]>
Commit | Line | Data |
---|---|---|
df5ae3eb S |
1 | import re |
2 | ||
1309b396 | 3 | from .common import InfoExtractor |
8e4d3f83 | 4 | from ..compat import compat_str |
1309b396 PH |
5 | from ..utils import ( |
6 | determine_ext, | |
e4f90ea0 | 7 | dict_get, |
23bdae09 | 8 | int_or_none, |
a0566bbf | 9 | unified_timestamp, |
43e79947 | 10 | str_or_none, |
7b393f9c | 11 | strip_or_none, |
23bdae09 | 12 | try_get, |
1309b396 PH |
13 | ) |
14 | ||
15 | ||
79998cd5 | 16 | class SVTBaseIE(InfoExtractor): |
4248dad9 | 17 | _GEO_COUNTRIES = ['SE'] |
6d4c2597 | 18 | |
23bdae09 | 19 | def _extract_video(self, video_info, video_id): |
488ff2dd | 20 | is_live = dict_get(video_info, ('live', 'simulcast'), default=False) |
21 | m3u8_protocol = 'm3u8' if is_live else 'm3u8_native' | |
1309b396 | 22 | formats = [] |
3f047fc4 | 23 | subtitles = {} |
1309b396 | 24 | for vr in video_info['videoReferences']: |
21d21b0c | 25 | player_type = vr.get('playerType') or vr.get('format') |
1309b396 | 26 | vurl = vr['url'] |
df5ae3eb S |
27 | ext = determine_ext(vurl) |
28 | if ext == 'm3u8': | |
3f047fc4 | 29 | fmts, subs = self._extract_m3u8_formats_and_subtitles( |
1309b396 | 30 | vurl, video_id, |
488ff2dd | 31 | ext='mp4', entry_protocol=m3u8_protocol, |
3f047fc4 F |
32 | m3u8_id=player_type, fatal=False) |
33 | formats.extend(fmts) | |
34 | self._merge_subtitles(subs, target=subtitles) | |
df5ae3eb S |
35 | elif ext == 'f4m': |
36 | formats.extend(self._extract_f4m_formats( | |
37 | vurl + '?hdcore=3.3.0', video_id, | |
edfd9351 | 38 | f4m_id=player_type, fatal=False)) |
39 | elif ext == 'mpd': | |
3f047fc4 F |
40 | fmts, subs = self._extract_mpd_formats_and_subtitles( |
41 | vurl, video_id, mpd_id=player_type, fatal=False) | |
42 | formats.extend(fmts) | |
43 | self._merge_subtitles(subs, target=subtitles) | |
1309b396 PH |
44 | else: |
45 | formats.append({ | |
edfd9351 | 46 | 'format_id': player_type, |
1309b396 PH |
47 | 'url': vurl, |
48 | }) | |
a0566bbf | 49 | rights = try_get(video_info, lambda x: x['rights'], dict) or {} |
50 | if not formats and rights.get('geoBlockedSweden'): | |
04d906ea | 51 | self.raise_geo_restricted( |
4248dad9 | 52 | 'This video is only available in Sweden', |
b7da73eb | 53 | countries=self._GEO_COUNTRIES, metadata_available=True) |
1309b396 | 54 | |
e4f90ea0 | 55 | subtitle_references = dict_get(video_info, ('subtitles', 'subtitleReferences')) |
594c4d79 S |
56 | if isinstance(subtitle_references, list): |
57 | for sr in subtitle_references: | |
58 | subtitle_url = sr.get('url') | |
e4f90ea0 | 59 | subtitle_lang = sr.get('language', 'sv') |
594c4d79 | 60 | if subtitle_url: |
3f047fc4 F |
61 | sub = { |
62 | 'url': subtitle_url, | |
63 | } | |
e4f90ea0 | 64 | if determine_ext(subtitle_url) == 'm3u8': |
3f047fc4 F |
65 | # XXX: no way of testing, is it ever hit? |
66 | sub['ext'] = 'vtt' | |
