Upgrade yt_dlp and download script

This commit is contained in:
2025-05-02 16:11:08 -05:00
parent 3a2e8eeb08
commit d68d9ce4f9
1194 changed files with 60099 additions and 44436 deletions

View File

@@ -1,30 +1,30 @@
from datetime import datetime
import base64
import itertools
from .common import InfoExtractor
from ..networking import HEADRequest
from ..utils import int_or_none, urlencode_postdata
from ..utils import int_or_none, traverse_obj, url_or_none, urljoin
class TenPlayIE(InfoExtractor):
_VALID_URL = r'https?://(?:www\.)?10play\.com\.au/(?:[^/]+/)+(?P<id>tpv\d{6}[a-z]{5})'
_NETRC_MACHINE = '10play'
_TESTS = [{
'url': 'https://10play.com.au/neighbours/web-extras/season-39/nathan-borg-is-the-first-aussie-actor-with-a-cochlear-implant-to-join-neighbours/tpv210128qupwd',
'url': 'https://10play.com.au/neighbours/web-extras/season-41/heres-a-first-look-at-mischa-bartons-neighbours-debut/tpv230911hyxnz',
'info_dict': {
'id': '6226844312001',
'id': '6336940246112',
'ext': 'mp4',
'title': 'Nathan Borg Is The First Aussie Actor With A Cochlear Implant To Join Neighbours',
'alt_title': 'Nathan Borg Is The First Aussie Actor With A Cochlear Implant To Join Neighbours',
'description': 'md5:a02d0199c901c2dd4c796f1e7dd0de43',
'duration': 186,
'season': 39,
'title': 'Here\'s A First Look At Mischa Barton\'s Neighbours Debut',
'alt_title': 'Here\'s A First Look At Mischa Barton\'s Neighbours Debut',
'description': 'Neighbours Premieres Monday, September 18 At 4:30pm On 10 And 10 Play And 6:30pm On 10 Peach',
'duration': 74,
'season': 'Season 41',
'season_number': 41,
'series': 'Neighbours',
'thumbnail': r're:https://.*\.jpg',
'uploader': 'Channel 10',
'age_limit': 15,
'timestamp': 1611810000,
'upload_date': '20210128',
'timestamp': 1694386800,
'upload_date': '20230910',
'uploader_id': '2199827728001',
},
'params': {
@@ -32,21 +32,30 @@ class TenPlayIE(InfoExtractor):
},
'skip': 'Only available in Australia',
}, {
'url': 'https://10play.com.au/todd-sampsons-body-hack/episodes/season-4/episode-7/tpv200921kvngh',
'url': 'https://10play.com.au/neighbours/episodes/season-42/episode-9107/tpv240902nzqyp',
'info_dict': {
'id': '6192880312001',
'id': '9000000000091177',
'ext': 'mp4',
'title': "Todd Sampson's Body Hack - S4 Ep. 2",
'description': 'md5:fa278820ad90f08ea187f9458316ac74',
'title': 'Neighbours - S42 Ep. 9107',
'alt_title': 'Thu 05 Sep',
'description': 'md5:37a1f4271be34b9ee2b533426a5fbaef',
'duration': 1388,
'episode': 'Episode 9107',
'episode_number': 9107,
'season': 'Season 42',
'season_number': 42,
'series': 'Neighbours',
'thumbnail': r're:https://.*\.jpg',
'age_limit': 15,
'timestamp': 1600770600,
'upload_date': '20200922',
'timestamp': 1725517860,
'upload_date': '20240905',
'uploader': 'Channel 10',
'uploader_id': '2199827728001'
'uploader_id': '2199827728001',
},
'params': {
'skip_download': True,
}
},
'skip': 'Only available in Australia',
}, {
'url': 'https://10play.com.au/how-to-stay-married/web-extras/season-1/terrys-talks-ep-1-embracing-change/tpv190915ylupc',
'only_matching': True,
@@ -60,56 +69,95 @@ class TenPlayIE(InfoExtractor):
'MA': 15,
'MA15+': 15,
'R': 18,
'X': 18
'X': 18,
}
def _get_bearer_token(self, video_id):
username, password = self._get_login_info()
if username is None or password is None:
self.raise_login_required('Your 10play account\'s details must be provided with --username and --password.')
_timestamp = datetime.now().strftime('%Y%m%d000000')
_auth_header = base64.b64encode(_timestamp.encode('ascii')).decode('ascii')
data = self._download_json('https://10play.com.au/api/user/auth', video_id, 'Getting bearer token', headers={
'X-Network-Ten-Auth': _auth_header,
}, data=urlencode_postdata({
'email': username,
'password': password,
}))
return 'Bearer ' + data['jwt']['accessToken']
