Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/__init__.py

   1 #!/usr/bin/env python
   2 # -*- coding: utf-8 -*-
   3
   4 from __future__ import unicode_literals
   5
   6 __license__ = 'Public Domain'
   7
   8 import codecs
   9 import io
  10 import os
  11 import random
  12 import sys
  13
  14
  15 from .options import (
  16     parseOpts,
  17 )
  18 from .compat import (
  19     compat_expanduser,
  20     compat_getpass,
  21     compat_print,
  22     workaround_optparse_bug9161,
  23 )
  24 from .utils import (
  25     DateRange,
  26     decodeOption,
  27     DEFAULT_OUTTMPL,
  28     DownloadError,
  29     match_filter_func,
  30     MaxDownloadsReached,
  31     preferredencoding,
  32     read_batch_urls,
  33     SameFileError,
  34     setproctitle,
  35     std_headers,
  36     write_string,
  37 )
  38 from .update import update_self
  39 from .downloader import (
  40     FileDownloader,
  41 )
  42 from .extractor import gen_extractors, list_extractors
  43 from .YoutubeDL import YoutubeDL
  44
  45
  46 def _real_main(argv=None):
  47     # Compatibility fixes for Windows
  48     if sys.platform == 'win32':
  49         # https://github.com/rg3/youtube-dl/issues/820
  50         codecs.register(lambda name: codecs.lookup('utf-8') if name == 'cp65001' else None)
  51
  52     workaround_optparse_bug9161()
  53
  54     setproctitle('youtube-dl')
  55
  56     parser, opts, args = parseOpts(argv)
  57
  58     # Set user agent
  59     if opts.user_agent is not None:
  60         std_headers['User-Agent'] = opts.user_agent
  61
  62     # Set referer
  63     if opts.referer is not None:
  64         std_headers['Referer'] = opts.referer
  65
  66     # Custom HTTP headers
  67     if opts.headers is not None:
  68         for h in opts.headers:
  69             if h.find(':', 1) < 0:
  70                 parser.error('wrong header formatting, it should be key:value, not "%s"' % h)
  71             key, value = h.split(':', 2)
  72             if opts.verbose:
  73                 write_string('[debug] Adding header from command line option %s:%s\n' % (key, value))
  74             std_headers[key] = value
  75
  76     # Dump user agent
  77     if opts.dump_user_agent:
  78         compat_print(std_headers['User-Agent'])
  79         sys.exit(0)
  80
  81     # Batch file verification
  82     batch_urls = []
  83     if opts.batchfile is not None:
  84         try:
  85             if opts.batchfile == '-':
  86                 batchfd = sys.stdin
  87             else:
  88                 batchfd = io.open(opts.batchfile, 'r', encoding='utf-8', errors='ignore')
  89             batch_urls = read_batch_urls(batchfd)
  90             if opts.verbose:
  91                 write_string('[debug] Batch file urls: ' + repr(batch_urls) + '\n')
  92         except IOError:
  93             sys.exit('ERROR: batch file could not be read')
  94     all_urls = batch_urls + args
  95     all_urls = [url.strip() for url in all_urls]
  96     _enc = preferredencoding()
  97     all_urls = [url.decode(_enc, 'ignore') if isinstance(url, bytes) else url for url in all_urls]
  98
  99     if opts.list_extractors:
 100         for ie in list_extractors(opts.age_limit):
 101             compat_print(ie.IE_NAME + (' (CURRENTLY BROKEN)' if not ie._WORKING else ''))
 102             matchedUrls = [url for url in all_urls if ie.suitable(url)]
 103             for mu in matchedUrls:
 104                 compat_print('  ' + mu)
 105         sys.exit(0)
 106     if opts.list_extractor_descriptions:
 107         for ie in list_extractors(opts.age_limit):
 108             if not ie._WORKING:
 109                 continue
 110             desc = getattr(ie, 'IE_DESC', ie.IE_NAME)
 111             if desc is False:
 112                 continue
 113             if hasattr(ie, 'SEARCH_KEY'):
 114                 _SEARCHES = ('cute kittens', 'slithering pythons', 'falling cat', 'angry poodle', 'purple fish', 'running tortoise', 'sleeping bunny', 'burping cow')
