]> jfr.im git - yt-dlp.git/blame - yt_dlp/extractor/viewlift.py
[extractors] Use new framework for existing embeds (#4307)
[yt-dlp.git] / yt_dlp / extractor / viewlift.py
CommitLineData
4877ffc0 1import json
654fd03c 2
03339b7b 3from .common import InfoExtractor
4877ffc0 4from ..compat import compat_HTTPError
654fd03c 5from ..utils import (
9fbfc9bd 6 ExtractorError,
654fd03c 7 int_or_none,
57d67920 8 parse_age_limit,
5be76d1a 9 traverse_obj,
654fd03c 10)
03339b7b 11
654fd03c 12
67167920 13class ViewLiftBaseIE(InfoExtractor):
4877ffc0
RA
14 _API_BASE = 'https://prod-api.viewlift.com/'
15 _DOMAINS_REGEX = r'(?:(?:main\.)?snagfilms|snagxtreme|funnyforfree|kiddovid|winnersview|(?:monumental|lax)sportsnetwork|vayafilm|failarmy|ftfnext|lnppass\.legapallacanestro|moviespree|app\.myoutdoortv|neoufitness|pflmma|theidentitytb)\.com|(?:hoichoi|app\.horseandcountry|kronon|marquee|supercrosslive)\.tv'
16 _SITE_MAP = {
17 'ftfnext': 'lax',
18 'funnyforfree': 'snagfilms',
19 'hoichoi': 'hoichoitv',
20 'kiddovid': 'snagfilms',
21 'laxsportsnetwork': 'lax',
22 'legapallacanestro': 'lnp',
23 'marquee': 'marquee-tv',
24 'monumentalsportsnetwork': 'monumental-network',
25 'moviespree': 'bingeflix',
26 'pflmma': 'pfl',
27 'snagxtreme': 'snagfilms',
28 'theidentitytb': 'tampabay',
29 'vayafilm': 'snagfilms',
30 }
31 _TOKENS = {}
32
5be76d1a 33 def _fetch_token(self, site, url):
34 if self._TOKENS.get(site):
35 return
5be76d1a 36
37 cookies = self._get_cookies(url)
38 if cookies and cookies.get('token'):
39 self._TOKENS[site] = self._search_regex(r'22authorizationToken\%22:\%22([^\%]+)\%22', cookies['token'].value, 'token')
40 if not self._TOKENS.get(site):
41 self.raise_login_required('Cookies (not necessarily logged in) are needed to download from this website', method='cookies')
42
43 def _call_api(self, site, path, video_id, url, query):
44 self._fetch_token(site, url)
45 try:
46 return self._download_json(
47 self._API_BASE + path, video_id, headers={'Authorization': self._TOKENS.get(site)}, query=query)
48 except ExtractorError as e:
49 if isinstance(e.cause, compat_HTTPError) and e.cause.code == 403:
50 webpage = e.cause.read().decode()
51 try:
52 error_message = traverse_obj(json.loads(webpage), 'errorMessage', 'message')
53 except json.JSONDecodeError:
54 raise ExtractorError(f'{site} said: {webpage}', cause=e.cause)
55 if error_message:
56 if 'has not purchased' in error_message:
57 self.raise_login_required(method='cookies')
58 raise ExtractorError(error_message, expected=True)
59 raise
67167920 60
61
62class ViewLiftEmbedIE(ViewLiftBaseIE):
4877ffc0
RA
63 IE_NAME = 'viewlift:embed'
64 _VALID_URL = r'https?://(?:(?:www|embed)\.)?(?P<domain>%s)/embed/player\?.*\bfilmId=(?P<id>[\da-f]{8}-(?:[\da-f]{4}-){3}[\da-f]{12})' % ViewLiftBaseIE._DOMAINS_REGEX
bfd973ec 65 _EMBED_REGEX = [r'<iframe[^>]+?src=(["\'])(?P<url>(?:https?:)?//(?:embed\.)?(?:%s)/embed/player.+?)\1' % ViewLiftBaseIE._DOMAINS_REGEX]
7d7d4690 66 _TESTS = [{
654fd03c
S
67 'url': 'http://embed.snagfilms.com/embed/player?filmId=74849a00-85a9-11e1-9660-123139220831&w=500',
68 'md5': '2924e9215c6eff7a55ed35b72276bd93',
69 'info_dict': {
7d7d4690 70 'id': '74849a00-85a9-11e1-9660-123139220831',
7d7d4690 71 'ext': 'mp4',
72 'title': '#whilewewatch',
4877ffc0
RA
