1
0
Fork 0
mirror of https://github.com/ytdl-org/youtube-dl.git synced 2024-11-22 02:01:50 +00:00

[Clipchamp] Add new extractor back-ported from yt-dlp

This commit is contained in:
dirkf 2023-05-04 00:08:26 +01:00
parent 4339910df3
commit 8465222041
2 changed files with 77 additions and 0 deletions

View file

@ -0,0 +1,76 @@
# coding: utf-8
from __future__ import unicode_literals
from .common import InfoExtractor
from ..compat import compat_str
from ..utils import (
ExtractorError,
merge_dicts,
T,
traverse_obj,
unified_timestamp,
url_or_none,
)
class ClipchampIE(InfoExtractor):
_VALID_URL = r'https?://(?:www\.)?clipchamp\.com/watch/(?P<id>[\w-]+)'
_TESTS = [{
'url': 'https://clipchamp.com/watch/gRXZ4ZhdDaU',
'info_dict': {
'id': 'gRXZ4ZhdDaU',
'ext': 'mp4',
'title': 'Untitled video',
'uploader': 'Alexander Schwartz',
'timestamp': 1680805580,
'upload_date': '20230406',
'thumbnail': r're:^https?://.+\.jpg',
},
'params': {
'skip_download': 'm3u8',
'format': 'bestvideo',
},
}]
_STREAM_URL_TMPL = 'https://%s.cloudflarestream.com/%s/manifest/video.%s'
_STREAM_URL_QUERY = {'parentOrigin': 'https://clipchamp.com'}
def _search_nextjs_data(self, webpage, video_id, **kw):
return self._parse_json(
self._search_regex(
r'(?s)<script[^>]+id=[\'"]__NEXT_DATA__[\'"][^>]*>([^<]+)</script>',
webpage, 'next.js data', **kw),
video_id, **kw)
def _real_extract(self, url):
video_id = self._match_id(url)
webpage = self._download_webpage(url, video_id)
data = self._search_nextjs_data(webpage, video_id)['props']['pageProps']['video']
storage_location = data.get('storage_location')
if storage_location != 'cf_stream':
raise ExtractorError('Unsupported clip storage location "%s"' % (storage_location,))
path = data['download_url']
iframe = self._download_webpage(
'https://iframe.cloudflarestream.com/' + path, video_id, 'Downloading player iframe')
subdomain = self._search_regex(
r'''\bcustomer-domain-prefix\s*=\s*("|')(?P<sd>[\w-]+)\1''', iframe,
'subdomain', group='sd', fatal=False) or 'customer-2ut9yn3y6fta1yxe'
formats = self._extract_mpd_formats(
self._STREAM_URL_TMPL % (subdomain, path, 'mpd'), video_id,
query=self._STREAM_URL_QUERY, fatal=False, mpd_id='dash')
formats.extend(self._extract_m3u8_formats(
self._STREAM_URL_TMPL % (subdomain, path, 'm3u8'), video_id, 'mp4',
query=self._STREAM_URL_QUERY, fatal=False, m3u8_id='hls'))
return merge_dicts({
'id': video_id,
'formats': formats,
'uploader': ' '.join(traverse_obj(data, ('creator', ('first_name', 'last_name'), T(compat_str)))) or None,
}, traverse_obj(data, {
'title': ('project', 'project_name', T(compat_str)),
'timestamp': ('created_at', T(unified_timestamp)),
'thumbnail': ('thumbnail_url', T(url_or_none)),
}), rev=True)

View file

@ -226,6 +226,7 @@ from .ciscolive import (
CiscoLiveSearchIE, CiscoLiveSearchIE,
) )
from .cjsw import CJSWIE from .cjsw import CJSWIE
from .clipchamp import ClipchampIE
from .cliphunter import CliphunterIE from .cliphunter import CliphunterIE
from .clippit import ClippitIE from .clippit import ClippitIE
from .cliprs import ClipRsIE from .cliprs import ClipRsIE