 115                 _COUNTS = ('', '5', '10', 'all')
 116                 desc += ' (Example: "%s%s:%s" )' % (ie.SEARCH_KEY, random.choice(_COUNTS), random.choice(_SEARCHES))
 117             compat_print(desc)
 118         sys.exit(0)
 119
 120     # Conflicting, missing and erroneous options
 121     if opts.usenetrc and (opts.username is not None or opts.password is not None):
 122         parser.error('using .netrc conflicts with giving username/password')
 123     if opts.password is not None and opts.username is None:
 124         parser.error('account username missing\n')
 125     if opts.outtmpl is not None and (opts.usetitle or opts.autonumber or opts.useid):
 126         parser.error('using output template conflicts with using title, video ID or auto number')
 127     if opts.usetitle and opts.useid:
 128         parser.error('using title conflicts with using video ID')
 129     if opts.username is not None and opts.password is None:
 130         opts.password = compat_getpass('Type account password and press [Return]: ')
 131     if opts.ratelimit is not None:
 132         numeric_limit = FileDownloader.parse_bytes(opts.ratelimit)
 133         if numeric_limit is None:
 134             parser.error('invalid rate limit specified')
 135         opts.ratelimit = numeric_limit
 136     if opts.min_filesize is not None:
 137         numeric_limit = FileDownloader.parse_bytes(opts.min_filesize)
 138         if numeric_limit is None:
 139             parser.error('invalid min_filesize specified')
 140         opts.min_filesize = numeric_limit
 141     if opts.max_filesize is not None:
 142         numeric_limit = FileDownloader.parse_bytes(opts.max_filesize)
 143         if numeric_limit is None:
 144             parser.error('invalid max_filesize specified')
 145         opts.max_filesize = numeric_limit
 146     if opts.retries is not None:
 147         if opts.retries in ('inf', 'infinite'):
 148             opts_retries = float('inf')
 149         else:
 150             try:
 151                 opts_retries = int(opts.retries)
 152             except (TypeError, ValueError):
 153                 parser.error('invalid retry count specified')
 154     if opts.buffersize is not None:
 155         numeric_buffersize = FileDownloader.parse_bytes(opts.buffersize)
 156         if numeric_buffersize is None:
 157             parser.error('invalid buffer size specified')
 158         opts.buffersize = numeric_buffersize
 159     if opts.playliststart <= 0:
 160         raise ValueError('Playlist start must be positive')
 161     if opts.playlistend not in (-1, None) and opts.playlistend < opts.playliststart:
 162         raise ValueError('Playlist end must be greater than playlist start')
 163     if opts.extractaudio:
 164         if opts.audioformat not in ['best', 'aac', 'mp3', 'm4a', 'opus', 'vorbis', 'wav']:
 165             parser.error('invalid audio format specified')
 166     if opts.audioquality:
 167         opts.audioquality = opts.audioquality.strip('k').strip('K')
 168         if not opts.audioquality.isdigit():
 169             parser.error('invalid audio quality specified')
 170     if opts.recodevideo is not None:
 171         if opts.recodevideo not in ['mp4', 'flv', 'webm', 'ogg', 'mkv']:
 172             parser.error('invalid video recode format specified')
 173     if opts.convertsubtitles is not None:
 174         if opts.convertsubtitles not in ['srt', 'vtt', 'ass']:
 175             parser.error('invalid subtitle format specified')
 176
 177     if opts.date is not None:
 178         date = DateRange.day(opts.date)
 179     else:
 180         date = DateRange(opts.dateafter, opts.datebefore)
 181
 182     # Do not download videos when there are audio-only formats
 183     if opts.extractaudio and not opts.keepvideo and opts.format is None:
 184         opts.format = 'bestaudio/best'
 185
 186     # --all-sub automatically sets --write-sub if --write-auto-sub is not given
 187     # this was the old behaviour if only --all-sub was given.
