from ..swfinterp import SWFInterpreter
from ..compat import (
compat_chr,
+ compat_kwargs,
compat_parse_qs,
compat_urllib_parse_unquote,
compat_urllib_parse_unquote_plus,
return True
+ def _download_webpage(self, *args, **kwargs):
+ kwargs.setdefault('query', {})['disable_polymer'] = 'true'
+ return super(YoutubeBaseInfoExtractor, self)._download_webpage(
+ *args, **compat_kwargs(kwargs))
+
def _real_initialize(self):
if self._downloader is None:
return
(?:(?:(?:(?:\w+\.)?[yY][oO][uU][tT][uU][bB][eE](?:-nocookie)?\.com/|
(?:www\.)?deturl\.com/www\.youtube\.com/|
(?:www\.)?pwnyoutube\.com/|
+ (?:www\.)?hooktube\.com/|
(?:www\.)?yourepeat\.com/|
tube\.majestyc\.net/|
youtube\.googleapis\.com/) # the various hostnames, with wildcard subdomains
},
},
# video_info is None (https://github.com/rg3/youtube-dl/issues/4421)
+ # YouTube Red ad is not captured for creator
{
'url': '__2ABJjxzNo',
'info_dict': {
'Skipping DASH manifest',
],
},
+ {
+ # The following content has been identified by the YouTube community
+ # as inappropriate or offensive to some audiences.
+ 'url': 'https://www.youtube.com/watch?v=6SJNVb0GnPI',
+ 'info_dict': {
+ 'id': '6SJNVb0GnPI',
+ 'ext': 'mp4',
+ 'title': 'Race Differences in Intelligence',
+ 'description': 'md5:5d161533167390427a1f8ee89a1fc6f1',
+ 'duration': 965,
+ 'upload_date': '20140124',
+ 'uploader': 'New Century Foundation',
+ 'uploader_id': 'UCEJYpZGqgUob0zVVEaLhvVg',
+ 'uploader_url': r're:https?://(?:www\.)?youtube\.com/channel/UCEJYpZGqgUob0zVVEaLhvVg',
+ 'license': 'Standard YouTube License',
+ 'view_count': int,
+ },
+ 'params': {
+ 'skip_download': True,
+ },
+ },
{
# itag 212
'url': '1t24XAntNCY',
sub_lang_list[sub_lang] = sub_formats
return sub_lang_list
+ def make_captions(sub_url, sub_langs):
+ parsed_sub_url = compat_urllib_parse_urlparse(sub_url)
+ caption_qs = compat_parse_qs(parsed_sub_url.query)
+ captions = {}
+ for sub_lang in sub_langs:
+ sub_formats = []
+ for ext in self._SUBTITLE_FORMATS:
+ caption_qs.update({
+ 'tlang': [sub_lang],
+ 'fmt': [ext],
+ })
+ sub_url = compat_urlparse.urlunparse(parsed_sub_url._replace(
+ query=compat_urllib_parse_urlencode(caption_qs, True)))
+ sub_formats.append({
+ 'url': sub_url,
+ 'ext': ext,
+ })
+ captions[sub_lang] = sub_formats
+ return captions
+
+ # New captions format as of 22.06.2017
+ player_response = args.get('player_response')
+ if player_response and isinstance(player_response, compat_str):
+ player_response = self._parse_json(
+ player_response, video_id, fatal=False)
+ if player_response:
+ renderer = player_response['captions']['playerCaptionsTracklistRenderer']
+ base_url = renderer['captionTracks'][0]['baseUrl']
+ sub_lang_list = []
+ for lang in renderer['translationLanguages']:
+ lang_code = lang.get('languageCode')
+ if lang_code:
+ sub_lang_list.append(lang_code)
+ return make_captions(base_url, sub_lang_list)
+
# Some videos don't provide ttsurl but rather caption_tracks and
# caption_translation_languages (e.g. 20LmZk1hakA)
+ # Does not used anymore as of 22.06.2017
caption_tracks = args['caption_tracks']
caption_translation_languages = args['caption_translation_languages']
caption_url = compat_parse_qs(caption_tracks.split(',')[0])['u'][0]
- parsed_caption_url = compat_urllib_parse_urlparse(caption_url)
- caption_qs = compat_parse_qs(parsed_caption_url.query)
-
- sub_lang_list = {}
+ sub_lang_list = []
for lang in caption_translation_languages.split(','):
lang_qs = compat_parse_qs(compat_urllib_parse_unquote_plus(lang))
sub_lang = lang_qs.get('lc', [None])[0]
- if not sub_lang:
- continue
- sub_formats = []
- for ext in self._SUBTITLE_FORMATS:
- caption_qs.update({
- 'tlang': [sub_lang],
- 'fmt': [ext],
- })
- sub_url = compat_urlparse.urlunparse(parsed_caption_url._replace(
- query=compat_urllib_parse_urlencode(caption_qs, True)))
- sub_formats.append({
- 'url': sub_url,
- 'ext': ext,
- })
- sub_lang_list[sub_lang] = sub_formats
- return sub_lang_list
+ if sub_lang:
+ sub_lang_list.append(sub_lang)
+ return make_captions(caption_url, sub_lang_list)
# An extractor error can be raise by the download process if there are
# no automatic captions but there are subtitles
- except (KeyError, ExtractorError):
+ except (KeyError, IndexError, ExtractorError):
self._downloader.report_warning(err_msg)
return {}
playback_url, video_id, 'Marking watched',
'Unable to mark watched', fatal=False)
+ @staticmethod
+ def _extract_urls(webpage):
+ # Embedded YouTube player
+ entries = [
+ unescapeHTML(mobj.group('url'))
+ for mobj in re.finditer(r'''(?x)
+ (?:
+ <iframe[^>]+?src=|
+ data-video-url=|
+ <embed[^>]+?src=|
+ embedSWF\(?:\s*|
+ <object[^>]+data=|
+ new\s+SWFObject\(
+ )
+ (["\'])
+ (?P<url>(?:https?:)?//(?:www\.)?youtube(?:-nocookie)?\.com/
+ (?:embed|v|p)/.+?)