73 'description': 'md5:b542bef32a6f657dadd0df06e26fb0c8',
74 'timestamp': 1334350096,
75 'upload_date': '20120413',
7d7d4690 76 }
496ce6b3
S
77 }, {
78 # invalid labels, 360p is better that 480p
79 'url': 'http://www.snagfilms.com/embed/player?filmId=17ca0950-a74a-11e0-a92a-0026bb61d036',
80 'md5': '882fca19b9eb27ef865efeeaed376a48',
81 'info_dict': {
82 'id': '17ca0950-a74a-11e0-a92a-0026bb61d036',
83 'ext': 'mp4',
84 'title': 'Life in Limbo',
4877ffc0
RA
85 },
86 'skip': 'The video does not exist',
654fd03c
S
87 }, {
88 'url': 'http://www.snagfilms.com/embed/player?filmId=0000014c-de2f-d5d6-abcf-ffef58af0017',
89 'only_matching': True,
7d7d4690 90 }]
03339b7b 91
7e0480ae 92 def _real_extract(self, url):
5ad28e7f 93 domain, film_id = self._match_valid_url(url).groups()
4877ffc0
RA
94 site = domain.split('.')[-2]
95 if site in self._SITE_MAP:
96 site = self._SITE_MAP[site]
5be76d1a 97
98 content_data = self._call_api(
99 site, 'entitlement/video/status', film_id, url, {
100 'id': film_id
101 })['video']
4877ffc0
RA
102 gist = content_data['gist']
103 title = gist['title']
104 video_assets = content_data['streamingInfo']['videoAssets']
9fbfc9bd 105
5be76d1a 106 hls_url = video_assets.get('hls')
107 formats, subtitles = [], {}
108 if hls_url:
109 formats, subtitles = self._extract_m3u8_formats_and_subtitles(
110 hls_url, film_id, 'mp4', 'm3u8_native', m3u8_id='hls', fatal=False)
111
112 for video_asset in video_assets.get('mpeg') or []:
4877ffc0 113 video_asset_url = video_asset.get('url')
5be76d1a 114 if not video_asset_url:
654fd03c 115 continue
4877ffc0
RA
116 bitrate = int_or_none(video_asset.get('bitrate'))
117 height = int_or_none(self._search_regex(
118 r'^_?(\d+)[pP]$', video_asset.get('renditionValue'),
119 'height', default=None))
120 formats.append({
121 'url': video_asset_url,
122 'format_id': 'http%s' % ('-%d' % bitrate if bitrate else ''),
123 'tbr': bitrate,
124 'height': height,
125 'vcodec': video_asset.get('codec'),
126 })
127
5be76d1a 128 subs = {}
129 for sub in traverse_obj(content_data, ('contentDetails', 'closedCaptions')) or []:
130 sub_url = sub.get('url')
131 if not sub_url:
132 continue
133 subs.setdefault(sub.get('language', 'English'), []).append({
134 'url': sub_url,
135 })
4877ffc0 136
5be76d1a 137 self._sort_formats(formats)
138 return {
4877ffc0 139 'id': film_id,
654fd03c 140 'title': title,
4877ffc0
RA
141 'description': gist.get('description'),
142 'thumbnail': gist.get('videoImageUrl'),
143 'duration': int_or_none(gist.get('runtime')),
144 'age_limit': parse_age_limit(content_data.get('parentalRating')),
145 'timestamp': int_or_none(gist.get('publishDate'), 1000),
654fd03c 146 'formats': formats,
5be76d1a 147 'subtitles': self._merge_subtitles(subs, subtitles),
148 'categories': traverse_obj(content_data, ('categories', ..., 'title')),
149 'tags': traverse_obj(content_data, ('tags', ..., 'title')),
654fd03c
S
150 }
151
152
67167920 153class ViewLiftIE(ViewLiftBaseIE):
4877ffc0 154 IE_NAME = 'viewlift'
5be76d1a 155 _API_BASE = 'https://prod-api-cached-2.viewlift.com/'
4877ffc0 156 _VALID_URL = r'https?://(?:www\.)?(?P<domain>%s)(?P<path>(?:/(?:films/title|show|(?:news/)?videos?|watch))?/(?P<id>[^?#]+))' % ViewLiftBaseIE._DOMAINS_REGEX
242a998b 157 _TESTS = [{
654fd03c
S
