]> Raphaƫl G. Git Repositories - youtubedl/blobdiff - youtube_dl/extractor/pornhub.py
Initiate a new release.
[youtubedl] / youtube_dl / extractor / pornhub.py
index fdda69f33064fe4bd4e89775c5e3c7d56e4946d2..fb2032832e4757e328d016ab289e892721d73af2 100644 (file)
@@ -4,10 +4,14 @@ import os
 import re
 
 from .common import InfoExtractor
-from ..utils import (
+from ..compat import (
+    compat_urllib_parse,
     compat_urllib_parse_urlparse,
     compat_urllib_request,
-    compat_urllib_parse,
+)
+from ..utils import (
+    ExtractorError,
+    str_to_int,
 )
 from ..aes import (
     aes_decrypt_text
@@ -15,36 +19,58 @@ from ..aes import (
 
 
 class PornHubIE(InfoExtractor):
-    _VALID_URL = r'^(?:https?://)?(?:www\.)?(?P<url>pornhub\.com/view_video\.php\?viewkey=(?P<videoid>[0-9a-f]+))'
+    _VALID_URL = r'https?://(?:www\.)?pornhub\.com/view_video\.php\?viewkey=(?P<id>[0-9a-f]+)'
     _TEST = {
         'url': 'http://www.pornhub.com/view_video.php?viewkey=648719015',
-        'file': '648719015.mp4',
         'md5': '882f488fa1f0026f023f33576004a2ed',
         'info_dict': {
-            "uploader": "BABES-COM",
+            'id': '648719015',
+            'ext': 'mp4',
+            "uploader": "Babes",
             "title": "Seductive Indian beauty strips down and fingers her pink pussy",
             "age_limit": 18
         }
     }
 
+    def _extract_count(self, pattern, webpage, name):
+        count = self._html_search_regex(pattern, webpage, '%s count' % name, fatal=False)
+        if count:
+            count = str_to_int(count)
+        return count
+
     def _real_extract(self, url):
-        mobj = re.match(self._VALID_URL, url)
-        video_id = mobj.group('videoid')
-        url = 'http://www.' + mobj.group('url')
+        video_id = self._match_id(url)
 
         req = compat_urllib_request.Request(url)
         req.add_header('Cookie', 'age_verified=1')
         webpage = self._download_webpage(req, video_id)
 
+        error_msg = self._html_search_regex(
+            r'(?s)<div class="userMessageSection[^"]*".*?>(.*?)</div>',
+            webpage, 'error message', default=None)
+        if error_msg:
+            error_msg = re.sub(r'\s+', ' ', error_msg)
+            raise ExtractorError(
+                'PornHub said: %s' % error_msg,
+                expected=True, video_id=video_id)
+
         video_title = self._html_search_regex(r'<h1 [^>]+>([^<]+)', webpage, 'title')
-        video_uploader = self._html_search_regex(r'<b>From: </b>(?:\s|<[^>]*>)*(.+?)<', webpage, 'uploader', fatal=False)
+        video_uploader = self._html_search_regex(
+            r'(?s)From:&nbsp;.+?<(?:a href="/users/|a href="/channels/|<span class="username)[^>]+>(.+?)<',
+            webpage, 'uploader', fatal=False)
         thumbnail = self._html_search_regex(r'"image_url":"([^"]+)', webpage, 'thumbnail', fatal=False)
         if thumbnail:
             thumbnail = compat_urllib_parse.unquote(thumbnail)
 
-        video_urls = list(map(compat_urllib_parse.unquote , re.findall(r'"quality_[0-9]{3}p":"([^"]+)', webpage)))
+        view_count = self._extract_count(r'<span class="count">([\d,\.]+)</span> views', webpage, 'view')
+        like_count = self._extract_count(r'<span class="votesUp">([\d,\.]+)</span>', webpage, 'like')
+        dislike_count = self._extract_count(r'<span class="votesDown">([\d,\.]+)</span>', webpage, 'dislike')
+        comment_count = self._extract_count(
+            r'All comments \(<var class="videoCommentCount">([\d,\.]+)</var>', webpage, 'comment')
+
+        video_urls = list(map(compat_urllib_parse.unquote, re.findall(r'"quality_[0-9]{3}p":"([^"]+)', webpage)))
         if webpage.find('"encrypted":true') != -1:
-            password = self._html_search_regex(r'"video_title":"([^"]+)', webpage, 'password').replace('+', ' ')
+            password = compat_urllib_parse.unquote_plus(self._html_search_regex(r'"video_title":"([^"]+)', webpage, 'password'))
             video_urls = list(map(lambda s: aes_decrypt_text(s, password, 32).decode('utf-8'), video_urls))
 
         formats = []
@@ -77,6 +103,10 @@ class PornHubIE(InfoExtractor):
             'uploader': video_uploader,
             'title': video_title,
             'thumbnail': thumbnail,
+            'view_count': view_count,
+            'like_count': like_count,
+            'dislike_count': dislike_count,
+            'comment_count': comment_count,
             'formats': formats,
             'age_limit': 18,
         }