[vk] improve extraction

- add support for Odnoklassniki embeds
- update tests
- extract more video from user lists(closes #4470)
- fix wall post audio extraction(closes #18332)
- improve error detection(closes #22568)

1 files changed, 173 insertions, 156 deletions

diff --git a/youtube_dl/extractor/vk.py b/youtube_dl/extractor/vk.py
index 8b6dc0e24..c289fcad3 100644
--- a/youtube_dl/extractor/vk.py
+++ b/youtube_dl/extractor/vk.py
@@ -12,7 +12,6 @@ from ..utils import (
     get_element_by_class,
     int_or_none,
     orderedSet,
-    remove_start,
     str_or_none,
     str_to_int,
     unescapeHTML,
@@ -21,6 +20,7 @@ from ..utils import (
     urlencode_postdata,
 )
 from .dailymotion import DailymotionIE
+from .odnoklassniki import OdnoklassnikiIE
 from .pladform import PladformIE
 from .vimeo import VimeoIE
 from .youtube import YoutubeIE
@@ -60,6 +60,18 @@ class VKBaseIE(InfoExtractor):
     def _real_initialize(self):
         self._login()
 
+    def _download_payload(self, path, video_id, data, fatal=True):
+        data['al'] = 1
+        code, payload = self._download_json(
+            'https://vk.com/%s.php' % path, video_id,
+            data=urlencode_postdata(data), fatal=fatal,
+            headers={'X-Requested-With': 'XMLHttpRequest'})['payload']
+        if code == '3':
+            self.raise_login_required()
+        elif code == '8':
+            raise ExtractorError(clean_html(payload[0][1:-1]), expected=True)
+        return payload
+
 
 class VKIE(VKBaseIE):
     IE_NAME = 'vk'
@@ -96,7 +108,6 @@ class VKIE(VKBaseIE):
         },
         {
             'url': 'http://vk.com/video205387401_165548505',
-            'md5': '6c0aeb2e90396ba97035b9cbde548700',
             'info_dict': {
                 'id': '205387401_165548505',
                 'ext': 'mp4',
@@ -110,18 +121,18 @@ class VKIE(VKBaseIE):
         },
         {
             'note': 'Embedded video',
-            'url': 'http://vk.com/video_ext.php?oid=32194266&id=162925554&hash=7d8c2e0d5e05aeaa&hd=1',
-            'md5': 'c7ce8f1f87bec05b3de07fdeafe21a0a',
+            'url': 'https://vk.com/video_ext.php?oid=-77521&id=162222515&hash=87b046504ccd8bfa',
+            'md5': '7babad3b85ea2e91948005b1b8b0cb84',
             'info_dict': {
-                'id': '32194266_162925554',
+                'id': '-77521_162222515',
                 'ext': 'mp4',
-                'uploader': 'Vladimir Gavrin',
-                'title': 'Lin Dan',
-                'duration': 101,
-                'upload_date': '20120730',
-                'view_count': int,
+                'uploader': 're:(?:Noize MC|Alexander Ilyashenko).*',
+                'title': 'ProtivoGunz - Хуёвая песня',
+                'duration': 195,
+                'upload_date': '20120212',
+                'timestamp': 1329049880,
+                'uploader_id': '-77521',
             },
-            'skip': 'This video has been removed from public access.',
         },
         {
             # VIDEO NOW REMOVED
@@ -138,18 +149,19 @@ class VKIE(VKBaseIE):
                 'upload_date': '20121218',
                 'view_count': int,
             },
-            'skip': 'Requires vk account credentials',
+            'skip': 'Removed',
         },
         {
             'url': 'http://vk.com/hd_kino_mania?z=video-43215063_168067957%2F15c66b9b533119788d',
-            'md5': '4d7a5ef8cf114dfa09577e57b2993202',
             'info_dict': {
                 'id': '-43215063_168067957',
                 'ext': 'mp4',
-                'uploader': 'Киномания - лучшее из мира кино',
+                'uploader': 'Bro Mazter',
                 'title': ' ',
                 'duration': 7291,
                 'upload_date': '20140328',
+                'uploader_id': '223413403',
+                'timestamp': 1396018030,
             },
             'skip': 'Requires vk account credentials',
         },
@@ -165,7 +177,7 @@ class VKIE(VKBaseIE):
                 'upload_date': '20140626',
                 'view_count': int,
             },
-            'skip': 'Only works from Russia',
+            'skip': 'Removed',
         },
         {
             # video (removed?) only available with list id
@@ -247,6 +259,9 @@ class VKIE(VKBaseIE):
                 'uploader_id': '-387766',
                 'timestamp': 1475137527,
             },
+            'params': {
+                'skip_download': True,
+            },
         },
         {
             # live stream, hls and rtmp links, most likely already finished live
@@ -288,80 +303,94 @@ class VKIE(VKBaseIE):
         mobj = re.match(self._VALID_URL, url)
         video_id = mobj.group('videoid')
 