158 'url': 'http://www.snagfilms.com/films/title/lost_for_life',
159 'md5': '19844f897b35af219773fd63bdec2942',
160 'info_dict': {
161 'id': '0000014c-de2f-d5d6-abcf-ffef58af0017',
162 'display_id': 'lost_for_life',
163 'ext': 'mp4',
164 'title': 'Lost for Life',
57d67920 165 'description': 'md5:ea10b5a50405ae1f7b5269a6ec594102',
ec85ded8 166 'thumbnail': r're:^https?://.*\.jpg',
654fd03c 167 'duration': 4489,
57d67920
RA
168 'categories': 'mincount:3',
169 'age_limit': 14,
170 'upload_date': '20150421',
326ae4ff 171 'timestamp': 1429656820,
654fd03c 172 }
242a998b
S
173 }, {
174 'url': 'http://www.snagfilms.com/show/the_world_cut_project/india',
175 'md5': 'e6292e5b837642bbda82d7f8bf3fbdfd',
176 'info_dict': {
177 'id': '00000145-d75c-d96e-a9c7-ff5c67b20000',
178 'display_id': 'the_world_cut_project/india',
179 'ext': 'mp4',
180 'title': 'India',
181 'description': 'md5:5c168c5a8f4719c146aad2e0dfac6f5f',
ec85ded8 182 'thumbnail': r're:^https?://.*\.jpg',
242a998b 183 'duration': 979,
57d67920
RA
184 'timestamp': 1399478279,
185 'upload_date': '20140507',
242a998b 186 }
326ae4ff
S
187 }, {
188 'url': 'http://main.snagfilms.com/augie_alone/s_2_ep_12_love',
189 'info_dict': {
190 'id': '00000148-7b53-de26-a9fb-fbf306f70020',
191 'display_id': 'augie_alone/s_2_ep_12_love',
192 'ext': 'mp4',
4877ffc0
RA
193 'title': 'S. 2 Ep. 12 - Love',
194 'description': 'Augie finds love.',
326ae4ff
S
195 'thumbnail': r're:^https?://.*\.jpg',
196 'duration': 107,
4877ffc0
RA
197 'upload_date': '20141012',
198 'timestamp': 1413129540,
199 'age_limit': 17,
326ae4ff
S
200 },
201 'params': {
202 'skip_download': True,
203 },
204 }, {
205 'url': 'http://main.snagfilms.com/films/title/the_freebie',
206 'only_matching': True,
a9de9517
S
207 }, {
208 # Film is not playable in your area.
209 'url': 'http://www.snagfilms.com/films/title/inside_mecca',
210 'only_matching': True,
211 }, {
212 # Film is not available.
213 'url': 'http://www.snagfilms.com/show/augie_alone/flirting',
214 'only_matching': True,
67167920 215 }, {
216 'url': 'http://www.winnersview.com/videos/the-good-son',
217 'only_matching': True,
28bab133
YCH
218 }, {
219 # Was once Kaltura embed
220 'url': 'https://www.monumentalsportsnetwork.com/videos/john-carlson-postgame-2-25-15',
221 'only_matching': True,
4877ffc0
RA
222 }, {
223 'url': 'https://www.marquee.tv/watch/sadlerswells-sacredmonsters',
224 'only_matching': True,
ab630a57 225 }, { # Free film with langauge code
226 'url': 'https://www.hoichoi.tv/bn/films/title/shuyopoka',
227 'info_dict': {
228 'id': '7a7a9d33-1f4c-4771-9173-ee4fb6dbf196',
229 'ext': 'mp4',
230 'title': 'Shuyopoka',
231 'description': 'md5:e28f2fb8680096a69c944d37c1fa5ffc',
232 'thumbnail': r're:^https?://.*\.jpg$',
233 'upload_date': '20211006',
234 'series': None
235 },
236 'params': {'skip_download': True},
237 }, { # Free film
238 'url': 'https://www.hoichoi.tv/films/title/dadu-no1',
239 'info_dict': {
240 'id': '0000015b-b009-d126-a1db-b81ff3780000',
241 'ext': 'mp4',
242 'title': 'Dadu No.1',
243 'description': 'md5:605cba408e51a79dafcb824bdeded51e',
244 'thumbnail': r're:^https?://.*\.jpg$',
245 'upload_date': '20210827',
246 'series': None
247 },
248 'params': {'skip_download': True},
249 }, { # Free episode
250 'url': 'https://www.hoichoi.tv/webseries/case-jaundice-s01-e01',