 188     if opts.allsubtitles and not opts.writeautomaticsub:
 189         opts.writesubtitles = True
 190
 191     if sys.version_info < (3,):
 192         # In Python 2, sys.argv is a bytestring (also note http://bugs.python.org/issue2128 for Windows systems)
 193         if opts.outtmpl is not None:
 194             opts.outtmpl = opts.outtmpl.decode(preferredencoding())
 195     outtmpl = ((opts.outtmpl is not None and opts.outtmpl) or
 196                (opts.format == '-1' and opts.usetitle and '%(title)s-%(id)s-%(format)s.%(ext)s') or
 197                (opts.format == '-1' and '%(id)s-%(format)s.%(ext)s') or
 198                (opts.usetitle and opts.autonumber and '%(autonumber)s-%(title)s-%(id)s.%(ext)s') or
 199                (opts.usetitle and '%(title)s-%(id)s.%(ext)s') or
 200                (opts.useid and '%(id)s.%(ext)s') or
 201                (opts.autonumber and '%(autonumber)s-%(id)s.%(ext)s') or
 202                DEFAULT_OUTTMPL)
 203     if not os.path.splitext(outtmpl)[1] and opts.extractaudio:
 204         parser.error('Cannot download a video and extract audio into the same'
 205                      ' file! Use "{0}.%(ext)s" instead of "{0}" as the output'
 206                      ' template'.format(outtmpl))
 207
 208     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
 209     any_printing = opts.print_json
 210     download_archive_fn = compat_expanduser(opts.download_archive) if opts.download_archive is not None else opts.download_archive
 211
 212     # PostProcessors
 213     postprocessors = []
 214     # Add the metadata pp first, the other pps will copy it
 215     if opts.addmetadata:
 216         postprocessors.append({'key': 'FFmpegMetadata'})
 217     if opts.extractaudio:
 218         postprocessors.append({
 219             'key': 'FFmpegExtractAudio',
 220             'preferredcodec': opts.audioformat,
 221             'preferredquality': opts.audioquality,
 222             'nopostoverwrites': opts.nopostoverwrites,
 223         })
 224     if opts.recodevideo:
 225         postprocessors.append({
 226             'key': 'FFmpegVideoConvertor',
 227             'preferedformat': opts.recodevideo,
 228         })
 229     if opts.convertsubtitles:
 230         postprocessors.append({
 231             'key': 'FFmpegSubtitlesConvertor',
 232             'format': opts.convertsubtitles,
 233         })
 234     if opts.embedsubtitles:
 235         postprocessors.append({
 236             'key': 'FFmpegEmbedSubtitle',
 237         })
 238     if opts.xattrs:
 239         postprocessors.append({'key': 'XAttrMetadata'})
 240     if opts.embedthumbnail:
 241         if not opts.addmetadata:
 242             postprocessors.append({'key': 'FFmpegAudioFix'})
 243         postprocessors.append({'key': 'AtomicParsley'})
 244     # Please keep ExecAfterDownload towards the bottom as it allows the user to modify the final file in any way.
 245     # So if the user is able to remove the file before your postprocessor runs it might cause a few problems.
 246     if opts.exec_cmd:
 247         postprocessors.append({
 248             'key': 'ExecAfterDownload',
 249             'verboseOutput': opts.verbose,
 250             'exec_cmd': opts.exec_cmd,
 251         })
 252     if opts.xattr_set_filesize:
 253         try:
 254             import xattr
 255             xattr  # Confuse flake8
 256         except ImportError:
 257             parser.error('setting filesize xattr requested but python-xattr is not available')
 258     match_filter = (
 259         None if opts.match_filter is None
 260         else match_filter_func(opts.match_filter))
 261
 262     ydl_opts = {
 263         'usenetrc': opts.usenetrc,
 264         'username': opts.username,
 265         'password': opts.password,
 266         'twofactor': opts.twofactor,
 267         'videopassword': opts.videopassword,
 268         'quiet': (opts.quiet or any_getting or any_printing),
 269         'no_warnings': opts.no_warnings,
 270         'forceurl': opts.geturl,
 271         'forcetitle': opts.gettitle,
 272         'forceid': opts.getid,
 273         'forcethumbnail': opts.getthumbnail,
 274         'forcedescription': opts.getdescription,
 275         'forceduration': opts.getduration,
 276         'forcefilename': opts.getfilename,
 277         'forceformat': opts.getformat,
 278         'forcejson': opts.dumpjson or opts.print_json,
 279         'dump_single_json': opts.dump_single_json,
 280         'simulate': opts.simulate or any_getting,
 281         'skip_download': opts.skip_download,
 282         'format': opts.format,
 283         'format_limit': opts.format_limit,
 284         'listformats': opts.listformats,
 285         'outtmpl': outtmpl,
 286         'autonumber_size': opts.autonumber_size,
 287         'restrictfilenames': opts.restrictfilenames,
 288         'ignoreerrors': opts.ignoreerrors,
 289         'ratelimit': opts.ratelimit,
 290         'nooverwrites': opts.nooverwrites,
 291         'retries': opts_retries,
 292         'buffersize': opts.buffersize,
