瀏覽代碼

[pornhub] Add support for playlists (Closes #4995)

Sergey M․ 10 年之前
父節點
當前提交
e66e1a0046
共有 2 個文件被更改,包括 34 次插入1 次删除
  1. 4 1
      youtube_dl/extractor/__init__.py
  2. 30 0
      youtube_dl/extractor/pornhub.py

+ 4 - 1
youtube_dl/extractor/__init__.py

@@ -353,7 +353,10 @@ from .playfm import PlayFMIE
 from .playvid import PlayvidIE
 from .playvid import PlayvidIE
 from .podomatic import PodomaticIE
 from .podomatic import PodomaticIE
 from .pornhd import PornHdIE
 from .pornhd import PornHdIE
-from .pornhub import PornHubIE
+from .pornhub import (
+    PornHubIE,
+    PornHubPlaylistIE,
+)
 from .pornotube import PornotubeIE
 from .pornotube import PornotubeIE
 from .pornoxo import PornoXOIE
 from .pornoxo import PornoXOIE
 from .promptfile import PromptFileIE
 from .promptfile import PromptFileIE

+ 30 - 0
youtube_dl/extractor/pornhub.py

@@ -110,3 +110,33 @@ class PornHubIE(InfoExtractor):
             'formats': formats,
             'formats': formats,
             'age_limit': 18,
             'age_limit': 18,
         }
         }
+
+
+class PornHubPlaylistIE(InfoExtractor):
+    _VALID_URL = r'https?://(?:www\.)?pornhub\.com/playlist/(?P<id>\d+)'
+    _TESTS = [{
+        'url': 'http://www.pornhub.com/playlist/6201671',
+        'info_dict': {
+            'id': '6201671',
+            'title': 'P0p4',
+        },
+        'playlist_mincount': 35,
+    }]
+
+    def _real_extract(self, url):
+        playlist_id = self._match_id(url)
+
+        webpage = self._download_webpage(url, playlist_id)
+
+        entries = [
+            self.url_result('http://www.pornhub.com/%s' % video_url, 'PornHub')
+            for video_url in set(re.findall('href="/?(view_video\.php\?viewkey=\d+[^"]*)"', webpage))
+        ]
+
+        playlist = self._parse_json(
+            self._search_regex(
+                r'playlistObject\s*=\s*({.+?});', webpage, 'playlist'),
+            playlist_id)
+
+        return self.playlist_result(
+            entries, playlist_id, playlist.get('title'), playlist.get('description'))