit, |
17093b83 | 312 | 'download_archive': download_archive_fn, |
dca08720 PH |
313 | 'cookiefile': opts.cookiefile, |
314 | 'nocheckcertificate': opts.no_check_certificate, | |
7e8c0af0 | 315 | 'prefer_insecure': opts.prefer_insecure, |
c2e52508 | 316 | 'proxy': opts.proxy, |
6ad14cab | 317 | 'socket_timeout': opts.socket_timeout, |
0783b09b | 318 | 'bidi_workaround': opts.bidi_workaround, |
a0ddb8a2 | 319 | 'debug_printtraffic': opts.debug_printtraffic, |
76b1bd67 | 320 | 'prefer_ffmpeg': opts.prefer_ffmpeg, |
7b0817e8 | 321 | 'include_ads': opts.include_ads, |
04b4d394 | 322 | 'default_search': opts.default_search, |
4919603f | 323 | 'youtube_include_dash_manifest': opts.youtube_include_dash_manifest, |
62fec3b2 | 324 | 'encoding': opts.encoding, |
8d31fa3c | 325 | 'exec_cmd': opts.exec_cmd, |
057a5206 | 326 | 'extract_flat': opts.extract_flat, |
4f026faf | 327 | 'postprocessors': postprocessors, |
bdde425c | 328 | } |
59ae15a5 | 329 | |
bdde425c | 330 | with YoutubeDL(ydl_opts) as ydl: |
bdde425c PH |
331 | # Update version |
332 | if opts.update_self: | |
333 | update_self(ydl.to_screen, opts.verbose) | |
334 | ||
052421ff PH |
335 | # Remove cache dir |
336 | if opts.rm_cachedir: | |
a0e07d31 | 337 | ydl.cache.remove() |
052421ff | 338 | |
bdde425c | 339 | # Maybe do nothing |
1dcc4c0c | 340 | if (len(all_urls) < 1) and (opts.load_info_filename is None): |
7d4111ed | 341 | if opts.update_self or opts.rm_cachedir: |
bdde425c | 342 | sys.exit() |
59ae15a5 | 343 | |
7d4111ed PH |
344 | ydl.warn_if_short_id(sys.argv[1:] if argv is None else argv) |
345 | parser.error('you must provide at least one URL') | |
346 | ||
bdde425c | 347 | try: |
1dcc4c0c JMF |
348 | if opts.load_info_filename is not None: |
349 | retcode = ydl.download_with_info_file(opts.load_info_filename) | |
350 | else: | |
351 | retcode = ydl.download(all_urls) | |
bdde425c | 352 | except MaxDownloadsReached: |
a4bc4336 | 353 | ydl.to_screen('--max-download limit reached, aborting.') |
bdde425c | 354 | retcode = 101 |
59ae15a5 | 355 | |
59ae15a5 | 356 | sys.exit(retcode) |
235b3ba4 | 357 | |
a27b9e8b | 358 | |
b8ad4f02 | 359 | def main(argv=None): |
59ae15a5 | 360 | try: |
b8ad4f02 | 361 | _real_main(argv) |
59ae15a5 PH |
362 | except DownloadError: |
363 | sys.exit(1) | |
364 | except SameFileError: | |
a4bc4336 | 365 | sys.exit('ERROR: fixed output name but more than one file to download') |
59ae15a5 | 366 | except KeyboardInterrupt: |
a4bc4336 | 367 | sys.exit('\nERROR: Interrupted by user') |