[xhamster] Add support for mobile URLs and fix thumbnail extraction
authorWindom <windom@users.noreply.github.com>
Tue, 5 Dec 2017 17:08:31 +0000 (19:08 +0200)
committerSergey M <dstftw@gmail.com>
Tue, 5 Dec 2017 17:08:31 +0000 (00:08 +0700)
youtube_dl/extractor/xhamster.py

index 52f8ded2f8b94f872d5dc8799ec4fbe2829c0754..68652a22fc7453d01802334b3af021557cc1cedd 100644 (file)
@@ -75,6 +75,10 @@ class XHamsterIE(InfoExtractor):
         'params': {
             'skip_download': True,
         },
+    }, {
+        # mobile site
+        'url': 'https://m.xhamster.com/videos/cute-teen-jacqueline-solo-masturbation-8559111',
+        'only_matching': True,
     }, {
         'url': 'https://xhamster.com/movies/2272726/amber_slayed_by_the_knight.html',
         'only_matching': True,
@@ -93,7 +97,8 @@ class XHamsterIE(InfoExtractor):
         video_id = mobj.group('id') or mobj.group('id_2')
         display_id = mobj.group('display_id') or mobj.group('display_id_2')
 
-        webpage = self._download_webpage(url, video_id)
+        desktop_url = re.sub(r'^(https?://(?:.+?\.)?)m\.', r'\1', url)
+        webpage = self._download_webpage(desktop_url, video_id)
 
         error = self._html_search_regex(
             r'<div[^>]+id=["\']videoClosed["\'][^>]*>(.+?)</div>',
@@ -229,8 +234,8 @@ class XHamsterIE(InfoExtractor):
             webpage, 'uploader', default='anonymous')
 
         thumbnail = self._search_regex(
-            [r'''thumb\s*:\s*(?P<q>["'])(?P<thumbnail>.+?)(?P=q)''',
-             r'''<video[^>]+poster=(?P<q>["'])(?P<thumbnail>.+?)(?P=q)[^>]*>'''],
+            [r'''["']thumbUrl["']\s*:\s*(?P<q>["'])(?P<thumbnail>.+?)(?P=q)''',
+             r'''<video[^>]+"poster"=(?P<q>["'])(?P<thumbnail>.+?)(?P=q)[^>]*>'''],
             webpage, 'thumbnail', fatal=False, group='thumbnail')
 
         duration = parse_duration(self._search_regex(
@@ -274,15 +279,16 @@ class XHamsterIE(InfoExtractor):
 
 
 class XHamsterEmbedIE(InfoExtractor):
-    _VALID_URL = r'https?://(?:www\.)?xhamster\.com/xembed\.php\?video=(?P<id>\d+)'
+    _VALID_URL = r'https?://(?:.+?\.)?xhamster\.com/xembed\.php\?video=(?P<id>\d+)'
     _TEST = {
         'url': 'http://xhamster.com/xembed.php?video=3328539',
         'info_dict': {
             'id': '3328539',
             'ext': 'mp4',
             'title': 'Pen Masturbation',
+            'timestamp': 1406581861,
             'upload_date': '20140728',
-            'uploader_id': 'anonymous',
+            'uploader': 'ManyakisArt',
             'duration': 5,
             'age_limit': 18,
         }