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