def _real_extract(self, url):
content_id = self._match_id(url)
data = self._download_json(
'https://10play.com.au/api/v1/videos/' + content_id, content_id)
headers = {}
if data.get('memberGated') is True:
_token = self._get_bearer_token(content_id)
headers = {'Authorization': _token}
_video_url = self._download_json(
data.get('playbackApiEndpoint'), content_id, 'Downloading video JSON',
headers=headers).get('source')
m3u8_url = self._request_webpage(HEADRequest(
_video_url), content_id).url
video_data = self._download_json(
f'https://vod.ten.com.au/api/videos/bcquery?command=find_videos_by_id&video_id={data["altId"]}',
content_id, 'Downloading video JSON')
m3u8_url = self._request_webpage(
HEADRequest(video_data['items'][0]['HLSURL']),
content_id, 'Checking stream URL').url
if '10play-not-in-oz' in m3u8_url:
self.raise_geo_restricted(countries=['AU'])
# Attempt to get a higher quality stream
m3u8_url = m3u8_url.replace(',150,75,55,0000', ',300,150,75,55,0000')
formats = self._extract_m3u8_formats(m3u8_url, content_id, 'mp4')
return {
'id': content_id,
'formats': formats,
'subtitles': {'en': [{'url': data.get('captionUrl')}]} if data.get('captionUrl') else None,
'id': data.get('altId') or content_id,
'duration': data.get('duration'),
'title': data.get('subtitle'),
'alt_title': data.get('title'),
'description': data.get('description'),
'age_limit': self._AUS_AGES.get(data.get('classification')),
'series': data.get('tvShow'),
'season': int_or_none(data.get('season')),
'episode_number': int_or_none(data.get('episode')),
'timestamp': data.get('published'),
'thumbnail': data.get('imageUrl'),
'subtitles': {'en': [{'url': data['captionUrl']}]} if url_or_none(data.get('captionUrl')) else None,
'uploader': 'Channel 10',
'uploader_id': '2199827728001',
**traverse_obj(data, {
'id': ('altId', {str}),
'duration': ('duration', {int_or_none}),
'title': ('subtitle', {str}),
'alt_title': ('title', {str}),
'description': ('description', {str}),
'age_limit': ('classification', {self._AUS_AGES.get}),
'series': ('tvShow', {str}),
'season_number': ('season', {int_or_none}),
'episode_number': ('episode', {int_or_none}),
'timestamp': ('published', {int_or_none}),
'thumbnail': ('imageUrl', {url_or_none}),
}),
}
class TenPlaySeasonIE(InfoExtractor):
_VALID_URL = r'https?://(?:www\.)?10play\.com\.au/(?P<show>[^/?#]+)/episodes/(?P<season>[^/?#]+)/?(?:$|[?#])'
_TESTS = [{
'url': 'https://10play.com.au/masterchef/episodes/season-14',
'info_dict': {
'title': 'Season 14',
'id': 'MjMyOTIy',
},
'playlist_mincount': 64,
}, {
'url': 'https://10play.com.au/the-bold-and-the-beautiful-fast-tracked/episodes/season-2022',
'info_dict': {
'title': 'Season 2022',
'id': 'Mjc0OTIw',
},
'playlist_mincount': 256,
}]
def _entries(self, load_more_url, display_id=None):
skip_ids = []
for page in itertools.count(1):
episodes_carousel = self._download_json(
load_more_url, display_id, query={'skipIds[]': skip_ids},
note=f'Fetching episodes page {page}')
episodes_chunk = episodes_carousel['items']
skip_ids.extend(ep['id'] for ep in episodes_chunk)
for ep in episodes_chunk:
yield ep['cardLink']
if not episodes_carousel['hasMore']:
break
def _real_extract(self, url):
show, season = self._match_valid_url(url).group('show', 'season')
season_info = self._download_json(
f'https://10play.com.au/api/shows/{show}/episodes/{season}', f'{show}/{season}')
episodes_carousel = traverse_obj(season_info, (
'content', 0, 'components', (
lambda _, v: v['title'].lower() == 'episodes',
(..., {dict}),
)), get_all=False) or {}
playlist_id = episodes_carousel['tpId']
return self.playlist_from_matches(
self._entries(urljoin(url, episodes_carousel['loadMoreUrl']), playlist_id),
playlist_id, traverse_obj(season_info, ('content', 0, 'title', {str})),
getter=urljoin(url))