67 | subtitles.setdefault(subtitle_lang, []).append(sub) | |
1f16b958 | 68 | |
23bdae09 S |
69 | title = video_info.get('title') |
70 | ||
71 | series = video_info.get('programTitle') | |
72 | season_number = int_or_none(video_info.get('season')) | |
73 | episode = video_info.get('episodeTitle') | |
74 | episode_number = int_or_none(video_info.get('episodeNumber')) | |
75 | ||
a0566bbf | 76 | timestamp = unified_timestamp(rights.get('validFrom')) |
23bdae09 S |
77 | duration = int_or_none(dict_get(video_info, ('materialLength', 'contentDuration'))) |
78 | age_limit = None | |
79 | adult = dict_get( | |
80 | video_info, ('inappropriateForChildren', 'blockedForChildren'), | |
81 | skip_false_values=False) | |
82 | if adult is not None: | |
83 | age_limit = 18 if adult else 0 | |
1309b396 PH |
84 | |
85 | return { | |
86 | 'id': video_id, | |
23bdae09 | 87 | 'title': title, |
1309b396 | 88 | 'formats': formats, |
1f16b958 | 89 | 'subtitles': subtitles, |
1309b396 | 90 | 'duration': duration, |
a0566bbf | 91 | 'timestamp': timestamp, |
df5ae3eb | 92 | 'age_limit': age_limit, |
23bdae09 S |
93 | 'series': series, |
94 | 'season_number': season_number, | |
95 | 'episode': episode, | |
96 | 'episode_number': episode_number, | |
488ff2dd | 97 | 'is_live': is_live, |
1309b396 | 98 | } |
79998cd5 S |
99 | |
100 | ||
101 | class SVTIE(SVTBaseIE): | |
102 | _VALID_URL = r'https?://(?:www\.)?svt\.se/wd\?(?:.*?&)?widgetId=(?P<widget_id>\d+)&.*?\barticleId=(?P<id>\d+)' | |
bfd973ec | 103 | _EMBED_REGEX = [r'(?:<iframe src|href)="(?P<url>%s[^"]*)"' % _VALID_URL] |
79998cd5 S |
104 | _TEST = { |
105 | 'url': 'http://www.svt.se/wd?widgetId=23991§ionId=541&articleId=2900353&type=embed&contextSectionId=123&autostart=false', | |
e4f90ea0 | 106 | 'md5': '33e9a5d8f646523ce0868ecfb0eed77d', |
79998cd5 S |
107 | 'info_dict': { |
108 | 'id': '2900353', | |
e4f90ea0 YCH |
109 | 'ext': 'mp4', |
110 | 'title': 'Stjärnorna skojar till det - under SVT-intervjun', | |
79998cd5 S |
111 | 'duration': 27, |
112 | 'age_limit': 0, | |
113 | }, | |
114 | } | |
115 | ||
116 | def _real_extract(self, url): | |
5ad28e7f | 117 | mobj = self._match_valid_url(url) |
79998cd5 S |
118 | widget_id = mobj.group('widget_id') |
119 | article_id = mobj.group('id') | |
e4f90ea0 YCH |
120 | |
121 | info = self._download_json( | |
79998cd5 S |
122 | 'http://www.svt.se/wd?widgetId=%s&articleId=%s&format=json&type=embed&output=json' % (widget_id, article_id), |
123 | article_id) | |
124 | ||
23bdae09 | 125 | info_dict = self._extract_video(info['video'], article_id) |
e4f90ea0 YCH |
126 | info_dict['title'] = info['context']['title'] |
127 | return info_dict | |
128 | ||
79998cd5 | 129 | |
1236ac6b S |
130 | class SVTPlayBaseIE(SVTBaseIE): |
131 | _SVTPLAY_RE = r'root\s*\[\s*(["\'])_*svtplay\1\s*\]\s*=\s*(?P<json>{.+?})\s*;\s*\n' | |
132 | ||
133 | ||
134 | class SVTPlayIE(SVTPlayBaseIE): | |
