X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;ds=sidebyside;f=youtube_dl%2Fextractor%2Fyoutube.py;h=6bdea1c44bc5201d2c0b9a6d01279b9414668cbe;hb=9a2dc4f7ac308725afd651b1f0f92291d1906496;hp=3c8f7f7a2a65720e72c9efeb0fbf017202902de5;hpb=ad3bc6acd5d6724875b9fa59f9b5cdb9b904ec91;p=youtube-dl
diff --git a/youtube_dl/extractor/youtube.py b/youtube_dl/extractor/youtube.py
index 3c8f7f7a2..6bdea1c44 100644
--- a/youtube_dl/extractor/youtube.py
+++ b/youtube_dl/extractor/youtube.py
@@ -223,6 +223,8 @@ class YoutubeIE(YoutubeBaseInfoExtractor, SubtitlesInfoExtractor):
'246': {'ext': 'webm', 'height': 480, 'format_note': 'DASH video', 'acodec': 'none', 'preference': -40},
'247': {'ext': 'webm', 'height': 720, 'format_note': 'DASH video', 'acodec': 'none', 'preference': -40},
'248': {'ext': 'webm', 'height': 1080, 'format_note': 'DASH video', 'acodec': 'none', 'preference': -40},
+ '271': {'ext': 'webm', 'height': 1440, 'format_note': 'DASH video', 'acodec': 'none', 'preference': -40},
+ '272': {'ext': 'webm', 'height': 2160, 'format_note': 'DASH video', 'acodec': 'none', 'preference': -40},
# Dash webm audio
'171': {'ext': 'webm', 'vcodec': 'none', 'format_note': 'DASH audio', 'abr': 48, 'preference': -50},
@@ -439,7 +441,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor, SubtitlesInfoExtractor):
def _parse_sig_js(self, jscode):
funcname = self._search_regex(
- r'signature=([a-zA-Z]+)', jscode,
+ r'signature=([$a-zA-Z]+)', jscode,
u'Initial JS player signature function name')
jsi = JSInterpreter(jscode)
@@ -1138,6 +1140,10 @@ class YoutubeIE(YoutubeBaseInfoExtractor, SubtitlesInfoExtractor):
# upload date
upload_date = None
mobj = re.search(r'(?s)id="eow-date.*?>(.*?)', video_webpage)
+ if mobj is None:
+ mobj = re.search(
+ r'(?s)id="watch-uploader-info".*?>.*?(?:Published|Uploaded|Streamed live) on (.*?)',
+ video_webpage)
if mobj is not None:
upload_date = ' '.join(re.sub(r'[/,-]', r' ', mobj.group(1)).split())
upload_date = unified_strdate(upload_date)
@@ -1145,7 +1151,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor, SubtitlesInfoExtractor):
m_cat_container = get_element_by_id("eow-category", video_webpage)
if m_cat_container:
category = self._html_search_regex(
- r'(?s)(.*?)', m_cat_container, 'cateory',
+ r'(?s)(.*?)', m_cat_container, 'category',
default=None)
video_categories = None if category is None else [category]
else:
@@ -1381,13 +1387,13 @@ class YoutubePlaylistIE(YoutubeBaseInfoExtractor):
| p/
)
(
- (?:PL|EC|UU|FL|RD)?[0-9A-Za-z-_]{10,}
+ (?:PL|LL|EC|UU|FL|RD)?[0-9A-Za-z-_]{10,}
# Top tracks, they can also include dots
|(?:MC)[\w\.]*
)
.*
|
- ((?:PL|EC|UU|FL|RD)[0-9A-Za-z-_]{10,})
+ ((?:PL|LL|EC|UU|FL|RD)[0-9A-Za-z-_]{10,})
)"""
_TEMPLATE_URL = 'https://www.youtube.com/playlist?list=%s'
_MORE_PAGES_INDICATOR = r'data-link-type="next"'
@@ -1410,11 +1416,9 @@ class YoutubePlaylistIE(YoutubeBaseInfoExtractor):
title_span = (search_title('playlist-title') or
search_title('title long-title') or search_title('title'))
title = clean_html(title_span)
- video_re = r'''(?x)data-video-username="(.*?)".*?
+ video_re = r'''(?x)data-video-username=".*?".*?
href="/watch\?v=([0-9A-Za-z_-]{11})&[^"]*?list=%s''' % re.escape(playlist_id)
- matches = orderedSet(re.findall(video_re, webpage, flags=re.DOTALL))
- # Some of the videos may have been deleted, their username field is empty
- ids = [video_id for (username, video_id) in matches if username]
+ ids = orderedSet(re.findall(video_re, webpage, flags=re.DOTALL))
url_results = self._ids_to_results(ids)
return self.playlist_result(url_results, playlist_id, title)
@@ -1771,9 +1775,12 @@ class YoutubeFeedsInfoExtractor(YoutubeBaseInfoExtractor):
feed_entries.extend(
self.url_result(video_id, 'Youtube', video_id=video_id)
for video_id in ids)
- if info['paging'] is None:
+ mobj = re.search(
+ r'data-uix-load-more-href="/?[^"]+paging=(?P\d+)',
+ feed_html)
+ if mobj is None:
break
- paging = info['paging']
+ paging = mobj.group('paging')
return self.playlist_result(feed_entries, playlist_title=self._PLAYLIST_TITLE)
class YoutubeSubscriptionsIE(YoutubeFeedsInfoExtractor):