gitlab-com-dstftw/youtube-dl
1
0
Fork 0
mirror of https://gitlab.com/dstftw/youtube-dl.git synced 2020-11-16 09:42:26 +00:00
Code Issues Releases Wiki Activity

[tiktok] Add support for new URL schema (closes #20573)

Browse source
This commit is contained in:
Sergey M․ 2019-04-07 21:05:50 +07:00
parent d562cac9dc
commit 7c2ecbc1cc
No known key found for this signature in database
GPG key ID: 2C393E0F18A9236D
1 changed files with 28 additions and 7 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

35
youtube_dl/extractor/tiktok.py
View file

@ -65,8 +65,15 @@ class TikTokBaseIE(InfoExtractor):
class TikTokIE(TikTokBaseIE): class TikTokIE(TikTokBaseIE):
_VALID_URL = r'https?://(?:m\.)?tiktok\.com/v/(?P<id>\d+)' _VALID_URL = r'''(?x)
_TEST = { https?://
(?:
(?:m\.)?tiktok\.com/v|
(?:www\.)?tiktok\.com/share/video
)
/(?P<id>\d+)
'''
_TESTS = [{
'url': 'https://m.tiktok.com/v/6606727368545406213.html', 'url': 'https://m.tiktok.com/v/6606727368545406213.html',
'md5': 'd584b572e92fcd48888051f238022420', 'md5': 'd584b572e92fcd48888051f238022420',
'info_dict': { 'info_dict': {
@ -81,25 +88,39 @@ class TikTokIE(TikTokBaseIE):
'comment_count': int, 'comment_count': int,
'repost_count': int, 'repost_count': int,
} }
} }, {
'url': 'https://www.tiktok.com/share/video/6606727368545406213',
'only_matching': True,
}]
def _real_extract(self, url): def _real_extract(self, url):
video_id = self._match_id(url) video_id = self._match_id(url)
webpage = self._download_webpage(url, video_id) webpage = self._download_webpage(
'https://m.tiktok.com/v/%s.html' % video_id, video_id)
data = self._parse_json(self._search_regex( data = self._parse_json(self._search_regex(
r'\bdata\s*=\s*({.+?})\s*;', webpage, 'data'), video_id) r'\bdata\s*=\s*({.+?})\s*;', webpage, 'data'), video_id)
return self._extract_aweme(data) return self._extract_aweme(data)
class TikTokUserIE(TikTokBaseIE): class TikTokUserIE(TikTokBaseIE):
_VALID_URL = r'https?://(?:m\.)?tiktok\.com/h5/share/usr/(?P<id>\d+)' _VALID_URL = r'''(?x)
_TEST = { https?://
(?:
(?:m\.)?tiktok\.com/h5/share/usr|
(?:www\.)?tiktok\.com/share/user
)
/(?P<id>\d+)
'''
_TESTS = [{
'url': 'https://m.tiktok.com/h5/share/usr/188294915489964032.html', 'url': 'https://m.tiktok.com/h5/share/usr/188294915489964032.html',
'info_dict': { 'info_dict': {
'id': '188294915489964032', 'id': '188294915489964032',
}, },
'playlist_mincount': 24, 'playlist_mincount': 24,
} }, {
'url': 'https://www.tiktok.com/share/user/188294915489964032',
'only_matching': True,
}]
def _real_extract(self, url): def _real_extract(self, url):
user_id = self._match_id(url) user_id = self._match_id(url)