kumi/yt-dlp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

[tvp] Add TVPStreamIE (#1401)

Browse source 
Authored by: selfisekai
This commit is contained in:
Lauren Liberda 2021-10-31 11:03:04 +05:30 committed by pukkandan
parent c15de6ffe6
commit ebfab36fca
No known key found for this signature in database
GPG key ID: 0F00D95A001F4698
2 changed files with 47 additions and 0 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

1
yt_dlp/extractor/extractors.py
View file

@ -1571,6 +1571,7 @@ from .tvnow import (
from .tvp import (
TVPEmbedIE,
TVPIE,
TVPStreamIE,
TVPWebsiteIE,
)
from .tvplay import (

46
yt_dlp/extractor/tvp.py
View file

@ -251,6 +251,52 @@ class TVPIE(InfoExtractor):
}
class TVPStreamIE(InfoExtractor):
IE_NAME = 'tvp:stream'
_VALID_URL = r'(?:tvpstream:|https?://tvpstream\.vod\.tvp\.pl/(?:\?(?:[^&]+[&;])*channel_id=)?)(?P<id>\d*)'
_TESTS = [{
# untestable as "video" id changes many times across a day
'url': 'https://tvpstream.vod.tvp.pl/?channel_id=1455',
'only_matching': True,
}, {
'url': 'tvpstream:39821455',
'only_matching': True,
}, {
# the default stream when you provide no channel_id, most probably TVP Info
'url': 'tvpstream:',
'only_matching': True,
}, {
'url': 'https://tvpstream.vod.tvp.pl/',
'only_matching': True,
}]
_PLAYER_BOX_RE = r'<div\s[^>]*id\s*=\s*["\']?tvp_player_box["\']?[^>]+data-%s-id\s*=\s*["\']?(\d+)'
_BUTTON_RE = r'<div\s[^>]*data-channel-id=["\']?%s["\']?[^>]*\sdata-title=(?:"([^"]*)"|\'([^\']*)\')[^>]*\sdata-stationname=(?:"([^"]*)"|\'([^\']*)\')'
def _real_extract(self, url):
channel_id = self._match_id(url)
channel_url = self._proto_relative_url('//tvpstream.vod.tvp.pl/?channel_id=%s' % channel_id or 'default')
webpage = self._download_webpage(channel_url, channel_id, 'Downloading channel webpage')
if not channel_id:
channel_id = self._search_regex(self._PLAYER_BOX_RE % 'channel',
webpage, 'default channel id')
video_id = self._search_regex(self._PLAYER_BOX_RE % 'video',
webpage, 'video id')
audition_title, station_name = self._search_regex(
self._BUTTON_RE % (re.escape(channel_id)), webpage,
'audition title and station name',
group=(1, 2))
return {
'_type': 'url_transparent',
'id': channel_id,
'url': 'tvp:%s' % video_id,
'title': audition_title,
'alt_title': station_name,
'is_live': True,
'ie_key': 'TVPEmbed',
}
class TVPEmbedIE(InfoExtractor):
IE_NAME = 'tvp:embed'
IE_DESC = 'Telewizja Polska'