+        mv_data = {}
         if video_id:
-            info_url = 'https://vk.com/al_video.php?act=show_inline&al=1&video=' + video_id
+            data = {
+                'act': 'show_inline',
+                'video': video_id,
+            }
             # Some videos (removed?) can only be downloaded with list id specified
             list_id = mobj.group('list_id')
             if list_id:
-                info_url += '&list=%s' % list_id
+                data['list'] = list_id
+
+            payload = self._download_payload('al_video', video_id, data)
+            info_page = payload[1]
+            opts = payload[-1]
+            mv_data = opts.get('mvData') or {}
+            player = opts.get('player') or {}
         else:
-            info_url = 'http://vk.com/video_ext.php?' + mobj.group('embed_query')
             video_id = '%s_%s' % (mobj.group('oid'), mobj.group('id'))
 
-        info_page = self._download_webpage(info_url, video_id)
+            info_page = self._download_webpage(
+                'http://vk.com/video_ext.php?' + mobj.group('embed_query'), video_id)
 
-        error_message = self._html_search_regex(
-            [r'(?s)<!><div[^>]+class="video_layer_message"[^>]*>(.+?)</div>',
-                r'(?s)<div[^>]+id="video_ext_msg"[^>]*>(.+?)</div>'],
-            info_page, 'error message', default=None)
-        if error_message:
-            raise ExtractorError(error_message, expected=True)
+            error_message = self._html_search_regex(
+                [r'(?s)<!><div[^>]+class="video_layer_message"[^>]*>(.+?)</div>',
+                    r'(?s)<div[^>]+id="video_ext_msg"[^>]*>(.+?)</div>'],
+                info_page, 'error message', default=None)
+            if error_message:
+                raise ExtractorError(error_message, expected=True)
 
-        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)
+            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)
 
-        ERROR_COPYRIGHT = 'Video %s has been removed from public access due to rightholder complaint.'
+            ERROR_COPYRIGHT = 'Video %s has been removed from public access due to rightholder complaint.'
 
-        ERRORS = {
-            r'>Видеозапись .*? была изъята из публичного доступа в связи с обращением правообладателя.<':
-            ERROR_COPYRIGHT,
+            ERRORS = {
+                r'>Видеозапись .*? была изъята из публичного доступа в связи с обращением правообладателя.<':
+                ERROR_COPYRIGHT,
 
-            r'>The video .*? was removed from public access by request of the copyright holder.<':
-            ERROR_COPYRIGHT,
+                r'>The video .*? was removed from public access by request of the copyright holder.<':
+                ERROR_COPYRIGHT,
 
-            r'<!>Please log in or <':
-            'Video %s is only available for registered users, '
-            'use --username and --password options to provide account credentials.',
+                r'<!>Please log in or <':
+                'Video %s is only available for registered users, '
+                'use --username and --password options to provide account credentials.',
 
-            r'<!>Unknown error':
-            'Video %s does not exist.',
+                r'<!>Unknown error':
+                'Video %s does not exist.',
 
-            r'<!>Видео временно недоступно':
-            'Video %s is temporarily unavailable.',
+                r'<!>Видео временно недоступно':
+                'Video %s is temporarily unavailable.',
 
-            r'<!>Access denied':
-            'Access denied to video %s.',
+                r'<!>Access denied':
+                'Access denied to video %s.',
 
-            r'<!>Видеозапись недоступна, так как её автор был заблокирован.':
-            'Video %s is no longer available, because its author has been blocked.',
+                r'<!>Видеозапись недоступна, так как её автор был заблокирован.':
+                'Video %s is no longer available, because its author has been blocked.',
 
-            r'<!>This video is no longer available, because its author has been blocked.':
-            'Video %s is no longer available, because its author has been blocked.',
+                r'<!>This video is no longer available, because its author has been blocked.':
+                'Video %s is no longer available, because its author has been blocked.',
 
