X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;ds=inline;f=youtube_dl%2Fextractor%2Fxhamster.py;h=b3547174dd92beffafaf8f220b50b94a25f2fa2b;hb=b22ca76204e1a05e1c4b07d24cb6a0dbbc09d18e;hp=ccee77359761b5214eb4ff24f8051f3dda1995eb;hpb=6a16fd4a1ad1dbd9372f75ddffce1e9fe95b002c;p=youtube-dl
diff --git a/youtube_dl/extractor/xhamster.py b/youtube_dl/extractor/xhamster.py
index ccee77359..b3547174d 100644
--- a/youtube_dl/extractor/xhamster.py
+++ b/youtube_dl/extractor/xhamster.py
@@ -4,9 +4,9 @@ import re
from .common import InfoExtractor
from ..utils import (
+ dict_get,
float_or_none,
int_or_none,
- str_to_int,
unified_strdate,
)
@@ -22,7 +22,7 @@ class XHamsterIE(InfoExtractor):
'title': 'FemaleAgent Shy beauty takes the bait',
'upload_date': '20121014',
'uploader': 'Ruseful2011',
- 'duration': 893,
+ 'duration': 893.52,
'age_limit': 18,
}
},
@@ -34,7 +34,7 @@ class XHamsterIE(InfoExtractor):
'title': 'Britney Spears Sexy Booty',
'upload_date': '20130914',
'uploader': 'jojo747400',
- 'duration': 200,
+ 'duration': 200.48,
'age_limit': 18,
}
},
@@ -64,9 +64,10 @@ class XHamsterIE(InfoExtractor):
webpage = self._download_webpage(mrss_url, video_id)
title = self._html_search_regex(
- [r'
]*>(.+?)(?:,\s*[^,]*?\s*Porn\s*[^,]*?:\s*xHamster[^<]*| - xHamster\.com)',
- r']*>([^<]+)
',
- r']+itemprop=".*?caption.*?"[^>]+content="(.+?)"'], webpage, 'title')
+ [r']*>([^<]+)
',
+ r']+itemprop=".*?caption.*?"[^>]+content="(.+?)"',
+ r']*>(.+?)(?:,\s*[^,]*?\s*Porn\s*[^,]*?:\s*xHamster[^<]*| - xHamster\.com)'],
+ webpage, 'title')
# Only a few videos have an description
mobj = re.search(r'Description: ([^<]+)', webpage)
@@ -77,7 +78,7 @@ class XHamsterIE(InfoExtractor):
webpage, 'upload date', fatal=False))
uploader = self._html_search_regex(
- r"]+>(?P[^<]+)",
+ r']+itemprop=["\']author[^>]+>]+href=["\'].+?xhamster\.com/user/[^>]+>(?P.+?)',
webpage, 'uploader', default='anonymous')
thumbnail = self._search_regex(
@@ -170,6 +171,12 @@ class XHamsterEmbedIE(InfoExtractor):
video_url = self._search_regex(
r'href="(https?://xhamster\.com/movies/%s/[^"]+\.html[^"]*)"' % video_id,
- webpage, 'xhamster url')
+ webpage, 'xhamster url', default=None)
+
+ if not video_url:
+ vars = self._parse_json(
+ self._search_regex(r'vars\s*:\s*({.+?})\s*,\s*\n', webpage, 'vars'),
+ video_id)
+ video_url = dict_get(vars, ('downloadLink', 'homepageLink', 'commentsLink', 'shareUrl'))
return self.url_result(video_url, 'XHamster')