+ \1''', webpage)]
+
+ # lazyYT YouTube embed
+ entries.extend(list(map(
+ unescapeHTML,
+ re.findall(r'class="lazyYT" data-youtube-id="([^"]+)"', webpage))))
+
+ # Wordpress "YouTube Video Importer" plugin
+ matches = re.findall(r'''(?x)<div[^>]+
+ class=(?P<q1>[\'"])[^\'"]*\byvii_single_video_player\b[^\'"]*(?P=q1)[^>]+
+ data-video_id=(?P<q2>[\'"])([^\'"]+)(?P=q2)''', webpage)
+ entries.extend(m[-1] for m in matches)
+
+ return entries
+
+ @staticmethod
+ def _extract_url(webpage):
+ urls = YoutubeIE._extract_urls(webpage)
+ return urls[0] if urls else None
+
@classmethod
def extract_id(cls, url):
mobj = re.match(cls._VALID_URL, url, re.VERBOSE)
if dash_mpd and dash_mpd[0] not in dash_mpds:
dash_mpds.append(dash_mpd[0])
+ is_live = None
+ view_count = None
+
+ def extract_view_count(v_info):
+ return int_or_none(try_get(v_info, lambda x: x['view_count'][0]))
+
# Get video info
embed_webpage = None
- is_live = None
if re.search(r'player-age-gate-content">', video_webpage) is not None:
age_gate = True
# We simulate the access to the video from www.youtube.com/v/{video_id}
continue
get_video_info = compat_parse_qs(video_info_webpage)
add_dash_mpd(get_video_info)
+ if view_count is None:
+ view_count = extract_view_count(get_video_info)
if not video_info:
video_info = get_video_info
if 'token' in get_video_info:
return self.playlist_result(entries, video_id, video_title, video_description)
self.to_screen('Downloading just video %s because of --no-playlist' % video_id)
- if 'view_count' in video_info:
- view_count = int(video_info['view_count'][0])
- else:
- view_count = None
+ if view_count is None:
+ view_count = extract_view_count(video_info)
# Check for "rental" videos
if 'ypc_video_rental_bar_text' in video_info and 'author' not in video_info:
if not upload_date:
upload_date = self._search_regex(
[r'(?s)id="eow-date.*?>(.*?)</span>',
- r'id="watch-uploader-info".*?>.*?(?:Published|Uploaded|Streamed live|Started) on (.+?)</strong>'],
+ r'(?:id="watch-uploader-info".*?>.*?|["\']simpleText["\']\s*:\s*["\'])(?:Published|Uploaded|Streamed live|Started) on (.+?)[<"\']'],
video_webpage, 'upload date', default=None)
- if upload_date:
- upload_date = ' '.join(re.sub(r'[/,-]', r' ', mobj.group(1)).split())
upload_date = unified_strdate(upload_date)
video_license = self._html_search_regex(
video_webpage, 'license', default=None)
m_music = re.search(
- r'<h4[^>]+class="title"[^>]*>\s*Music\s*</h4>\s*<ul[^>]*>\s*<li>(?P<title>.+?) by (?P<creator>.+?)(?:\(.+?\))?</li',
+ r'''(?x)
+ <h4[^>]+class="title"[^>]*>\s*Music\s*</h4>\s*
+ <ul[^>]*>\s*
+ <li>(?P<title>.+?)
+ by (?P<creator>.+?)
+ (?:
+ \(.+?\)|
+ <a[^>]*
+ (?:
+ \bhref=["\']/red[^>]*>| # drop possible
+ >\s*Listen ad-free with YouTube Red # YouTube Red ad
+ )
+ .*?
+ )?</li
+ ''',
video_webpage)
if m_music:
video_alt_title = remove_quotes(unescapeHTML(m_music.group('title')))
|
(%(playlist_id)s)
)""" % {'playlist_id': YoutubeBaseInfoExtractor._PLAYLIST_ID_RE}
- _TEMPLATE_URL = 'https://www.youtube.com/playlist?list=%s&disable_polymer=true'
+ _TEMPLATE_URL = 'https://www.youtube.com/playlist?list=%s'
_VIDEO_RE = r'href="\s*/watch\?v=(?P<id>[0-9A-Za-z_-]{11})&[^"]*?index=(?P<index>\d+)(?:[^>]+>(?P<title>[^<]+))?'
IE_NAME = 'youtube:playlist'
_TESTS = [{