[extractor/yle_areena] Support restricted videos (#5735)

* and improve metadata

Closes #5734
Authored by: docbender
pull/5757/head
Vita 2 years ago committed by GitHub
parent 3d79ebc8b7
commit b05f0a50e0
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23

@ -1,40 +1,79 @@
from .common import InfoExtractor from .common import InfoExtractor
from .kaltura import KalturaIE from .kaltura import KalturaIE
from ..utils import int_or_none, traverse_obj, url_or_none from ..utils import (
int_or_none,
smuggle_url,
traverse_obj,
unified_strdate,
url_or_none,
)
class YleAreenaIE(InfoExtractor): class YleAreenaIE(InfoExtractor):
_VALID_URL = r'https?://areena\.yle\.fi/(?P<id>[\d-]+)' _VALID_URL = r'https?://areena\.yle\.fi/(?P<id>[\d-]+)'
_TESTS = [{ _TESTS = [
'url': 'https://areena.yle.fi/1-4371942', {
'md5': '932edda0ecf5dfd6423804182d32f8ac', 'url': 'https://areena.yle.fi/1-4371942',
'info_dict': { 'md5': '932edda0ecf5dfd6423804182d32f8ac',
'id': '0_a3tjk92c', 'info_dict': {
'ext': 'mp4', 'id': '0_a3tjk92c',
'title': 'Pouchit', 'ext': 'mp4',
'description': 'md5:d487309c3abbe5650265bbd1742d2f82', 'title': 'Pouchit',
'series': 'Modernit miehet', 'description': 'md5:d487309c3abbe5650265bbd1742d2f82',
'season': 'Season 1', 'series': 'Modernit miehet',
'season_number': 1, 'season': 'Season 1',
'episode': 'Episode 2', 'season_number': 1,
'episode_number': 2, 'episode': 'Episode 2',
'thumbnail': 'http://cfvod.kaltura.com/p/1955031/sp/195503100/thumbnail/entry_id/0_a3tjk92c/version/100061', 'episode_number': 2,
'uploader_id': 'ovp@yle.fi', 'thumbnail': 'http://cfvod.kaltura.com/p/1955031/sp/195503100/thumbnail/entry_id/0_a3tjk92c/version/100061',
'duration': 1435, 'uploader_id': 'ovp@yle.fi',
'view_count': int, 'duration': 1435,
'upload_date': '20181204', 'view_count': int,
'timestamp': 1543916210, 'upload_date': '20181204',
'subtitles': {'fin': [{'url': r're:^https?://', 'ext': 'srt'}]}, 'release_date': '20190106',
'age_limit': 7, 'timestamp': 1543916210,
'subtitles': {'fin': [{'url': r're:^https?://', 'ext': 'srt'}]},
'age_limit': 7,
'webpage_url': 'https://areena.yle.fi/1-4371942'
}
},
{
'url': 'https://areena.yle.fi/1-2158940',
'md5': 'cecb603661004e36af8c5188b5212b12',
'info_dict': {
'id': '1_l38iz9ur',
'ext': 'mp4',
'title': 'Albi haluaa vessan',
'description': 'md5:15236d810c837bed861fae0e88663c33',
'series': 'Albi Lumiukko',
'season': None,
'season_number': None,
'episode': None,
'episode_number': None,
'thumbnail': 'http://cfvod.kaltura.com/p/1955031/sp/195503100/thumbnail/entry_id/1_l38iz9ur/version/100021',
'uploader_id': 'ovp@yle.fi',
'duration': 319,
'view_count': int,
'upload_date': '20211202',
'release_date': '20211215',
'timestamp': 1638448202,
'subtitles': {},
'age_limit': 0,
'webpage_url': 'https://areena.yle.fi/1-2158940'
}
} }
}] ]
def _real_extract(self, url): def _real_extract(self, url):
video_id = self._match_id(url) video_id = self._match_id(url)
info = self._search_json_ld(self._download_webpage(url, video_id), video_id, default={}) info = self._search_json_ld(self._download_webpage(url, video_id), video_id, default={})
video_data = self._download_json( video_data = self._download_json(
f'https://player.api.yle.fi/v1/preview/{video_id}.json?app_id=player_static_prod&app_key=8930d72170e48303cf5f3867780d549b', f'https://player.api.yle.fi/v1/preview/{video_id}.json?app_id=player_static_prod&app_key=8930d72170e48303cf5f3867780d549b',
video_id) video_id, headers={
'origin': 'https://areena.yle.fi',
'referer': 'https://areena.yle.fi/',
'content-type': 'application/json'
})
# Example title: 'K1, J2: Pouchit | Modernit miehet' # Example title: 'K1, J2: Pouchit | Modernit miehet'
series, season_number, episode_number, episode = self._search_regex( series, season_number, episode_number, episode = self._search_regex(
@ -54,7 +93,9 @@ class YleAreenaIE(InfoExtractor):
return { return {
'_type': 'url_transparent', '_type': 'url_transparent',
'url': 'kaltura:1955031:%s' % traverse_obj(video_data, ('data', 'ongoing_ondemand', 'kaltura', 'id')), 'url': smuggle_url(
f'kaltura:1955031:{video_data["data"]["ongoing_ondemand"]["kaltura"]["id"]}',
{'source_url': url}),
'ie_key': KalturaIE.ie_key(), 'ie_key': KalturaIE.ie_key(),
'title': (traverse_obj(video_data, ('data', 'ongoing_ondemand', 'title', 'fin'), expected_type=str) 'title': (traverse_obj(video_data, ('data', 'ongoing_ondemand', 'title', 'fin'), expected_type=str)
or episode or info.get('title')), or episode or info.get('title')),
@ -62,10 +103,11 @@ class YleAreenaIE(InfoExtractor):
'series': (traverse_obj(video_data, ('data', 'ongoing_ondemand', 'series', 'title', 'fin'), expected_type=str) 'series': (traverse_obj(video_data, ('data', 'ongoing_ondemand', 'series', 'title', 'fin'), expected_type=str)
or series), or series),
'season_number': (int_or_none(self._search_regex(r'Kausi (\d+)', description, 'season number', default=None)) 'season_number': (int_or_none(self._search_regex(r'Kausi (\d+)', description, 'season number', default=None))
or int(season_number)), or int_or_none(season_number)),
'episode_number': (traverse_obj(video_data, ('data', 'ongoing_ondemand', 'episode_number'), expected_type=int_or_none) 'episode_number': (traverse_obj(video_data, ('data', 'ongoing_ondemand', 'episode_number'), expected_type=int_or_none)
or int(episode_number)), or int_or_none(episode_number)),
'thumbnails': traverse_obj(info, ('thumbnails', ..., {'url': 'url'})), 'thumbnails': traverse_obj(info, ('thumbnails', ..., {'url': 'url'})),
'age_limit': traverse_obj(video_data, ('data', 'ongoing_ondemand', 'content_rating', 'age_restriction'), expected_type=int_or_none), 'age_limit': traverse_obj(video_data, ('data', 'ongoing_ondemand', 'content_rating', 'age_restriction'), expected_type=int_or_none),
'subtitles': subtitles, 'subtitles': subtitles,
'release_date': unified_strdate(traverse_obj(video_data, ('data', 'ongoing_ondemand', 'start_time'), expected_type=str)),
} }

Loading…
Cancel
Save