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
 174     if opts.date is not None:
 175         date = DateRange.day(opts.date)
 176     else:
 177         date = DateRange(opts.dateafter, opts.datebefore)
 178
 179     # Do not download videos when there are audio-only formats
 180     if opts.extractaudio and not opts.keepvideo and opts.format is None:
 181         opts.format = 'bestaudio/best'
 182
 183     # --all-sub automatically sets --write-sub if --write-auto-sub is not given
 184     # this was the old behaviour if only --all-sub was given.
 185     if opts.allsubtitles and not opts.writeautomaticsub:
 186         opts.writesubtitles = True
 187
 188     if sys.version_info < (3,):
 189         # In Python 2, sys.argv is a bytestring (also note http://bugs.python.org/issue2128 for Windows systems)
 190         if opts.outtmpl is not None:
 191             opts.outtmpl = opts.outtmpl.decode(preferredencoding())
 192     outtmpl = ((opts.outtmpl is not None and opts.outtmpl) or
 193                (opts.format == '-1' and opts.usetitle and '%(title)s-%(id)s-%(format)s.%(ext)s') or
 194                (opts.format == '-1' and '%(id)s-%(format)s.%(ext)s') or
 195                (opts.usetitle and opts.autonumber and '%(autonumber)s-%(title)s-%(id)s.%(ext)s') or
 196                (opts.usetitle and '%(title)s-%(id)s.%(ext)s') or
 197                (opts.useid and '%(id)s.%(ext)s') or
 198                (opts.autonumber and '%(autonumber)s-%(id)s.%(ext)s') or
 199                DEFAULT_OUTTMPL)
 200     if not os.path.splitext(outtmpl)[1] and opts.extractaudio:
 201         parser.error('Cannot download a video and extract audio into the same'
 202                      ' file! Use "{0}.%(ext)s" instead of "{0}" as the output'
 203                      ' template'.format(outtmpl))
 204
 205     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
 206     any_printing = opts.print_json
 207     download_archive_fn = compat_expanduser(opts.download_archive) if opts.download_archive is not None else opts.download_archive
 208
 209     # PostProcessors
 210     postprocessors = []
 211     # Add the metadata pp first, the other pps will copy it
 212     if opts.addmetadata:
 213         postprocessors.append({'key': 'FFmpegMetadata'})
 214     if opts.extractaudio:
 215         postprocessors.append({
 216             'key': 'FFmpegExtractAudio',
 217             'preferredcodec': opts.audioformat,
 218             'preferredquality': opts.audioquality,
 219             'nopostoverwrites': opts.nopostoverwrites,
 220         })
 221     if opts.recodevideo:
 222         postprocessors.append({
 223             'key': 'FFmpegVideoConvertor',
 224             'preferedformat': opts.recodevideo,
 225         })
 226     if opts.embedsubtitles:
 227         postprocessors.append({
 228             'key': 'FFmpegEmbedSubtitle',
 229         })
 230     if opts.xattrs:
 231         postprocessors.append({'key': 'XAttrMetadata'})
 232     if opts.embedthumbnail:
 233         if not opts.addmetadata:
 234             postprocessors.append({'key': 'FFmpegAudioFix'})
 235         postprocessors.append({'key': 'AtomicParsley'})
 236     # Please keep ExecAfterDownload towards the bottom as it allows the user to modify the final file in any way.
 237     # So if the user is able to remove the file before your postprocessor runs it might cause a few problems.
 238     if opts.exec_cmd:
 239         postprocessors.append({
 240             'key': 'ExecAfterDownload',
 241             'verboseOutput': opts.verbose,
 242             'exec_cmd': opts.exec_cmd,
 243         })
 244     if opts.xattr_set_filesize:
 245         try:
 246             import xattr
 247             xattr  # Confuse flake8
 248         except ImportError:
 249             parser.error('setting filesize xattr requested but python-xattr is not available')
 250     match_filter = (
 251         None if opts.match_filter is None
 252         else match_filter_func(opts.match_filter))
 253
 254     ydl_opts = {
 255         'usenetrc': opts.usenetrc,
 256         'username': opts.username,
 257         'password': opts.password,
 258         'twofactor': opts.twofactor,
 259         'videopassword': opts.videopassword,
 260         'quiet': (opts.quiet or any_getting or any_printing),
 261         'no_warnings': opts.no_warnings,
 262         'forceurl': opts.geturl,
 263         'forcetitle': opts.gettitle,
 264         'forceid': opts.getid,
 265         'forcethumbnail': opts.getthumbnail,
 266         'forcedescription': opts.getdescription,
 267         'forceduration': opts.getduration,
 268         'forcefilename': opts.getfilename,
 269         'forceformat': opts.getformat,
 270         'forcejson': opts.dumpjson or opts.print_json,
 271         'dump_single_json': opts.dump_single_json,
 272         'simulate': opts.simulate or any_getting,
 273         'skip_download': opts.skip_download,
 274         'format': opts.format,
 275         'format_limit': opts.format_limit,
 276         'listformats': opts.listformats,
 277         'outtmpl': outtmpl,
 278         'autonumber_size': opts.autonumber_size,
 279         'restrictfilenames': opts.restrictfilenames,
