import json
from .common import InfoExtractor
-from ..utils import (
+from ..compat import (
compat_urllib_request,
+)
+from ..utils import (
parse_duration,
str_to_int,
)
class XTubeIE(InfoExtractor):
- _VALID_URL = r'https?://(?:www\.)?(?P<url>xtube\.com/watch\.php\?v=(?P<videoid>[^/?&]+))'
+ _VALID_URL = r'https?://(?:www\.)?(?P<url>xtube\.com/watch\.php\?v=(?P<id>[^/?&]+))'
_TEST = {
'url': 'http://www.xtube.com/watch.php?v=kVTUy_G222_',
'md5': '092fbdd3cbe292c920ef6fc6a8a9cdab',
'id': 'kVTUy_G222_',
'ext': 'mp4',
'title': 'strange erotica',
- 'description': 'surreal gay themed erotica...almost an ET kind of thing',
+ 'description': 'http://www.xtube.com an ET kind of thing',
'uploader': 'greenshowers',
'duration': 450,
'age_limit': 18,
def _real_extract(self, url):
mobj = re.match(self._VALID_URL, url)
- video_id = mobj.group('videoid')
+ video_id = mobj.group('id')
url = 'http://www.' + mobj.group('url')
req = compat_urllib_request.Request(url)
'comment_count': comment_count,
'formats': formats,
'age_limit': 18,
- }
\ No newline at end of file
+ }
+
+
+class XTubeUserIE(InfoExtractor):
+ IE_DESC = 'XTube user profile'
+ _VALID_URL = r'https?://(?:www\.)?xtube\.com/community/profile\.php\?(.*?)user=(?P<username>[^&#]+)(?:$|[&#])'
+ _TEST = {
+ 'url': 'http://www.xtube.com/community/profile.php?user=greenshowers',
+ 'info_dict': {
+ 'id': 'greenshowers',
+ },
+ 'playlist_mincount': 155,
+ }
+
+ def _real_extract(self, url):
+ mobj = re.match(self._VALID_URL, url)
+ username = mobj.group('username')
+
+ profile_page = self._download_webpage(
+ url, username, note='Retrieving profile page')
+
+ video_count = int(self._search_regex(
+ r'<strong>%s\'s Videos \(([0-9]+)\)</strong>' % username, profile_page,
+ 'video count'))
+
+ PAGE_SIZE = 25
+ urls = []
+ page_count = (video_count + PAGE_SIZE + 1) // PAGE_SIZE
+ for n in range(1, page_count + 1):
+ lpage_url = 'http://www.xtube.com/user_videos.php?page=%d&u=%s' % (n, username)
+ lpage = self._download_webpage(
+ lpage_url, username,
+ note='Downloading page %d/%d' % (n, page_count))
+ urls.extend(
+ re.findall(r'addthis:url="([^"]+)"', lpage))
+
+ return {
+ '_type': 'playlist',
+ 'id': username,
+ 'entries': [{
+ '_type': 'url',
+ 'url': eurl,
+ 'ie_key': 'XTube',
+ } for eurl in urls]
+ }