251 'info_dict': {
252 'id': 'f779e07c-30c8-459c-8612-5a834ab5e5ba',
253 'ext': 'mp4',
254 'title': 'Humans Vs. Corona',
255 'description': 'md5:ca30a682b4528d02a3eb6d0427dd0f87',
256 'thumbnail': r're:^https?://.*\.jpg$',
257 'upload_date': '20210830',
258 'series': 'Case Jaundice'
259 },
260 'params': {'skip_download': True},
261 }, { # Free video
262 'url': 'https://www.hoichoi.tv/videos/1549072415320-six-episode-02-hindi',
263 'info_dict': {
264 'id': 'b41fa1ce-aca6-47b6-b208-283ff0a2de30',
265 'ext': 'mp4',
266 'title': 'Woman in red - Hindi',
267 'description': 'md5:9d21edc1827d32f8633eb67c2054fc31',
268 'thumbnail': r're:^https?://.*\.jpg$',
269 'upload_date': '20211006',
270 'series': 'Six (Hindi)'
271 },
272 'params': {'skip_download': True},
273 }, { # Free episode
274 'url': 'https://www.hoichoi.tv/shows/watch-asian-paints-moner-thikana-online-season-1-episode-1',
275 'info_dict': {
276 'id': '1f45d185-8500-455c-b88d-13252307c3eb',
277 'ext': 'mp4',
278 'title': 'Jisshu Sengupta',
279 'description': 'md5:ef6ffae01a3d83438597367400f824ed',
280 'thumbnail': r're:^https?://.*\.jpg$',
281 'upload_date': '20211004',
282 'series': 'Asian Paints Moner Thikana'
283 },
284 'params': {'skip_download': True},
285 }, { # Free series
286 'url': 'https://www.hoichoi.tv/shows/watch-moner-thikana-bengali-web-series-online',
287 'playlist_mincount': 5,
288 'info_dict': {
289 'id': 'watch-moner-thikana-bengali-web-series-online',
290 },
291 }, { # Premium series
292 'url': 'https://www.hoichoi.tv/shows/watch-byomkesh-bengali-web-series-online',
293 'playlist_mincount': 14,
294 'info_dict': {
295 'id': 'watch-byomkesh-bengali-web-series-online',
296 },
297 }, { # Premium movie
298 'url': 'https://www.hoichoi.tv/movies/detective-2020',
299 'only_matching': True
242a998b 300 }]
654fd03c 301
2906631e
S
302 @classmethod
303 def suitable(cls, url):
304 return False if ViewLiftEmbedIE.suitable(url) else super(ViewLiftIE, cls).suitable(url)
305
5be76d1a 306 def _show_entries(self, domain, seasons):
307 for season in seasons:
308 for episode in season.get('episodes') or []:
309 path = traverse_obj(episode, ('gist', 'permalink'))
310 if path:
311 yield self.url_result(f'https://www.{domain}{path}', ie=self.ie_key())
312
654fd03c 313 def _real_extract(self, url):
5ad28e7f 314 domain, path, display_id = self._match_valid_url(url).groups()
4877ffc0
RA
315 site = domain.split('.')[-2]
316 if site in self._SITE_MAP:
317 site = self._SITE_MAP[site]
318 modules = self._call_api(
5be76d1a 319 site, 'content/pages', display_id, url, {
4877ffc0
RA
320 'includeContent': 'true',
321 'moduleOffset': 1,
322 'path': path,
323 'site': site,
324 })['modules']
5be76d1a 325
326 seasons = next((m['contentData'][0]['seasons'] for m in modules if m.get('moduleType') == 'ShowDetailModule'), None)
327 if seasons:
328 return self.playlist_result(self._show_entries(domain, seasons), display_id)
329
4877ffc0
RA
330 film_id = next(m['contentData'][0]['gist']['id'] for m in modules if m.get('moduleType') == 'VideoDetailModule')
331 return {
332 '_type': 'url_transparent',
333 'url': 'http://%s/embed/player?filmId=%s' % (domain, film_id),
334 'id': film_id,
335 'display_id': display_id,
336 'ie_key': 'ViewLiftEmbed',
337 }