Mirrors/yt-dlp
1
0
Fork 0
mirror of https://github.com/yt-dlp/yt-dlp.git synced 2024-11-16 16:13:35 +01:00
Code Issues Releases Activity

[ie/facebook:reel] Improve metadata extraction

Browse Source 
Closes #9057, Closes #10824
Authored by: lengzuo
This commit is contained in:
Leng 2024-09-14 07:18:13 +08:00 committed by GitHub
parent 3dfd720d09
commit 0e1b941c6b
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 27 additions and 15 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

42
yt_dlp/extractor/facebook.py
View File

@ -84,7 +84,7 @@ class FacebookIE(InfoExtractor):
'timestamp': 1692346159, 'timestamp': 1692346159,
'thumbnail': r're:^https?://.*', 'thumbnail': r're:^https?://.*',
'uploader_id': '100063551323670', 'uploader_id': '100063551323670',
'duration': 3132.184, 'duration': 3133.583,
'view_count': int, 'view_count': int,
'concurrent_view_count': 0, 'concurrent_view_count': 0,
}, },
@ -112,9 +112,10 @@ class FacebookIE(InfoExtractor):
'upload_date': '20140506', 'upload_date': '20140506',
'timestamp': 1399398998, 'timestamp': 1399398998,
'thumbnail': r're:^https?://.*', 'thumbnail': r're:^https?://.*',
'uploader_id': 'pfbid028wxorhX2ErLFJ578N6P3crHD3PHmXTCqCvfBpsnbSLmbokwSY75p5hWBjHGkG4zxl', 'uploader_id': 'pfbid05AzrFTXgY37tqwaSgbFTTEpCLBjjEJHkigogwGiRPtKEpAsJYJpzE94H1RxYXWEtl',
'duration': 131.03, 'duration': 131.03,
'concurrent_view_count': int, 'concurrent_view_count': int,
'view_count': int,
}, },
}, { }, {
'note': 'Video with DASH manifest', 'note': 'Video with DASH manifest',
@ -167,7 +168,7 @@ class FacebookIE(InfoExtractor):
# have 1080P, but only up to 720p in swf params # have 1080P, but only up to 720p in swf params
# data.video.story.attachments[].media # data.video.story.attachments[].media
'url': 'https://www.facebook.com/cnn/videos/10155529876156509/', 'url': 'https://www.facebook.com/cnn/videos/10155529876156509/',
'md5': 'ca63897a90c9452efee5f8c40d080e25', 'md5': '1659aa21fb3dd1585874f668e81a72c8',
'info_dict': { 'info_dict': {
'id': '10155529876156509', 'id': '10155529876156509',
'ext': 'mp4', 'ext': 'mp4',
@ -180,9 +181,10 @@ class FacebookIE(InfoExtractor):
'view_count': int, 'view_count': int,
'uploader_id': '100059479812265', 'uploader_id': '100059479812265',
'concurrent_view_count': int, 'concurrent_view_count': int,
'duration': 44.478, 'duration': 44.181,
}, },
}, { }, {
# FIXME: unable to extract uploader, no formats found
# bigPipe.onPageletArrive ... onPageletArrive pagelet_group_mall # bigPipe.onPageletArrive ... onPageletArrive pagelet_group_mall
# data.node.comet_sections.content.story.attachments[].style_type_renderer.attachment.media # data.node.comet_sections.content.story.attachments[].style_type_renderer.attachment.media
'url': 'https://www.facebook.com/yaroslav.korpan/videos/1417995061575415/', 'url': 'https://www.facebook.com/yaroslav.korpan/videos/1417995061575415/',
@ -241,9 +243,9 @@ class FacebookIE(InfoExtractor):
'timestamp': 1511548260, 'timestamp': 1511548260,
'upload_date': '20171124', 'upload_date': '20171124',
'uploader': 'Vickie Gentry', 'uploader': 'Vickie Gentry',
'uploader_id': 'pfbid0FuZhHCeWDAxWxEbr3yKPFaRstXvRxgsp9uCPG6GjD4J2AitB35NUAuJ4Q75KcjiDl', 'uploader_id': 'pfbid0FkkycT95ySNNyfCw4Cho6u5G7WbbZEcxT496Hq8rtx1K3LcTCATpR3wnyYhmyGC5l',
'thumbnail': r're:^https?://.*', 'thumbnail': r're:^https?://.*',
'duration': 148.435, 'duration': 148.224,
}, },
}, { }, {
# data.node.comet_sections.content.story.attachments[].styles.attachment.media # data.node.comet_sections.content.story.attachments[].styles.attachment.media
@ -271,7 +273,7 @@ class FacebookIE(InfoExtractor):
'description': 'Today Makkovik\'s own Pilot Mandy Smith made her inaugural landing on the airstrip in her hometown. What a proud moment as we all cheered and...', 'description': 'Today Makkovik\'s own Pilot Mandy Smith made her inaugural landing on the airstrip in her hometown. What a proud moment as we all cheered and...',
'thumbnail': r're:^https?://.*', 'thumbnail': r're:^https?://.*',
'uploader': 'Lela Evans', 'uploader': 'Lela Evans',
'uploader_id': 'pfbid0shZJipuigyy5mqrUJn9ub5LJFWNHvan5prtyi3LrDuuuJ4NwrURgnQHYR9fywBepl', 'uploader_id': 'pfbid0swT2y7t6TAsZVBvcyeYPdhTMefGaS26mzUwML3vd1ma6ndGZKxsyS4Ssu3jitZLXl',
'upload_date': '20231228', 'upload_date': '20231228',
'timestamp': 1703804085, 'timestamp': 1703804085,
'duration': 394.347, 'duration': 394.347,
@ -322,7 +324,7 @@ class FacebookIE(InfoExtractor):
