0
0
Fork 0
mirror of https://github.com/yt-dlp/yt-dlp.git synced 2024-11-27 03:03:01 +00:00
yt-dlp/youtube_dl/extractor/adultswim.py
2019-04-05 11:45:49 +01:00

202 lines
7.7 KiB
Python

# coding: utf-8
from __future__ import unicode_literals
import json
import re
from .turner import TurnerBaseIE
from ..utils import (
determine_ext,
float_or_none,
int_or_none,
mimetype2ext,
parse_age_limit,
parse_iso8601,
strip_or_none,
try_get,
)
class AdultSwimIE(TurnerBaseIE):
_VALID_URL = r'https?://(?:www\.)?adultswim\.com/videos/(?P<show_path>[^/?#]+)(?:/(?P<episode_path>[^/?#]+))?'
_TESTS = [{
'url': 'http://adultswim.com/videos/rick-and-morty/pilot',
'info_dict': {
'id': 'rQxZvXQ4ROaSOqq-or2Mow',
'ext': 'mp4',
'title': 'Rick and Morty - Pilot',
'description': 'Rick moves in with his daughter\'s family and establishes himself as a bad influence on his grandson, Morty.',
'timestamp': 1543294800,
'upload_date': '20181127',
},
'params': {
# m3u8 download
'skip_download': True,
},
'expected_warnings': ['Unable to download f4m manifest'],
}, {
'url': 'http://www.adultswim.com/videos/tim-and-eric-awesome-show-great-job/dr-steve-brule-for-your-wine/',
'info_dict': {
'id': 'sY3cMUR_TbuE4YmdjzbIcQ',
'ext': 'mp4',
'title': 'Tim and Eric Awesome Show Great Job! - Dr. Steve Brule, For Your Wine',
'description': 'Dr. Brule reports live from Wine Country with a special report on wines. \nWatch Tim and Eric Awesome Show Great Job! episode #20, "Embarrassed" on Adult Swim.',
'upload_date': '20080124',
'timestamp': 1201150800,
},
'params': {
# m3u8 download
'skip_download': True,
},
'skip': '404 Not Found',
}, {
'url': 'http://www.adultswim.com/videos/decker/inside-decker-a-new-hero/',
'info_dict': {
'id': 'I0LQFQkaSUaFp8PnAWHhoQ',
'ext': 'mp4',
'title': 'Decker - Inside Decker: A New Hero',
'description': 'The guys recap the conclusion of the season. They announce a new hero, take a peek into the Victorville Film Archive and welcome back the talented James Dean.',
'timestamp': 1469480460,
'upload_date': '20160725',
},
'params': {
# m3u8 download
'skip_download': True,
},
'expected_warnings': ['Unable to download f4m manifest'],
}, {
'url': 'http://www.adultswim.com/videos/attack-on-titan',
'info_dict': {
'id': 'attack-on-titan',
'title': 'Attack on Titan',
'description': 'md5:41caa9416906d90711e31dc00cb7db7e',
},
'playlist_mincount': 12,
}, {
'url': 'http://www.adultswim.com/videos/streams/williams-stream',
'info_dict': {
'id': 'd8DEBj7QRfetLsRgFnGEyg',
'ext': 'mp4',
'title': r're:^Williams Stream \d{4}-\d{2}-\d{2} \d{2}:\d{2}$',
'description': 'original programming',
},
'params': {
# m3u8 download
'skip_download': True,
},
'skip': '404 Not Found',
}]
def _real_extract(self, url):
show_path, episode_path = re.match(self._VALID_URL, url).groups()
display_id = episode_path or show_path
query = '''query {
getShowBySlug(slug:"%s") {
%%s
}
}''' % show_path
if episode_path:
query = query % '''title
getVideoBySlug(slug:"%s") {
_id
auth
description
duration
episodeNumber
launchDate
mediaID
seasonNumber
poster
title
tvRating
}''' % episode_path
['getVideoBySlug']
else:
query = query % '''metaDescription
title
videos(first:1000,sort:["episode_number"]) {
edges {
node {
_id
slug
}
}
}'''
show_data = self._download_json(
'https://www.adultswim.com/api/search', display_id,
data=json.dumps({'query': query}).encode(),
headers={'Content-Type': 'application/json'})['data']['getShowBySlug']
if episode_path:
video_data = show_data['getVideoBySlug']
video_id = video_data['_id']
episode_title = title = video_data['title']
series = show_data.get('title')
if series:
title = '%s - %s' % (series, title)
info = {
'id': video_id,
'title': title,
'description': strip_or_none(video_data.get('description')),
'duration': float_or_none(video_data.get('duration')),
'formats': [],
'subtitles': {},
'age_limit': parse_age_limit(video_data.get('tvRating')),
'thumbnail': video_data.get('poster'),
'timestamp': parse_iso8601(video_data.get('launchDate')),
'series': series,
'season_number': int_or_none(video_data.get('seasonNumber')),
'episode': episode_title,
'episode_number': int_or_none(video_data.get('episodeNumber')),
}
auth = video_data.get('auth')
media_id = video_data.get('mediaID')
if media_id:
info.update(self._extract_ngtv_info(media_id, {
# CDN_TOKEN_APP_ID from:
# https://d2gg02c3xr550i.cloudfront.net/assets/asvp.e9c8bef24322d060ef87.bundle.js
'appId': 'eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJhcHBJZCI6ImFzLXR2ZS1kZXNrdG9wLXB0enQ2bSIsInByb2R1Y3QiOiJ0dmUiLCJuZXR3b3JrIjoiYXMiLCJwbGF0Zm9ybSI6ImRlc2t0b3AiLCJpYXQiOjE1MzI3MDIyNzl9.BzSCk-WYOZ2GMCIaeVb8zWnzhlgnXuJTCu0jGp_VaZE',
}, {
'url': url,
'site_name': 'AdultSwim',
'auth_required': auth,
}))
if not auth:
extract_data = self._download_json(
'https://www.adultswim.com/api/shows/v1/videos/' + video_id,
video_id, query={'fields': 'stream'}, fatal=False) or {}
assets = try_get(extract_data, lambda x: x['data']['video']['stream']['assets'], list) or []
for asset in assets:
asset_url = asset.get('url')
if not asset_url:
continue
ext = determine_ext(asset_url, mimetype2ext(asset.get('mime_type')))
if ext == 'm3u8':
info['formats'].extend(self._extract_m3u8_formats(
asset_url, video_id, 'mp4', m3u8_id='hls', fatal=False))
elif ext == 'f4m':
continue
# info['formats'].extend(self._extract_f4m_formats(
# asset_url, video_id, f4m_id='hds', fatal=False))
elif ext in ('scc', 'ttml', 'vtt'):
info['subtitles'].setdefault('en', []).append({
'url': asset_url,
})
self._sort_formats(info['formats'])
return info
else:
entries = []
for edge in show_data.get('videos', {}).get('edges', []):
video = edge.get('node') or {}
slug = video.get('slug')
if not slug:
continue
entries.append(self.url_result(
'http://adultswim.com/videos/%s/%s' % (show_path, slug),
'AdultSwim', video.get('_id')))
return self.playlist_result(
entries, show_path, show_data.get('title'),
strip_or_none(show_data.get('metaDescription')))