79998cd5 | 135 | IE_DESC = 'SVT Play and Öppet arkiv' |
7b393f9c S |
136 | _VALID_URL = r'''(?x) |
137 | (?: | |
a0566bbf | 138 | (?: |
139 | svt:| | |
140 | https?://(?:www\.)?svt\.se/barnkanalen/barnplay/[^/]+/ | |
141 | ) | |
142 | (?P<svt_id>[^/?#&]+)| | |
7b393f9c | 143 | https?://(?:www\.)?(?:svtplay|oppetarkiv)\.se/(?:video|klipp|kanaler)/(?P<id>[^/?#&]+) |
41d1cca3 | 144 | (?:.*?(?:modalId|id)=(?P<modal_id>[\da-zA-Z-]+))? |
7b393f9c S |
145 | ) |
146 | ''' | |
23bdae09 | 147 | _TESTS = [{ |
421a4595 | 148 | 'url': 'https://www.svtplay.se/video/30479064', |
a0566bbf | 149 | 'md5': '2382036fd6f8c994856c323fe51c426e', |
79998cd5 | 150 | 'info_dict': { |
421a4595 | 151 | 'id': '8zVbDPA', |
594c4d79 | 152 | 'ext': 'mp4', |
421a4595 | 153 | 'title': 'Designdrömmar i Stenungsund', |
154 | 'timestamp': 1615770000, | |
155 | 'upload_date': '20210315', | |
156 | 'duration': 3519, | |
a0566bbf | 157 | 'thumbnail': r're:^https?://(?:.*[\.-]jpg|www.svtstatic.se/image/.*)$', |
79998cd5 | 158 | 'age_limit': 0, |
594c4d79 S |
159 | 'subtitles': { |
160 | 'sv': [{ | |
a0566bbf | 161 | 'ext': 'vtt', |
594c4d79 S |
162 | }] |
163 | }, | |
79998cd5 | 164 | }, |
a0566bbf | 165 | 'params': { |
a0566bbf | 166 | # skip for now due to download test asserts that segment is > 10000 bytes and svt uses |
167 | # init segments that are smaller | |
168 | # AssertionError: Expected test_SVTPlay_jNwpV9P.mp4 to be at least 9.77KiB, but it's only 864.00B | |
169 | 'skip_download': True, | |
170 | }, | |
421a4595 | 171 | }, { |
172 | 'url': 'https://www.svtplay.se/video/30479064/husdrommar/husdrommar-sasong-8-designdrommar-i-stenungsund?modalId=8zVbDPA', | |
173 | 'only_matching': True, | |
41d1cca3 | 174 | }, { |
175 | 'url': 'https://www.svtplay.se/video/30684086/rapport/rapport-24-apr-18-00-7?id=e72gVpa', | |
176 | 'only_matching': True, | |
23bdae09 S |
177 | }, { |
178 | # geo restricted to Sweden | |
179 | 'url': 'http://www.oppetarkiv.se/video/5219710/trollflojten', | |
180 | 'only_matching': True, | |
3b34ab53 S |
181 | }, { |
182 | 'url': 'http://www.svtplay.se/klipp/9023742/stopptid-om-bjorn-borg', | |
183 | 'only_matching': True, | |
488ff2dd | 184 | }, { |
185 | 'url': 'https://www.svtplay.se/kanaler/svt1', | |
186 | 'only_matching': True, | |
7b393f9c S |
187 | }, { |
188 | 'url': 'svt:1376446-003A', | |
189 | 'only_matching': True, | |
190 | }, { | |
191 | 'url': 'svt:14278044', | |
192 | 'only_matching': True, | |
a0566bbf | 193 | }, { |
194 | 'url': 'https://www.svt.se/barnkanalen/barnplay/kar/eWv5MLX/', | |
195 | 'only_matching': True, | |
196 | }, { | |
197 | 'url': 'svt:eWv5MLX', | |
198 | 'only_matching': True, | |
23bdae09 | 199 | }] |
e4f90ea0 | 200 | |
7b393f9c S |
201 | def _extract_by_video_id(self, video_id, webpage=None): |
202 | data = self._download_json( | |
e6a25fea | 203 | 'https://api.svt.se/videoplayer-api/video/%s' % video_id, |
7b393f9c S |