'upload_date': '20180523', 'upload_date': '20180523',
'uploader': 'ESL One Dota 2', 'uploader': 'ESL One Dota 2',
'uploader_id': '100066514874195', 'uploader_id': '100066514874195',
'duration': 4524.212, 'duration': 4524.001,
'view_count': int, 'view_count': int,
'thumbnail': r're:^https?://.*', 'thumbnail': r're:^https?://.*',
'concurrent_view_count': int, 'concurrent_view_count': int,
@ -339,9 +341,9 @@ class FacebookIE(InfoExtractor):
'title': 'Josef', 'title': 'Josef',
'thumbnail': r're:^https?://.*', 'thumbnail': r're:^https?://.*',
'concurrent_view_count': int, 'concurrent_view_count': int,
'uploader_id': 'pfbid0cibUN6tV7DYgdbJdsUFN46wc4jKpVSPAvJQhFofGqBGmVn3V3JtAs2tfUwziw2hUl', 'uploader_id': 'pfbid02gpfwRM2XvdEJfsERupwQiNmBiDArc38RMRYZnap372q6Vs7MtFTVy72mmFWpJBTKl',
'timestamp': 1549275572, 'timestamp': 1549275572,
'duration': 3.413, 'duration': 3.283,
'uploader': 'Josef Novak', 'uploader': 'Josef Novak',
'description': '', 'description': '',
'upload_date': '20190204', 'upload_date': '20190204',
@ -396,6 +398,7 @@ class FacebookIE(InfoExtractor):
'playlist_count': 1, 'playlist_count': 1,
'skip': 'Requires logging in', 'skip': 'Requires logging in',
}, { }, {
# FIXME: Cannot parse data error
# data.event.cover_media_renderer.cover_video # data.event.cover_media_renderer.cover_video
'url': 'https://m.facebook.com/events/1509582499515440', 'url': 'https://m.facebook.com/events/1509582499515440',
'info_dict': { 'info_dict': {
@ -498,7 +501,8 @@ def extract_metadata(webpage):
or get_first(post, ('video', 'creation_story', 'attachments', ..., 'media', lambda k, v: k == 'owner' and v['name'])) or get_first(post, ('video', 'creation_story', 'attachments', ..., 'media', lambda k, v: k == 'owner' and v['name']))
or get_first(post, (..., 'video', lambda k, v: k == 'owner' and v['name'])) or get_first(post, (..., 'video', lambda k, v: k == 'owner' and v['name']))
or get_first(post, ('node', 'actors', ..., {dict})) or get_first(post, ('node', 'actors', ..., {dict}))
or get_first(post, ('event', 'event_creator', {dict})) or {}) or get_first(post, ('event', 'event_creator', {dict}))
or get_first(post, ('video', 'creation_story', 'short_form_video_context', 'video_owner', {dict})) or {})
uploader = uploader_data.get('name') or ( uploader = uploader_data.get('name') or (
clean_html(get_element_by_id('fbPhotoPageAuthorName', webpage)) clean_html(get_element_by_id('fbPhotoPageAuthorName', webpage))
or self._search_regex( or self._search_regex(
@ -524,6 +528,11 @@ def extract_metadata(webpage):
webpage, 'view count', default=None)), webpage, 'view count', default=None)),
'concurrent_view_count': get_first(post, ( 'concurrent_view_count': get_first(post, (
('video', (..., ..., 'attachments', ..., 'media')), 'liveViewerCount', {int_or_none})), ('video', (..., ..., 'attachments', ..., 'media')), 'liveViewerCount', {int_or_none})),
**traverse_obj(post, (lambda _, v: video_id in v['url'], 'feedback', {
'like_count': ('likers', 'count', {int}),
'comment_count': ('total_comment_count', {int}),
'repost_count': ('share_count_reduced', {parse_count}),
}), get_all=False),
} }
info_json_ld = self._search_json_ld(webpage, video_id, default={}) info_json_ld = self._search_json_ld(webpage, video_id, default={})
@ -932,18 +941,21 @@ class FacebookReelIE(InfoExtractor):
_TESTS = [{ _TESTS = [{
'url': 'https://www.facebook.com/reel/1195289147628387', 'url': 'https://www.facebook.com/reel/1195289147628387',
'md5': 'f13dd37f2633595982db5ed8765474d3', 'md5': 'a53256d10fc2105441fe0c4212ed8cea',
'info_dict': { 'info_dict': {
'id': '1195289147628387', 'id': '1195289147628387',
'ext': 'mp4', 'ext': 'mp4',
'title': 'md5:b05800b5b1ad56c0ca78bd3807b6a61e', 'title': r're:9\.6K views · 355 reactions .+ Let the “Slapathon” commence!! .+ LL COOL J · Mama Said Knock You Out$',
'description': 'md5:22f03309b216ac84720183961441d8db', 'description': r're:When your trying to help your partner .+ LL COOL J · Mama Said Knock You Out$',
'uploader': 'md5:723e6cb3091241160f20b3c5dc282af1', 'uploader': 'Beast Camp Training',
'uploader_id': '100040874179269', 'uploader_id': '100040874179269',
'duration': 9.579, 'duration': 9.579,
'timestamp': 1637502609, 'timestamp': 1637502609,
'upload_date': '20211121', 'upload_date': '20211121',
'thumbnail': r're:^https?://.*', 'thumbnail': r're:^https?://.*',
'like_count': int,
'comment_count': int,
'repost_count': int,
}, },
}] }]