-            r'<!>This video is no longer available, because it has been deleted.':
-            'Video %s is no longer available, because it has been deleted.',
+                r'<!>This video is no longer available, because it has been deleted.':
+                'Video %s is no longer available, because it has been deleted.',
 
-            r'<!>The video .+? is not available in your region.':
-            'Video %s is not available in your region.',
-        }
+                r'<!>The video .+? is not available in your region.':
+                'Video %s is not available in your region.',
+            }
+
+            for error_re, error_msg in ERRORS.items():
+                if re.search(error_re, info_page):
+                    raise ExtractorError(error_msg % video_id, expected=True)
 
-        for error_re, error_msg in ERRORS.items():
-            if re.search(error_re, info_page):
-                raise ExtractorError(error_msg % video_id, expected=True)
+            player = self._parse_json(self._search_regex(
+                r'var\s+playerParams\s*=\s*({.+?})\s*;\s*\n',
+                info_page, 'player params'), video_id)
 
         youtube_url = YoutubeIE._extract_url(info_page)
         if youtube_url:
-            return self.url_result(youtube_url, ie=YoutubeIE.ie_key())
+            return self.url_result(youtube_url, YoutubeIE.ie_key())
 
         vimeo_url = VimeoIE._extract_url(url, info_page)
         if vimeo_url is not None:
-            return self.url_result(vimeo_url)
+            return self.url_result(vimeo_url, VimeoIE.ie_key())
 
         pladform_url = PladformIE._extract_url(info_page)
         if pladform_url:
-            return self.url_result(pladform_url)
+            return self.url_result(pladform_url, PladformIE.ie_key())
 
         m_rutube = re.search(
             r'\ssrc="((?:https?:)?//rutube\.ru\\?/(?:video|play)\\?/embed(?:.*?))\\?"', info_page)
@@ -374,6 +403,10 @@ class VKIE(VKBaseIE):
         if dailymotion_urls:
             return self.url_result(dailymotion_urls[0], DailymotionIE.ie_key())
 
+        odnoklassniki_url = OdnoklassnikiIE._extract_url(info_page)
+        if odnoklassniki_url:
+            return self.url_result(odnoklassniki_url, OdnoklassnikiIE.ie_key())
+
         m_opts = re.search(r'(?s)var\s+opts\s*=\s*({.+?});', info_page)
         if m_opts:
             m_opts_url = re.search(r"url\s*:\s*'((?!/\b)[^']+)", m_opts.group(1))
@@ -383,38 +416,7 @@ class VKIE(VKBaseIE):
                     opts_url = 'http:' + opts_url
                 return self.url_result(opts_url)
 
-        # vars does not look to be served anymore since 24.10.2016
-        data = self._parse_json(
-            self._search_regex(
-                r'var\s+vars\s*=\s*({.+?});', info_page, 'vars', default='{}'),
-            video_id, fatal=False)
-
-        # <!json> is served instead
-        if not data:
-            data = self._parse_json(
-                self._search_regex(
-                    [r'<!json>\s*({.+?})\s*<!>', r'<!json>\s*({.+})'],
-                    info_page, 'json', default='{}'),
-                video_id)
-            if data:
-                data = data['player']['params'][0]
-
-        if not data:
-            data = self._parse_json(
-                self._search_regex(
-                    r'var\s+playerParams\s*=\s*({.+?})\s*;\s*\n', info_page,
-                    'player params', default='{}'),
-                video_id)
-            if data:
-                data = data['params'][0]
-
-        # <!--{...}
-        if not data:
-            data = self._parse_json(
-                self._search_regex(
-                    r'<!--\s*({.+})', info_page, 'payload'),
-                video_id)['payload'][-1][-1]['player']['params'][0]
-
+        data = player['params'][0]
         title = unescapeHTML(data['md_title'])
 
         # 2 = live
@@ -463,12 +465,12 @@ class VKIE(VKBaseIE):
             'title': title,
             'thumbnail': data.get('jpg'),
             'uploader': data.get('md_author'),
-            'uploader_id': str_or_none(data.get('author_id')),
-            'duration': data.get('duration'),
+            'uploader_id': str_or_none(data.get('author_id') or mv_data.get('authorId')),
+            'duration': int_or_none(data.get('duration') or mv_data.get('duration')),
             'timestamp': timestamp,
             'view_count': view_count,
-            'like_count': int_or_none(data.get('liked')),
-            'dislike_count': int_or_none(data.get('nolikes')),
+            'like_count': int_or_none(mv_data.get('likes')),
+            'comment_count': int_or_none(mv_data.get('commcount')),
             'is_live': is_live,
         }
 