204 | video_id, headers=self.geo_verification_headers()) |
205 | info_dict = self._extract_video(data, video_id) | |
206 | if not info_dict.get('title'): | |
207 | title = dict_get(info_dict, ('episode', 'series')) | |
208 | if not title and webpage: | |
209 | title = re.sub( | |
210 | r'\s*\|\s*.+?$', '', self._og_search_title(webpage)) | |
211 | if not title: | |
212 | title = video_id | |
213 | info_dict['title'] = title | |
7b393f9c S |
214 | return info_dict |
215 | ||
79998cd5 | 216 | def _real_extract(self, url): |
5ad28e7f | 217 | mobj = self._match_valid_url(url) |
421a4595 | 218 | video_id = mobj.group('id') |
219 | svt_id = mobj.group('svt_id') or mobj.group('modal_id') | |
7b393f9c S |
220 | |
221 | if svt_id: | |
222 | return self._extract_by_video_id(svt_id) | |
e4f90ea0 YCH |
223 | |
224 | webpage = self._download_webpage(url, video_id) | |
225 | ||
23bdae09 S |
226 | data = self._parse_json( |
227 | self._search_regex( | |
1236ac6b S |
228 | self._SVTPLAY_RE, webpage, 'embedded data', default='{}', |
229 | group='json'), | |
23bdae09 | 230 | video_id, fatal=False) |
e4f90ea0 YCH |
231 | |
232 | thumbnail = self._og_search_thumbnail(webpage) | |
233 | ||
23bdae09 S |
234 | if data: |
235 | video_info = try_get( | |
236 | data, lambda x: x['context']['dispatcher']['stores']['VideoTitlePageStore']['data']['video'], | |
237 | dict) | |
238 | if video_info: | |
239 | info_dict = self._extract_video(video_info, video_id) | |
240 | info_dict.update({ | |
241 | 'title': data['context']['dispatcher']['stores']['MetaStore']['title'], | |
242 | 'thumbnail': thumbnail, | |
243 | }) | |
244 | return info_dict | |
245 | ||
5ed05f26 S |
246 | svt_id = try_get( |
247 | data, lambda x: x['statistics']['dataLake']['content']['id'], | |
248 | compat_str) | |
249 | ||
250 | if not svt_id: | |
251 | svt_id = self._search_regex( | |
252 | (r'<video[^>]+data-video-id=["\']([\da-zA-Z-]+)', | |
41d1cca3 | 253 | r'<[^>]+\bdata-rt=["\']top-area-play-button["\'][^>]+\bhref=["\'][^"\']*video/%s/[^"\']*\b(?:modalId|id)=([\da-zA-Z-]+)' % re.escape(video_id), |
95c98100 | 254 | r'["\']videoSvtId["\']\s*:\s*["\']([\da-zA-Z-]+)', |
2181983a | 255 | r'["\']videoSvtId\\?["\']\s*:\s*\\?["\']([\da-zA-Z-]+)', |
95c98100 | 256 | r'"content"\s*:\s*{.*?"id"\s*:\s*"([\da-zA-Z-]+)"', |
2181983a | 257 | r'["\']svtId["\']\s*:\s*["\']([\da-zA-Z-]+)', |
258 | r'["\']svtId\\?["\']\s*:\s*\\?["\']([\da-zA-Z-]+)'), | |
5ed05f26 | 259 | webpage, 'video id') |
23bdae09 | 260 | |
a0566bbf | 261 | info_dict = self._extract_by_video_id(svt_id, webpage) |
262 | info_dict['thumbnail'] = thumbnail | |
263 | ||
264 | return info_dict | |
fd97fa7b MW |
265 | |
266 | ||
1236ac6b | 267 | class SVTSeriesIE(SVTPlayBaseIE): |
8e4d3f83 | 268 | _VALID_URL = r'https?://(?:www\.)?svtplay\.se/(?P<id>[^/?&#]+)(?:.+?\btab=(?P<season_slug>[^&#]+))?' |
fd97fa7b MW |
269 | _TESTS = [{ |
