[vk] Use `_form_hidden_inputs` when logging in

[youtube-dl] / youtube_dl / extractor / vk.py

diff --git a/youtube_dl/extractor/vk.py b/youtube_dl/extractor/vk.py
index f974f8fef8415920cb171f0601189fb1e1b5ebc9..c0292095bab55d792249fedd03a7d927a98763e5 100644 (file)
--- a/youtube_dl/extractor/vk.py
+++ b/youtube_dl/extractor/vk.py
@@ -13,6 +13,7 @@ from ..compat import (
 from ..utils import (
     ExtractorError,
     orderedSet,
+    str_to_int,
     unescapeHTML,
     unified_strdate,
 )
@@ -20,7 +21,17 @@ from ..utils import (
 
 class VKIE(InfoExtractor):
     IE_NAME = 'vk.com'
-    _VALID_URL = r'https?://(?:m\.)?vk\.com/(?:video_ext\.php\?.*?\boid=(?P<oid>-?\d+).*?\bid=(?P<id>\d+)|(?:.+?\?.*?z=)?video(?P<videoid>[^s].*?)(?:\?|%2F|$))'
+    _VALID_URL = r'''(?x)
+                    https?://
+                        (?:
+                            (?:m\.)?vk\.com/video_ext\.php\?.*?\boid=(?P<oid>-?\d+).*?\bid=(?P<id>\d+)|
+                            (?:
+                                (?:m\.)?vk\.com/(?:.+?\?.*?z=)?video|
+                                (?:www\.)?biqle\.ru/watch/
+                            )
+                            (?P<videoid>[^s].*?)(?:\?|%2F|$)
+                        )
+                    '''
     _NETRC_MACHINE = 'vk'
 
     _TESTS = [
@@ -34,6 +45,7 @@ class VKIE(InfoExtractor):
                 'uploader': 're:(?:Noize MC|Alexander Ilyashenko).*',
                 'duration': 195,
                 'upload_date': '20120212',
+                'view_count': int,
             },
         },
         {
@@ -45,7 +57,8 @@ class VKIE(InfoExtractor):
                 'uploader': 'Tom Cruise',
                 'title': 'No name',
                 'duration': 9,
-                'upload_date': '20130721'
+                'upload_date': '20130721',
+                'view_count': int,
             }
         },
         {
@@ -59,6 +72,7 @@ class VKIE(InfoExtractor):
                 'title': 'Lin Dan',
                 'duration': 101,
                 'upload_date': '20120730',
+                'view_count': int,
             }
         },
         {
@@ -73,7 +87,8 @@ class VKIE(InfoExtractor):
                 'uploader': 'Триллеры',
                 'title': '► Бойцовский клуб / Fight Club 1999 [HD 720]',
                 'duration': 8352,
-                'upload_date': '20121218'
+                'upload_date': '20121218',
+                'view_count': int,
             },
             'skip': 'Requires vk account credentials',
         },
@@ -100,14 +115,35 @@ class VKIE(InfoExtractor):
                 'title': 'Книга Илая',
                 'duration': 6771,
                 'upload_date': '20140626',
+                'view_count': int,
             },
             'skip': 'Only works from Russia',
         },
+        {
+            # youtube embed
+            'url': 'https://vk.com/video276849682_170681728',
+            'info_dict': {
+                'id': 'V3K4mi0SYkc',
+                'ext': 'mp4',
+                'title': "DSWD Awards 'Children's Joy Foundation, Inc.' Certificate of Registration and License to Operate",
+                'description': 'md5:bf9c26cfa4acdfb146362682edd3827a',
+                'duration': 179,
+                'upload_date': '20130116',
+                'uploader': "Children's Joy Foundation",
+                'uploader_id': 'thecjf',
+                'view_count': int,
+            },
+        },
         {
             # removed video, just testing that we match the pattern
             'url': 'http://vk.com/feed?z=video-43215063_166094326%2Fbb50cacd3177146d7a',
             'only_matching': True,
         },
+        {
+            # vk wrapper
+            'url': 'http://www.biqle.ru/watch/847655_160197695',
+            'only_matching': True,
+        }
     ]
 
     def _login(self):
@@ -115,20 +151,25 @@ class VKIE(InfoExtractor):
         if username is None:
             return
 
