From a181cd0c600409a11f3f5e8448eaef4c63fdf1ed Mon Sep 17 00:00:00 2001 From: pukkandan Date: Mon, 23 Aug 2021 21:51:42 +0530 Subject: [PATCH] [facebook] Fix metadata extraction Original PR: https://github.com/ytdl-org/youtube-dl/pull/29796 Closes #453, https://github.com/ytdl-org/youtube-dl/issues/29421, https://github.com/ytdl-org/youtube-dl/issues/23627, https://github.com/ytdl-org/youtube-dl/issues/23180, https://github.com/ytdl-org/youtube-dl/issues/14156 Authored by: kikuyan --- yt_dlp/extractor/facebook.py | 122 ++++++++++++++++++++++++----------- 1 file changed, 84 insertions(+), 38 deletions(-) diff --git a/yt_dlp/extractor/facebook.py b/yt_dlp/extractor/facebook.py index d0295c881..e5bdb335a 100644 --- a/yt_dlp/extractor/facebook.py +++ b/yt_dlp/extractor/facebook.py @@ -20,6 +20,7 @@ int_or_none, js_to_json, limit_length, + merge_dicts, network_exceptions, parse_count, qualities, @@ -80,7 +81,8 @@ class FacebookIE(InfoExtractor): 'info_dict': { 'id': '274175099429670', 'ext': 'mp4', - 'title': 're:^Asif Nawab Butt posted a video', + 'title': 'Asif Nawab Butt', + 'description': 'Asif Nawab Butt', 'uploader': 'Asif Nawab Butt', 'upload_date': '20140506', 'timestamp': 1399398998, @@ -135,15 +137,17 @@ class FacebookIE(InfoExtractor): 'upload_date': '20160223', 'uploader': 'Barack Obama', }, + 'skip': 'Gif on giphy.com gone', }, { # have 1080P, but only up to 720p in swf params # data.video.story.attachments[].media 'url': 'https://www.facebook.com/cnn/videos/10155529876156509/', - 'md5': '9571fae53d4165bbbadb17a94651dcdc', + 'md5': '3f3798adb2b73423263e59376f1f5eb7', 'info_dict': { 'id': '10155529876156509', 'ext': 'mp4', - 'title': 'She survived the holocaust — and years later, she’s getting her citizenship s...', + 'title': 'Holocaust survivor becomes US citizen', + 'description': 'She survived the holocaust — and years later, she’s getting her citizenship so she can vote for Hillary Clinton http://cnn.it/2eERh5f', 'timestamp': 1477818095, 'upload_date': '20161030', 'uploader': 'CNN', @@ -157,15 +161,18 @@ class FacebookIE(InfoExtractor): 'info_dict': { 'id': '1417995061575415', 'ext': 'mp4', - 'title': 'md5:1db063d6a8c13faa8da727817339c857', - 'timestamp': 1486648217, + 'title': 'Yaroslav Korpan - Довгоочікуване відео', + 'description': 'Довгоочікуване відео', + 'timestamp': 1486648771, 'upload_date': '20170209', 'uploader': 'Yaroslav Korpan', + 'uploader_id': '100000948048708', }, 'params': { 'skip_download': True, }, }, { + # FIXME 'url': 'https://www.facebook.com/LaGuiaDelVaron/posts/1072691702860471', 'info_dict': { 'id': '1072691702860471', @@ -183,12 +190,14 @@ class FacebookIE(InfoExtractor): # data.node.comet_sections.content.story.attachments[].style_type_renderer.attachment.media 'url': 'https://www.facebook.com/groups/1024490957622648/permalink/1396382447100162/', 'info_dict': { - 'id': '1396382447100162', + 'id': '202882990186699', 'ext': 'mp4', - 'title': 'md5:19a428bbde91364e3de815383b54a235', - 'timestamp': 1486035494, + 'title': 'Elisabeth Ahtn - Hello? Yes your uber ride is here\n* Jukin...', + 'description': 'Hello? Yes your uber ride is here\n* Jukin Media Verified *\nFind this video and others like it by visiting...', + 'timestamp': 1486035513, 'upload_date': '20170202', 'uploader': 'Elisabeth Ahtn', + 'uploader_id': '100013949973717', }, 'params': { 'skip_download': True, @@ -229,8 +238,12 @@ class FacebookIE(InfoExtractor): 'info_dict': { 'id': '359649331226507', 'ext': 'mp4', - 'title': '#ESLOne VoD - Birmingham Finals Day#1 Fnatic vs. @Evil Geniuses', + 'title': 'Fnatic vs. EG - Group A - Opening Match - ESL One Birmingham Day 1', + 'description': '#ESLOne VoD - Birmingham Finals Day#1 Fnatic vs. @Evil Geniuses', + 'timestamp': 1527084179, + 'upload_date': '20180523', 'uploader': 'ESL One Dota 2', + 'uploader_id': '234218833769558', }, 'params': { 'skip_download': True, @@ -247,6 +260,7 @@ class FacebookIE(InfoExtractor): 'url': 'https://www.facebook.com/watch/?v=647537299265662', 'only_matching': True, }, { + # FIXME: https://github.com/yt-dlp/yt-dlp/issues/542 # data.node.comet_sections.content.story.attachments[].style_type_renderer.attachment.all_subattachments.nodes[].media 'url': 'https://www.facebook.com/PankajShahLondon/posts/10157667649866271', 'info_dict': { @@ -277,6 +291,7 @@ class FacebookIE(InfoExtractor): 'upload_date': '20161122', 'timestamp': 1479793574, }, + 'skip': 'No video', }, { # data.video.creation_story.attachments[].media 'url': 'https://www.facebook.com/watch/live/?v=1823658634322275', @@ -379,6 +394,56 @@ def _extract_from_url(self, url, video_id): webpage = self._download_webpage( url.replace('://m.facebook.com/', '://www.facebook.com/'), video_id) + def extract_metadata(webpage): + video_title = self._html_search_regex( + r']*class="uiHeaderTitle"[^>]*>([^<]*)', webpage, + 'title', default=None) + if not video_title: + video_title = self._html_search_regex( + r'(?s)(.