 280         'ignoreerrors': opts.ignoreerrors,
 281         'ratelimit': opts.ratelimit,
 282         'nooverwrites': opts.nooverwrites,
 283         'retries': opts_retries,
 284         'buffersize': opts.buffersize,
 285         'noresizebuffer': opts.noresizebuffer,
 286         'continuedl': opts.continue_dl,
 287         'noprogress': opts.noprogress,
 288         'progress_with_newline': opts.progress_with_newline,
 289         'playliststart': opts.playliststart,
 290         'playlistend': opts.playlistend,
 291         'playlistreverse': opts.playlist_reverse,
 292         'noplaylist': opts.noplaylist,
 293         'logtostderr': opts.outtmpl == '-',
 294         'consoletitle': opts.consoletitle,
 295         'nopart': opts.nopart,
 296         'updatetime': opts.updatetime,
 297         'writedescription': opts.writedescription,
 298         'writeannotations': opts.writeannotations,
 299         'writeinfojson': opts.writeinfojson,
 300         'writethumbnail': opts.writethumbnail,
 301         'write_all_thumbnails': opts.write_all_thumbnails,
 302         'writesubtitles': opts.writesubtitles,
 303         'writeautomaticsub': opts.writeautomaticsub,
 304         'allsubtitles': opts.allsubtitles,
 305         'listsubtitles': opts.listsubtitles,
 306         'subtitlesformat': opts.subtitlesformat,
 307         'subtitleslangs': opts.subtitleslangs,
 308         'matchtitle': decodeOption(opts.matchtitle),
 309         'rejecttitle': decodeOption(opts.rejecttitle),
 310         'max_downloads': opts.max_downloads,
 311         'prefer_free_formats': opts.prefer_free_formats,
 312         'verbose': opts.verbose,
 313         'dump_intermediate_pages': opts.dump_intermediate_pages,
 314         'write_pages': opts.write_pages,
 315         'test': opts.test,
 316         'keepvideo': opts.keepvideo,
 317         'min_filesize': opts.min_filesize,
 318         'max_filesize': opts.max_filesize,
 319         'min_views': opts.min_views,
 320         'max_views': opts.max_views,
 321         'daterange': date,
 322         'cachedir': opts.cachedir,
 323         'youtube_print_sig_code': opts.youtube_print_sig_code,
 324         'age_limit': opts.age_limit,
 325         'download_archive': download_archive_fn,
 326         'cookiefile': opts.cookiefile,
 327         'nocheckcertificate': opts.no_check_certificate,
 328         'prefer_insecure': opts.prefer_insecure,
 329         'proxy': opts.proxy,
 330         'socket_timeout': opts.socket_timeout,
 331         'bidi_workaround': opts.bidi_workaround,
 332         'debug_printtraffic': opts.debug_printtraffic,
 333         'prefer_ffmpeg': opts.prefer_ffmpeg,
 334         'include_ads': opts.include_ads,
 335         'default_search': opts.default_search,
 336         'youtube_include_dash_manifest': opts.youtube_include_dash_manifest,
 337         'encoding': opts.encoding,
 338         'exec_cmd': opts.exec_cmd,
 339         'extract_flat': opts.extract_flat,
 340         'merge_output_format': opts.merge_output_format,
 341         'postprocessors': postprocessors,
 342         'fixup': opts.fixup,
 343         'source_address': opts.source_address,
 344         'call_home': opts.call_home,
 345         'sleep_interval': opts.sleep_interval,
 346         'external_downloader': opts.external_downloader,
 347         'list_thumbnails': opts.list_thumbnails,
 348         'playlist_items': opts.playlist_items,
 349         'xattr_set_filesize': opts.xattr_set_filesize,
 350         'match_filter': match_filter,
 351         'no_color': opts.no_color,
 352         'ffmpeg_location': opts.ffmpeg_location,
 353         'hls_prefer_native': opts.hls_prefer_native,
 354     }
 355
 356     with YoutubeDL(ydl_opts) as ydl:
 357         # Update version
 358         if opts.update_self:
 359             update_self(ydl.to_screen, opts.verbose)
 360
 361         # Remove cache dir
 362         if opts.rm_cachedir:
 363             ydl.cache.remove()
 364
 365         # Maybe do nothing
 366         if (len(all_urls) < 1) and (opts.load_info_filename is None):
 367             if opts.update_self or opts.rm_cachedir:
 368                 sys.exit()
 369
 370             ydl.warn_if_short_id(sys.argv[1:] if argv is None else argv)
 371             parser.error(
 372                 'You must provide at least one URL.\n'
 373                 'Type youtube-dl --help to see a list of all options.')
 374
 375         try:
 376             if opts.load_info_filename is not None:
 377                 retcode = ydl.download_with_info_file(opts.load_info_filename)
 378             else:
 379                 retcode = ydl.download(all_urls)
 380         except MaxDownloadsReached:
 381             ydl.to_screen('--max-download limit reached, aborting.')
 382             retcode = 101
 383
 384     sys.exit(retcode)
 385
 386
 387 def main(argv=None):
 388     try:
 389         _real_main(argv)
 390     except DownloadError:
 391         sys.exit(1)
 392     except SameFileError:
 393         sys.exit('ERROR: fixed output name but more than one file to download')
 394     except KeyboardInterrupt:
 395         sys.exit('\nERROR: Interrupted by user')
 396
 397 __all__ = ['main', 'YoutubeDL', 'gen_extractors', 'list_extractors']