270 | 'url': 'https://www.svtplay.se/rederiet', | |
271 | 'info_dict': { | |
8e4d3f83 | 272 | 'id': '14445680', |
fd97fa7b | 273 | 'title': 'Rederiet', |
8e4d3f83 | 274 | 'description': 'md5:d9fdfff17f5d8f73468176ecd2836039', |
fd97fa7b MW |
275 | }, |
276 | 'playlist_mincount': 318, | |
df146eb2 | 277 | }, { |
8e4d3f83 | 278 | 'url': 'https://www.svtplay.se/rederiet?tab=season-2-14445680', |
df146eb2 | 279 | 'info_dict': { |
8e4d3f83 | 280 | 'id': 'season-2-14445680', |
df146eb2 | 281 | 'title': 'Rederiet - Säsong 2', |
8e4d3f83 | 282 | 'description': 'md5:d9fdfff17f5d8f73468176ecd2836039', |
df146eb2 | 283 | }, |
8e4d3f83 | 284 | 'playlist_mincount': 12, |
fd97fa7b MW |
285 | }] |
286 | ||
287 | @classmethod | |
288 | def suitable(cls, url): | |
b71bb3ba | 289 | return False if SVTIE.suitable(url) or SVTPlayIE.suitable(url) else super(SVTSeriesIE, cls).suitable(url) |
fd97fa7b MW |
290 | |
291 | def _real_extract(self, url): | |
5ad28e7f | 292 | series_slug, season_id = self._match_valid_url(url).groups() |
8e4d3f83 RA |
293 | |
294 | series = self._download_json( | |
295 | 'https://api.svt.se/contento/graphql', series_slug, | |
296 | 'Downloading series page', query={ | |
297 | 'query': '''{ | |
298 | listablesBySlug(slugs: ["%s"]) { | |
299 | associatedContent(include: [productionPeriod, season]) { | |
300 | items { | |
301 | item { | |
302 | ... on Episode { | |
303 | videoSvtId | |
304 | } | |
305 | } | |
306 | } | |
307 | id | |
308 | name | |
309 | } | |
310 | id | |
311 | longDescription | |
312 | name | |
313 | shortDescription | |
314 | } | |
315 | }''' % series_slug, | |
316 | })['data']['listablesBySlug'][0] | |
df146eb2 S |
317 | |
318 | season_name = None | |
fd97fa7b MW |
319 | |
320 | entries = [] | |
8e4d3f83 | 321 | for season in series['associatedContent']: |
df146eb2 S |
322 | if not isinstance(season, dict): |
323 | continue | |
8e4d3f83 RA |
324 | if season_id: |
325 | if season.get('id') != season_id: | |
df146eb2 S |
326 | continue |
327 | season_name = season.get('name') | |
8e4d3f83 RA |
328 | items = season.get('items') |
329 | if not isinstance(items, list): | |
fd97fa7b | 330 | continue |
8e4d3f83 RA |
331 | for item in items: |
332 | video = item.get('item') or {} | |
333 | content_id = video.get('videoSvtId') | |
334 | if not content_id or not isinstance(content_id, compat_str): | |
fd97fa7b | 335 | continue |
8e4d3f83 RA |
336 | entries.append(self.url_result( |
337 | 'svt:' + content_id, SVTPlayIE.ie_key(), content_id)) | |
b71bb3ba | 338 | |
8e4d3f83 RA |
339 | title = series.get('name') |
340 | season_name = season_name or season_id | |
df146eb2 S |
341 | |
342 | if title and season_name: | |
343 | title = '%s - %s' % (title, season_name) | |
8e4d3f83 RA |
344 | elif season_id: |
345 | title = season_id | |
df146eb2 | 346 | |
fd97fa7b | 347 | return self.playlist_result( |
8e4d3f83 RA |
