X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Fxhamster.py;h=0f7be6a7d93adc3a4fea8c6995cd8b58a084b9b4;hb=84213ea8d41d5fe1608333a16ac578dccdf9a915;hp=4297dffeee8dcbd3c637c04cbf522e520c6c66d4;hpb=8945b10f6e10337db0c9bf7a70758c8ecbb6c830;p=youtube-dl
diff --git a/youtube_dl/extractor/xhamster.py b/youtube_dl/extractor/xhamster.py
index 4297dffee..0f7be6a7d 100644
--- a/youtube_dl/extractor/xhamster.py
+++ b/youtube_dl/extractor/xhamster.py
@@ -1,5 +1,6 @@
from __future__ import unicode_literals
+import itertools
import re
from .common import InfoExtractor
@@ -8,6 +9,7 @@ from ..utils import (
clean_html,
determine_ext,
dict_get,
+ extract_attributes,
ExtractorError,
int_or_none,
parse_duration,
@@ -111,7 +113,7 @@ class XHamsterIE(InfoExtractor):
display_id = mobj.group('display_id') or mobj.group('display_id_2')
desktop_url = re.sub(r'^(https?://(?:.+?\.)?)m\.', r'\1', url)
- webpage = self._download_webpage(desktop_url, video_id)
+ webpage, urlh = self._download_webpage_handle(desktop_url, video_id)
error = self._html_search_regex(
r'
]+id=["\']videoClosed["\'][^>]*>(.+?)
',
@@ -159,6 +161,9 @@ class XHamsterIE(InfoExtractor):
'ext': determine_ext(format_url, 'mp4'),
'height': get_height(quality),
'filesize': filesize,
+ 'http_headers': {
+ 'Referer': urlh.geturl(),
+ },
})
self._sort_formats(formats)
@@ -331,3 +336,49 @@ class XHamsterEmbedIE(InfoExtractor):
video_url = dict_get(vars, ('downloadLink', 'homepageLink', 'commentsLink', 'shareUrl'))
return self.url_result(video_url, 'XHamster')
+
+
+class XHamsterUserIE(InfoExtractor):
+ _VALID_URL = r'https?://(?:.+?\.)?%s/users/(?P[^/?#&]+)' % XHamsterIE._DOMAINS
+ _TESTS = [{
+ # Paginated user profile
+ 'url': 'https://xhamster.com/users/netvideogirls/videos',
+ 'info_dict': {
+ 'id': 'netvideogirls',
+ },
+ 'playlist_mincount': 267,
+ }, {
+ # Non-paginated user profile
+ 'url': 'https://xhamster.com/users/firatkaan/videos',
+ 'info_dict': {
+ 'id': 'firatkaan',
+ },
+ 'playlist_mincount': 1,
+ }]
+
+ def _entries(self, user_id):
+ next_page_url = 'https://xhamster.com/users/%s/videos/1' % user_id
+ for pagenum in itertools.count(1):
+ page = self._download_webpage(
+ next_page_url, user_id, 'Downloading page %s' % pagenum)
+ for video_tag in re.findall(
+ r'(]+class=["\'].*?\bvideo-thumb__image-container[^>]+>)',
+ page):
+ video = extract_attributes(video_tag)
+ video_url = url_or_none(video.get('href'))
+ if not video_url or not XHamsterIE.suitable(video_url):
+ continue
+ video_id = XHamsterIE._match_id(video_url)
+ yield self.url_result(
+ video_url, ie=XHamsterIE.ie_key(), video_id=video_id)
+ mobj = re.search(r']+data-page=["\']next[^>]+>', page)
+ if not mobj:
+ break
+ next_page = extract_attributes(mobj.group(0))
+ next_page_url = url_or_none(next_page.get('href'))
+ if not next_page_url:
+ break
+
+ def _real_extract(self, url):
+ user_id = self._match_id(url)
+ return self.playlist_result(self._entries(user_id), user_id)