@@ -482,7 +484,6 @@ class VKUserVideosIE(VKBaseIE):
         'url': 'http://vk.com/videos205387401',
         'info_dict': {
             'id': '205387401',
-            'title': "Tom Cruise's Videos",
         },
         'playlist_mincount': 4,
     }, {
@@ -498,22 +499,25 @@ class VKUserVideosIE(VKBaseIE):
         'url': 'http://new.vk.com/videos205387401',
         'only_matching': True,
     }]
+    _VIDEO = collections.namedtuple(
+        'Video', ['owner_id', 'id', 'thumb', 'title', 'flags', 'duration', 'hash', 'moder_acts', 'owner', 'date', 'views', 'platform', 'blocked', 'music_video_meta'])
 
     def _real_extract(self, url):
         page_id = self._match_id(url)
 
-        webpage = self._download_webpage(url, page_id)
+        l = self._download_payload('al_video', page_id, {
+            'act': 'load_videos_silent',
+            'oid': page_id,
+        })[0]['']['list']
 
-        entries = [
-            self.url_result(
-                'http://vk.com/video' + video_id, 'VK', video_id=video_id)
-            for video_id in orderedSet(re.findall(r'href="/video(-?[0-9_]+)"', webpage))]
-
-        title = unescapeHTML(self._search_regex(
-            r'<title>\s*([^<]+?)\s+\|\s+\d+\s+videos',
-            webpage, 'title', default=page_id))
+        entries = []
+        for video in l:
+            v = self._VIDEO._make(video)
+            video_id = '%d_%d' % (v.owner_id, v.id)
+            entries.append(self.url_result(
+                'http://vk.com/video' + video_id, 'VK', video_id=video_id))
 
-        return self.playlist_result(entries, page_id, title)
+        return self.playlist_result(entries, page_id)
 
 
 class VKWallPostIE(VKBaseIE):
@@ -523,15 +527,15 @@ class VKWallPostIE(VKBaseIE):
         # public page URL, audio playlist
         'url': 'https://vk.com/bs.official?w=wall-23538238_35',
         'info_dict': {
-            'id': '23538238_35',
-            'title': 'Black Shadow - Wall post 23538238_35',
+            'id': '-23538238_35',
+            'title': 'Black Shadow - Wall post -23538238_35',
             'description': 'md5:3f84b9c4f9ef499731cf1ced9998cc0c',
         },
         'playlist': [{
             'md5': '5ba93864ec5b85f7ce19a9af4af080f6',
             'info_dict': {
                 'id': '135220665_111806521',
-                'ext': 'mp3',
+                'ext': 'mp4',
                 'title': 'Black Shadow - Слепое Верование',
                 'duration': 370,
                 'uploader': 'Black Shadow',
@@ -542,18 +546,16 @@ class VKWallPostIE(VKBaseIE):
             'md5': '4cc7e804579122b17ea95af7834c9233',
             'info_dict': {
                 'id': '135220665_111802303',
-                'ext': 'mp3',
+                'ext': 'mp4',
                 'title': 'Black Shadow - Война - Негасимое Бездны Пламя!',
                 'duration': 423,
                 'uploader': 'Black Shadow',
                 'artist': 'Black Shadow',
                 'track': 'Война - Негасимое Бездны Пламя!',
             },
-            'params': {
-                'skip_download': True,
-            },
         }],
         'params': {
+            'skip_download': True,
             'usenetrc': True,
         },
         'skip': 'Requires vk account credentials',
@@ -562,7 +564,7 @@ class VKWallPostIE(VKBaseIE):
         'url': 'https://vk.com/wall85155021_6319',
         'info_dict': {
             'id': '85155021_6319',
-            'title': 'Sergey Gorbunov - Wall post 85155021_6319',
+            'title': 'Сергей Горбунов - Wall post 85155021_6319',
         },
         'playlist_count': 1,
         'params': {
@@ -578,58 +580,73 @@ class VKWallPostIE(VKBaseIE):
         'url': 'https://m.vk.com/wall-23538238_35',
         'only_matching': True,
     }]
+    _BASE64_CHARS = 'abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMN0PQRSTUVWXYZO123456789+/='
+    _AUDIO = collections.namedtuple(
+        'Audio', ['id', 'owner_id', 'url', 'title', 'performer', 'duration', 'album_id', 'unk', 'author_link', 'lyrics', 'flags', 'context', 'extra', 'hashes', 'cover_url', 'ads', 'subtitle', 'main_artists', 'feat_artists', 'album', 'track_code', 'restriction', 'album_part', 'new_stats', 'access_key'])
+
+    def _decode(self, enc):
+        dec = ''
+        e = n = 0
+        for c in enc:
+            r = self._BASE64_CHARS.index(c)
+            cond = n % 4
+            e = 64 * e + r if cond else r
+            n += 1
+            if cond:
+                dec += chr(255 & e >> (-2 * n & 6))
+        return dec
+
+    def _unmask_url(self, mask_url, vk_id):
+        if 'audio_api_unavailable' in mask_url:
+            extra = mask_url.split('?extra=')[1].split('#')
+            func, base = self._decode(extra[1]).split(chr(11))
+            assert (func == 'i')
+            mask_url = list(self._decode(extra[0]))
+            url_len = len(mask_url)
+            indexes = [None] * url_len
+            index = int(base) ^ vk_id
+            for n in range(url_len - 1, -1, -1):
+                index = (url_len * (n + 1) ^ index + n) % url_len
+                indexes[n] = index
+            for n in range(1, url_len):
+                c = mask_url[n]
+                index = indexes[url_len - 1 - n]
+                mask_url[n] = mask_url[index]
+                mask_url[index] = c
+            mask_url = ''.join(mask_url)
+        return mask_url
 
