diff options
Diffstat (limited to 'youtube_dlc/extractor')
| -rw-r--r-- | youtube_dlc/extractor/tiktok.py | 28 | ||||
| -rw-r--r-- | youtube_dlc/extractor/youtube.py | 20 |
2 files changed, 36 insertions, 12 deletions
diff --git a/youtube_dlc/extractor/tiktok.py b/youtube_dlc/extractor/tiktok.py index 52e5f4f1f..0cfd2168a 100644 --- a/youtube_dlc/extractor/tiktok.py +++ b/youtube_dlc/extractor/tiktok.py @@ -12,19 +12,20 @@ from ..utils import ( class TikTokBaseIE(InfoExtractor): - def _extract_aweme(self, video_data, webpage, url): + def _extract_aweme(self, props_data, webpage, url): + video_data = try_get(props_data, lambda x: x['pageProps'], expected_type=dict) video_info = try_get( - video_data, lambda x: x['videoData']['itemInfos'], dict) + video_data, lambda x: x['itemInfo']['itemStruct'], dict) author_info = try_get( - video_data, lambda x: x['videoData']['authorInfos'], dict) - share_info = try_get(video_data, lambda x: x['shareMeta'], dict) + video_data, lambda x: x['itemInfo']['itemStruct']['author'], dict) + share_info = try_get(video_data, lambda x: x['itemInfo']['shareMeta'], dict) unique_id = str_or_none(author_info.get('uniqueId')) timestamp = try_get(video_info, lambda x: int(x['createTime']), int) date = datetime.fromtimestamp(timestamp).strftime('%Y%m%d') - height = try_get(video_info, lambda x: x['video']['videoMeta']['height'], int) - width = try_get(video_info, lambda x: x['video']['videoMeta']['width'], int) + height = try_get(video_info, lambda x: x['video']['height'], int) + width = try_get(video_info, lambda x: x['video']['width'], int) thumbnails = [] thumbnails.append({ 'url': video_info.get('thumbnail') or self._og_search_thumbnail(webpage), @@ -32,14 +33,20 @@ class TikTokBaseIE(InfoExtractor): 'height': height }) + url = '' + if not url: + url = try_get(video_info, lambda x: x['video']['playAddr']) + if not url: + url = try_get(video_info, lambda x: x['video']['downloadAddr']) formats = [] formats.append({ - 'url': try_get(video_info, lambda x: x['video']['urls'][0]), + 'url': url, 'ext': 'mp4', 'height': height, 'width': width }) + tracker = try_get(props_data, lambda x: x['initialProps']['$wid']) return { 'comment_count': int_or_none(video_info.get('commentCount')), 'duration': try_get(video_info, lambda x: x['video']['videoMeta']['duration'], int), @@ -63,6 +70,7 @@ class TikTokBaseIE(InfoExtractor): 'formats': formats, 'http_headers': { 'Referer': url, + 'Cookie': 'tt_webid=%s; tt_webid_v2=%s' % (tracker, tracker), } } @@ -130,10 +138,10 @@ class TikTokIE(TikTokBaseIE): r'id=\"__NEXT_DATA__\"\s+type=\"application\/json\"\s*[^>]+>\s*(?P<json_string_ld>[^<]+)', webpage, 'json_string', group='json_string_ld') json_data = self._parse_json(json_string, video_id) - video_data = try_get(json_data, lambda x: x['props']['pageProps'], expected_type=dict) + props_data = try_get(json_data, lambda x: x['props'], expected_type=dict) # Chech statusCode for success - if video_data.get('statusCode') == 0: - return self._extract_aweme(video_data, webpage, url) + if props_data.get('pageProps').get('statusCode') == 0: + return self._extract_aweme(props_data, webpage, url) raise ExtractorError('Video not available', video_id=video_id) diff --git a/youtube_dlc/extractor/youtube.py b/youtube_dlc/extractor/youtube.py index b5b229ccb..4fb49b864 100644 --- a/youtube_dlc/extractor/youtube.py +++ b/youtube_dlc/extractor/youtube.py @@ -1742,6 +1742,13 @@ class YoutubeIE(YoutubeBaseInfoExtractor): add_dash_mpd_pr(pl_response) return pl_response + def extract_embedded_config(embed_webpage, video_id): + embedded_config = self._search_regex( + r'setConfig\(({.*})\);', + embed_webpage, 'ytInitialData', default=None) + if embedded_config: + return embedded_config + player_response = {} # Get video info @@ -1755,8 +1762,17 @@ class YoutubeIE(YoutubeBaseInfoExtractor): # this can be viewed without login into Youtube url = proto + '://www.youtube.com/embed/%s' % video_id embed_webpage = self._download_webpage(url, video_id, 'Downloading embed webpage') - # check if video is only playable on youtube - if so it requires auth (cookies) - if re.search(r'player-unavailable">', embed_webpage) is not None: + ext = extract_embedded_config(embed_webpage, video_id) + # playabilityStatus = re.search(r'{\\\"status\\\":\\\"(?P<playabilityStatus>[^\"]+)\\\"', ext) + playable_in_embed = re.search(r'{\\\"playableInEmbed\\\":(?P<playableinEmbed>[^\,]+)', ext) + if not playable_in_embed: + self.to_screen('Could not determine whether playabale in embed for video %s' % video_id) + playable_in_embed = '' + else: + playable_in_embed = playable_in_embed.group('playableinEmbed') + # check if video is only playable on youtube in other words not playable in embed - if so it requires auth (cookies) + # if re.search(r'player-unavailable">', embed_webpage) is not None: + if playable_in_embed == 'false': ''' # TODO apply this patch when Support for Python 2.6(!) and above drops if ({'VISITOR_INFO1_LIVE', 'HSID', 'SSID', 'SID'} <= cookie_keys |