From 1a4fbe8462f5e531a891aeac7db6c0bde49c5536 Mon Sep 17 00:00:00 2001 From: FraFraFra-LongD <85188920+FraFraFra-LongD@users.noreply.github.com> Date: Sun, 13 Nov 2022 14:22:04 +0100 Subject: Added ThisVid.com support (#29187) * add ThisVidIE, ThisVidMemberIE, ThisVidPlaylistIE * redirect embed to main page for more metadata * use KVS extraction newly added to GenericIE and remove duplicate tests * also add MrDeepFake etc compat to GenericIE (closes #22390) Co-authored-by: dirkf --- youtube_dl/extractor/generic.py | 54 ++++++++++++++++++++--------------------- 1 file changed, 26 insertions(+), 28 deletions(-) (limited to 'youtube_dl/extractor/generic.py') diff --git a/youtube_dl/extractor/generic.py b/youtube_dl/extractor/generic.py index 01e406750..597611157 100644 --- a/youtube_dl/extractor/generic.py +++ b/youtube_dl/extractor/generic.py @@ -2252,31 +2252,7 @@ class GenericIE(InfoExtractor): 'skip_download': True, }, }, { - # KVS Player - 'url': 'https://thisvid.com/videos/fruit-is-healthy/', - 'md5': 'f83e52f409b9139a7efee58ef926a72e', - 'info_dict': { - 'id': '7079579', - 'display_id': 'fruit-is-healthy', - 'ext': 'mp4', - 'title': 'Fruit is healthy - ThisVid.com', - 'thumbnail': 'https://media.thisvid.com/contents/videos_screenshots/7079000/7079579/preview.jpg', - } - }, { - # KVS Player - 'url': 'https://thisvid.com/embed/7079579/', - 'info_dict': { - 'id': '7079579', - 'display_id': 'fruit-is-healthy', - 'ext': 'mp4', - 'title': 'Fruit is healthy - ThisVid.com', - 'thumbnail': 'https://media.thisvid.com/contents/videos_screenshots/7079000/7079579/preview.jpg', - }, - 'params': { - 'skip_download': True, - }, - }, { - # KVS Player + # KVS Player (tested also in thisvid.py) 'url': 'https://youix.com/video/leningrad-zoj/', 'md5': '94f96ba95706dc3880812b27b7d8a2b8', 'info_dict': { @@ -2306,6 +2282,7 @@ class GenericIE(InfoExtractor): 'display_id': '40-nochey-2016', 'ext': 'mp4', 'title': '40 ночей (2016) - BogMedia.org', + 'description': 'md5:4e6d7d622636eb7948275432eb256dc3', 'thumbnail': 'https://bogmedia.org/contents/videos_screenshots/21000/21217/preview_480p.mp4.jpg', }, }, { @@ -2319,6 +2296,18 @@ class GenericIE(InfoExtractor): 'title': 'Syren De Mer onlyfans_05-07-2020Have_a_happy_safe_holiday5f014e68a220979bdb8cd_source / Embed плеер', 'thumbnail': r're:https?://www\.camhub\.world/contents/videos_screenshots/389000/389508/preview\.mp4\.jpg', }, + }, { + 'url': 'https://mrdeepfakes.com/video/5/selena-gomez-pov-deep-fakes', + 'md5': 'fec4ad5ec150f655e0c74c696a4a2ff4', + 'info_dict': { + 'id': '5', + 'display_id': 'selena-gomez-pov-deep-fakes', + 'ext': 'mp4', + 'title': 'Selena Gomez POV (Deep Fakes) DeepFake Porn - MrDeepFakes', + 'description': 'md5:17d1f84b578c9c26875ac5ef9a932354', + 'height': 720, + 'age_limit': 18, + }, }, ] @@ -2491,6 +2480,7 @@ class GenericIE(InfoExtractor): 'url': getrealurl(flashvars[key], flashvars['license_code']), 'format_id': format_id, 'ext': 'mp4', + 'http_headers': {'Referer': url}, })) if not formats[-1].get('height'): formats[-1]['quality'] = 1 @@ -2713,9 +2703,15 @@ class GenericIE(InfoExtractor): # but actually don't. AGE_LIMIT_MARKERS = [ r'Proudly Labeled RTA', + r'>[^<]*you acknowledge you are at least (\d+) years old', ] - if any(re.search(marker, webpage) for marker in AGE_LIMIT_MARKERS): - age_limit = 18 + for marker in AGE_LIMIT_MARKERS: + m = re.search(marker, webpage) + if not m: + continue + age_limit = max( + age_limit or 0, + int_or_none(m.groups() and m.group(1), default=18)) # video uploader is domain name video_uploader = self._search_regex( @@ -3570,7 +3566,9 @@ class GenericIE(InfoExtractor): self.report_extraction('KVS Player') if found.group('maj_ver') not in ('4', '5', '6'): self.report_warning('Untested major version (%s) in player engine - download may fail.' % (found.group('ver'), )) - return self._extract_kvs(url, webpage, video_id) + return merge_dicts( + self._extract_kvs(url, webpage, video_id), + info_dict) # Looking for http://schema.org/VideoObject json_ld = self._search_json_ld( -- cgit v1.2.3