     def _real_extract(self, url):
         post_id = self._match_id(url)
 
-        wall_url = 'https://vk.com/wall%s' % post_id
-
-        post_id = remove_start(post_id, '-')
-
-        webpage = self._download_webpage(wall_url, post_id)
-
-        error = self._html_search_regex(
-            r'>Error</div>\s*<div[^>]+class=["\']body["\'][^>]*>([^<]+)',
-            webpage, 'error', default=None)
-        if error:
-            raise ExtractorError('VK said: %s' % error, expected=True)
+        webpage = self._download_payload('wkview', post_id, {
+            'act': 'show',
+            'w': 'wall' + post_id,
+        })[1]
 
         description = clean_html(get_element_by_class('wall_post_text', webpage))
         uploader = clean_html(get_element_by_class('author', webpage))
-        thumbnail = self._og_search_thumbnail(webpage)
 
         entries = []
 
-        audio_ids = re.findall(r'data-full-id=["\'](\d+_\d+)', webpage)
-        if audio_ids:
-            al_audio = self._download_webpage(
-                'https://vk.com/al_audio.php', post_id,
-                note='Downloading audio info', fatal=False,
-                data=urlencode_postdata({
-                    'act': 'reload_audio',
-                    'al': '1',
-                    'ids': ','.join(audio_ids)
-                }))
-            if al_audio:
-                Audio = collections.namedtuple(
-                    'Audio', ['id', 'user_id', 'url', 'track', 'artist', 'duration'])
-                audios = self._parse_json(
-                    self._search_regex(
-                        r'<!json>(.+?)<!>', al_audio, 'audios', default='[]'),
-                    post_id, fatal=False, transform_source=unescapeHTML)
-                if isinstance(audios, list):
-                    for audio in audios:
-                        a = Audio._make(audio[:6])
-                        entries.append({
-                            'id': '%s_%s' % (a.user_id, a.id),
-                            'url': a.url,
-                            'title': '%s - %s' % (a.artist, a.track) if a.artist and a.track else a.id,
-                            'thumbnail': thumbnail,
-                            'duration': a.duration,
-                            'uploader': uploader,
-                            'artist': a.artist,
-                            'track': a.track,
-                        })
+        for audio in re.findall(r'data-audio="([^"]+)', webpage):
+            audio = self._parse_json(unescapeHTML(audio), post_id)
+            a = self._AUDIO._make(audio)
+            if not a.url:
+                continue
+            title = unescapeHTML(a.title)
+            entries.append({
+                'id': '%s_%s' % (a.owner_id, a.id),
+                'url': self._unmask_url(a.url, a.ads['vk_id']),
+                'title': '%s - %s' % (a.performer, title) if a.performer else title,
+                'thumbnail': a.cover_url.split(',') if a.cover_url else None,
+                'duration': a.duration,
+                'uploader': uploader,
+                'artist': a.performer,
+                'track': title,
+                'ext': 'mp4',
+                'protocol': 'm3u8',
+            })
 
         for video in re.finditer(
                 r'<a[^>]+href=(["\'])(?P<url>/video(?:-?[\d_]+).*?)\1', webpage):