*?)', + webpage, 'alternative title', default=None) + if not video_title: + video_title = self._html_search_meta( + ['og:title', 'twitter:title', 'description'], + webpage, 'title', default=None) + if video_title: + video_title = limit_length(video_title, 80) + else: + video_title = 'Facebook video #%s' % video_id + description = self._html_search_meta( + ['description', 'og:description', 'twitter:description'], + webpage, 'description', default=None) + uploader = clean_html(get_element_by_id( + 'fbPhotoPageAuthorName', webpage)) or self._search_regex( + r'ownerName\s*:\s*"([^"]+)"', webpage, 'uploader', + default=None) or self._og_search_title(webpage, fatal=False) + timestamp = int_or_none(self._search_regex( + r']+data-utime=["\'](\d+)', webpage, + 'timestamp', default=None)) + thumbnail = self._html_search_meta( + ['og:image', 'twitter:image'], webpage, 'thumbnail', default=None) + # some webpages contain unretrievable thumbnail urls + # like https://lookaside.fbsbx.com/lookaside/crawler/media/?media_id=10155168902769113&get_thumbnail=1 + # in https://www.facebook.com/yaroslav.korpan/videos/1417995061575415/ + if thumbnail and not re.search(r'\.(?:jpg|png)', thumbnail): + thumbnail = None + view_count = parse_count(self._search_regex( + r'\bviewCount\s*:\s*["\']([\d,.]+)', webpage, 'view count', + default=None)) + info_dict = { + 'title': video_title, + 'description': description, + 'uploader': uploader, + 'timestamp': timestamp, + 'thumbnail': thumbnail, + 'view_count': view_count, + } + info_json_ld = self._search_json_ld(webpage, video_id, default={}) + if info_json_ld.get('title'): + info_json_ld['title'] = limit_length( + re.sub(r'\s*\|\s*Facebook$', '', info_json_ld['title']), 80) + return merge_dicts(info_json_ld, info_dict) + video_data = None def extract_video_data(instances): @@ -511,7 +576,15 @@ def parse_attachment(attachment, key='media'): if not entries: parse_graphql_video(video) - return self.playlist_result(entries, video_id) + if len(entries) > 1: + return self.playlist_result(entries, video_id) + + video_info = entries[0] + webpage_info = extract_metadata(webpage) + # honor precise duration in video info + if video_info.get('duration'): + webpage_info['duration'] = video_info['duration'] + return merge_dicts(webpage_info, video_info) if not video_data: m_msg = re.search(r'class="[^"]*uiInterstitialContent[^"]*">
(.*?)
', webpage) @@ -629,39 +702,12 @@ def parse_attachment(attachment, key='media'): process_formats(formats) - description = self._html_search_meta('description', webpage, default=None) - video_title = self._html_search_regex( - (r']*class="uiHeaderTitle"[^>]*>([^<]*)', - r'(?s)(.*?)'), - webpage, 'title', default=None) or self._og_search_title(webpage, default=None) or description - if video_title: - video_title = limit_length(video_title, 80) - else: - video_title = 'Facebook video #%s' % video_id - uploader = clean_html(get_element_by_id( - 'fbPhotoPageAuthorName', webpage)) or self._search_regex( - r'ownerName\s*:\s*"([^"]+)"', webpage, 'uploader', - default=None) or self._og_search_title(webpage, fatal=False) - timestamp = int_or_none(self._search_regex( - r']+data-utime=["\'](\d+)', webpage, - 'timestamp', default=None)) - thumbnail = self._html_search_meta(['og:image', 'twitter:image'], webpage) - - view_count = parse_count(self._search_regex( - r'\bviewCount\s*:\s*["\']([\d,.]+)', webpage, 'view count', - default=None)) - info_dict = { 'id': video_id, - 'title': video_title, 'formats': formats, - 'uploader': uploader, - 'timestamp': timestamp, - 'description': description, - 'thumbnail': thumbnail, - 'view_count': view_count, 'subtitles': subtitles, } + info_dict.update(extract_metadata(webpage)) return info_dict