348 | entries, season_id or series.get('id'), title, |
349 | dict_get(series, ('longDescription', 'shortDescription'))) | |
7b393f9c S |
350 | |
351 | ||
352 | class SVTPageIE(InfoExtractor): | |
43e79947 | 353 | _VALID_URL = r'https?://(?:www\.)?svt\.se/(?P<path>(?:[^/]+/)*(?P<id>[^/?&#]+))' |
7b393f9c | 354 | _TESTS = [{ |
43e79947 | 355 | 'url': 'https://www.svt.se/sport/ishockey/bakom-masken-lehners-kamp-mot-mental-ohalsa', |
7b393f9c | 356 | 'info_dict': { |
43e79947 RA |
357 | 'id': '25298267', |
358 | 'title': 'Bakom masken – Lehners kamp mot mental ohälsa', | |
7b393f9c | 359 | }, |
43e79947 | 360 | 'playlist_count': 4, |
7b393f9c | 361 | }, { |
43e79947 | 362 | 'url': 'https://www.svt.se/nyheter/utrikes/svenska-andrea-ar-en-mil-fran-branderna-i-kalifornien', |
7b393f9c | 363 | 'info_dict': { |
43e79947 RA |
364 | 'id': '24243746', |
365 | 'title': 'Svenska Andrea redo att fly sitt hem i Kalifornien', | |
7b393f9c | 366 | }, |
43e79947 | 367 | 'playlist_count': 2, |
7b393f9c S |
368 | }, { |
369 | # only programTitle | |
370 | 'url': 'http://www.svt.se/sport/ishockey/jagr-tacklar-giroux-under-intervjun', | |
371 | 'info_dict': { | |
43e79947 | 372 | 'id': '8439V2K', |
7b393f9c S |
373 | 'ext': 'mp4', |
374 | 'title': 'Stjärnorna skojar till det - under SVT-intervjun', | |
375 | 'duration': 27, | |
376 | 'age_limit': 0, | |
377 | }, | |
378 | }, { | |
379 | 'url': 'https://www.svt.se/nyheter/lokalt/vast/svt-testar-tar-nagon-upp-skrapet-1', | |
380 | 'only_matching': True, | |
381 | }, { | |
382 | 'url': 'https://www.svt.se/vader/manadskronikor/maj2018', | |
383 | 'only_matching': True, | |
384 | }] | |
385 | ||
386 | @classmethod | |
387 | def suitable(cls, url): | |
a0566bbf | 388 | return False if SVTIE.suitable(url) or SVTPlayIE.suitable(url) else super(SVTPageIE, cls).suitable(url) |
7b393f9c S |
389 | |
390 | def _real_extract(self, url): | |
5ad28e7f | 391 | path, display_id = self._match_valid_url(url).groups() |
7b393f9c | 392 | |
43e79947 RA |
393 | article = self._download_json( |
394 | 'https://api.svt.se/nss-api/page/' + path, display_id, | |
395 | query={'q': 'articles'})['articles']['content'][0] | |
7b393f9c | 396 | |
43e79947 | 397 | entries = [] |
7b393f9c | 398 | |
43e79947 RA |
399 | def _process_content(content): |
400 | if content.get('_type') in ('VIDEOCLIP', 'VIDEOEPISODE'): | |
401 | video_id = compat_str(content['image']['svtId']) | |
402 | entries.append(self.url_result( | |
403 | 'svt:' + video_id, SVTPlayIE.ie_key(), video_id)) | |
7b393f9c | 404 | |
43e79947 RA |
405 | for media in article.get('media', []): |
406 | _process_content(media) | |
407 | ||
408 | for obj in article.get('structuredBody', []): | |
409 | _process_content(obj.get('content') or {}) | |
410 | ||
411 | return self.playlist_result( | |
412 | entries, str_or_none(article.get('id')), | |
413 | strip_or_none(article.get('title'))) |