-        login_form = {
-            'act': 'login',
-            'role': 'al_frame',
-            'expire': '1',
+        login_page = self._download_webpage(
+            'https://vk.com', None, 'Downloading login page')
+
+        login_form = self._form_hidden_inputs(login_page)
+
+        login_form.update({
             'email': username.encode('cp1251'),
             'pass': password.encode('cp1251'),
-        }
+        })
 
-        request = compat_urllib_request.Request('https://login.vk.com/?act=login',
-                                                compat_urllib_parse.urlencode(login_form).encode('utf-8'))
-        login_page = self._download_webpage(request, None, note='Logging in as %s' % username)
+        request = compat_urllib_request.Request(
+            'https://login.vk.com/?act=login',
+            compat_urllib_parse.urlencode(login_form).encode('utf-8'))
+        login_page = self._download_webpage(
+            request, None, note='Logging in as %s' % username)
 
         if re.search(r'onLoginFailed', login_page):
-            raise ExtractorError('Unable to login, incorrect username and/or password', expected=True)
+            raise ExtractorError(
+                'Unable to login, incorrect username and/or password', expected=True)
 
     def _real_initialize(self):
         self._login()
@@ -140,9 +181,14 @@ class VKIE(InfoExtractor):
         if not video_id:
             video_id = '%s_%s' % (mobj.group('oid'), mobj.group('id'))
 
-        info_url = 'http://vk.com/al_video.php?act=show&al=1&module=video&video=%s' % video_id
+        info_url = 'https://vk.com/al_video.php?act=show&al=1&module=video&video=%s' % video_id
         info_page = self._download_webpage(info_url, video_id)
 
+        if re.search(r'<!>/login\.php\?.*\bact=security_check', info_page):
+            raise ExtractorError(
+                'You are trying to log in from an unusual location. You should confirm ownership at vk.com to log in with this IP.',
+                expected=True)
+
         ERRORS = {
             r'>Видеозапись .*? была изъята из публичного доступа в связи с обращением правообладателя.<':
             'Video %s has been removed from public access due to rightholder complaint.',
@@ -162,10 +208,11 @@ class VKIE(InfoExtractor):
             if re.search(error_re, info_page):
                 raise ExtractorError(error_msg % video_id, expected=True)
 
-        m_yt = re.search(r'src="(http://www.youtube.com/.*?)"', info_page)
-        if m_yt is not None:
-            self.to_screen('Youtube video detected')
-            return self.url_result(m_yt.group(1), 'Youtube')
+        youtube_url = self._search_regex(
+            r'<iframe[^>]+src="((?:https?:)?//www.youtube.com/embed/[^"]+)"',
+            info_page, 'youtube iframe', default=None)
+        if youtube_url:
+            return self.url_result(youtube_url, 'Youtube')
 
         m_rutube = re.search(
             r'\ssrc="((?:https?:)?//rutube\.ru\\?/video\\?/embed(?:.*?))\\?"', info_page)
@@ -194,6 +241,10 @@ class VKIE(InfoExtractor):
             mobj.group(1) + ' ' + mobj.group(2)
             upload_date = unified_strdate(mobj.group(1) + ' ' + mobj.group(2))
 
+        view_count = str_to_int(self._search_regex(
+            r'"mv_views_count_number"[^>]*>([\d,.]+) views<',
+            info_page, 'view count', fatal=False))
+
         formats = [{
             'format_id': k,
             'url': v,
@@ -210,6 +261,7 @@ class VKIE(InfoExtractor):
             'uploader': data.get('md_author'),
             'duration': data.get('duration'),
             'upload_date': upload_date,
+            'view_count': view_count,
         }