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