 293         'noresizebuffer': opts.noresizebuffer,
 294         'continuedl': opts.continue_dl,
 295         'noprogress': opts.noprogress,
 296         'progress_with_newline': opts.progress_with_newline,
 297         'playliststart': opts.playliststart,
 298         'playlistend': opts.playlistend,
 299         'playlistreverse': opts.playlist_reverse,
 300         'noplaylist': opts.noplaylist,
 301         'logtostderr': opts.outtmpl == '-',
 302         'consoletitle': opts.consoletitle,
 303         'nopart': opts.nopart,
 304         'updatetime': opts.updatetime,
 305         'writedescription': opts.writedescription,
 306         'writeannotations': opts.writeannotations,
 307         'writeinfojson': opts.writeinfojson,
 308         'writethumbnail': opts.writethumbnail,
 309         'write_all_thumbnails': opts.write_all_thumbnails,
 310         'writesubtitles': opts.writesubtitles,
 311         'writeautomaticsub': opts.writeautomaticsub,
 312         'allsubtitles': opts.allsubtitles,
 313         'listsubtitles': opts.listsubtitles,
 314         'subtitlesformat': opts.subtitlesformat,
 315         'subtitleslangs': opts.subtitleslangs,
 316         'matchtitle': decodeOption(opts.matchtitle),
 317         'rejecttitle': decodeOption(opts.rejecttitle),
 318         'max_downloads': opts.max_downloads,
 319         'prefer_free_formats': opts.prefer_free_formats,
 320         'verbose': opts.verbose,
 321         'dump_intermediate_pages': opts.dump_intermediate_pages,
 322         'write_pages': opts.write_pages,
 323         'test': opts.test,
 324         'keepvideo': opts.keepvideo,
 325         'min_filesize': opts.min_filesize,
 326         'max_filesize': opts.max_filesize,
 327         'min_views': opts.min_views,
 328         'max_views': opts.max_views,
 329         'daterange': date,
 330         'cachedir': opts.cachedir,
 331         'youtube_print_sig_code': opts.youtube_print_sig_code,
 332         'age_limit': opts.age_limit,
 333         'download_archive': download_archive_fn,
 334         'cookiefile': opts.cookiefile,
 335         'nocheckcertificate': opts.no_check_certificate,
 336         'prefer_insecure': opts.prefer_insecure,
 337         'proxy': opts.proxy,
 338         'socket_timeout': opts.socket_timeout,
 339         'bidi_workaround': opts.bidi_workaround,
 340         'debug_printtraffic': opts.debug_printtraffic,
 341         'prefer_ffmpeg': opts.prefer_ffmpeg,
 342         'include_ads': opts.include_ads,
 343         'default_search': opts.default_search,
 344         'youtube_include_dash_manifest': opts.youtube_include_dash_manifest,
 345         'encoding': opts.encoding,
 346         'exec_cmd': opts.exec_cmd,
 347         'extract_flat': opts.extract_flat,
 348         'merge_output_format': opts.merge_output_format,
 349         'postprocessors': postprocessors,
 350         'fixup': opts.fixup,
 351         'source_address': opts.source_address,
 352         'call_home': opts.call_home,
 353         'sleep_interval': opts.sleep_interval,
 354         'external_downloader': opts.external_downloader,
 355         'list_thumbnails': opts.list_thumbnails,
 356         'playlist_items': opts.playlist_items,
 357         'xattr_set_filesize': opts.xattr_set_filesize,
 358         'match_filter': match_filter,
 359         'no_color': opts.no_color,
 360         'ffmpeg_location': opts.ffmpeg_location,
 361         'hls_prefer_native': opts.hls_prefer_native,
 362     }
 363
 364     with YoutubeDL(ydl_opts) as ydl:
 365         # Update version
 366         if opts.update_self:
 367             update_self(ydl.to_screen, opts.verbose)
 368
 369         # Remove cache dir
 370         if opts.rm_cachedir:
 371             ydl.cache.remove()
 372
 373         # Maybe do nothing
 374         if (len(all_urls) < 1) and (opts.load_info_filename is None):
 375             if opts.update_self or opts.rm_cachedir:
 376                 sys.exit()
 377
 378             ydl.warn_if_short_id(sys.argv[1:] if argv is None else argv)
 379             parser.error(
 380                 'You must provide at least one URL.\n'
 381                 'Type youtube-dl --help to see a list of all options.')
 382
 383         try:
 384             if opts.load_info_filename is not None:
 385                 retcode = ydl.download_with_info_file(opts.load_info_filename)
 386             else:
 387                 retcode = ydl.download(all_urls)
 388         except MaxDownloadsReached:
 389             ydl.to_screen('--max-download limit reached, aborting.')
 390             retcode = 101
 391
 392     sys.exit(retcode)
 393
 394
 395 def main(argv=None):
 396     try:
 397         _real_main(argv)
 398     except DownloadError:
 399         sys.exit(1)
 400     except SameFileError:
 401         sys.exit('ERROR: fixed output name but more than one file to download')
 402     except KeyboardInterrupt:
 403         sys.exit('\nERROR: Interrupted by user')
 404
 405 __all__ = ['main', 'YoutubeDL', 'gen_extractors', 'list_extractors']