]>
Commit | Line | Data |
---|---|---|
78149a96 MK |
1 | import re |
2 | ||
3 | from .common import InfoExtractor | |
d50116b8 | 4 | from .xstream import XstreamIE |
4d454c5e S |
5 | from ..utils import ( |
6 | ExtractorError, | |
7 | float_or_none, | |
320d597c | 8 | try_get, |
4d454c5e | 9 | ) |
78149a96 | 10 | |
78149a96 | 11 | |
6368e2e6 | 12 | class VGTVIE(XstreamIE): # XXX: Do not subclass from concrete IE |
804afc58 | 13 | IE_DESC = 'VGTV, BTTV, FTV, Aftenposten and Aftonbladet' |
4248dad9 | 14 | _GEO_BYPASS = False |
41c3b34b | 15 | |
16 | _HOST_TO_APPNAME = { | |
dcbf7394 | 17 | 'tv.vg.no': 'vgtv', |
41c3b34b | 18 | 'vgtv.no': 'vgtv', |
19 | 'bt.no/tv': 'bttv', | |
20 | 'aftenbladet.no/tv': 'satv', | |
21 | 'fvn.no/fvntv': 'fvntv', | |
22 | 'aftenposten.no/webtv': 'aptv', | |
7177fd24 | 23 | 'ap.vgtv.no/webtv': 'aptv', |
1418a043 | 24 | 'tv.aftonbladet.se': 'abtv', |
25 | # obsolete URL schemas, kept in order to save one HTTP redirect | |
b0582fc8 | 26 | 'tv.aftonbladet.se/abtv': 'abtv', |
721a877d | 27 | 'www.aftonbladet.se/tv': 'abtv', |
41c3b34b | 28 | } |
29 | ||
30 | _APP_NAME_TO_VENDOR = { | |
31 | 'vgtv': 'vgtv', | |
32 | 'bttv': 'bt', | |
33 | 'satv': 'sa', | |
34 | 'fvntv': 'fvn', | |
35 | 'aptv': 'ap', | |
b0582fc8 | 36 | 'abtv': 'ab', |
41c3b34b | 37 | } |
38 | ||
0ceab847 | 39 | _VALID_URL = r'''(?x) |
41c3b34b | 40 | (?:https?://(?:www\.)? |
804afc58 | 41 | (?P<host> |
41c3b34b | 42 | %s |
0ceab847 | 43 | ) |
7177fd24 | 44 | /? |
0ceab847 | 45 | (?: |
cbf85239 | 46 | (?:\#!/)?(?:video|live)/| |
b0582fc8 | 47 | embed?.*id=| |
721a877d | 48 | a(?:rticles)?/ |
41c3b34b | 49 | )| |
50 | (?P<appname> | |
51 | %s | |
52 | ):) | |
9f6b5176 | 53 | (?P<id>\d+) |
41c3b34b | 54 | ''' % ('|'.join(_HOST_TO_APPNAME.keys()), '|'.join(_APP_NAME_TO_VENDOR.keys())) |
55 | ||
321c1e44 S |
56 | _TESTS = [ |
57 | { | |
58 | # streamType: vod | |
59 | 'url': 'http://www.vgtv.no/#!/video/84196/hevnen-er-soet-episode-10-abu', | |
60 | 'md5': 'b8be7a234cebb840c0d512c78013e02f', | |
61 | 'info_dict': { | |
62 | 'id': '84196', | |
63 | 'ext': 'mp4', | |
eecd6a46 | 64 | 'title': 'Hevnen er søt: Episode 10 - Abu', |
321c1e44 | 65 | 'description': 'md5:e25e4badb5f544b04341e14abdc72234', |
ec85ded8 | 66 | 'thumbnail': r're:^https?://.*\.jpg', |
321c1e44 S |
67 | 'duration': 648.000, |
68 | 'timestamp': 1404626400, | |
3fbeb95e S |
69 | 'upload_date': '20140706', |
70 | 'view_count': int, | |
321c1e44 S |
71 | }, |
72 | }, | |
73 | { | |
74 | # streamType: wasLive | |
75 | 'url': 'http://www.vgtv.no/#!/live/100764/opptak-vgtv-foelger-em-kvalifiseringen', | |
76 | 'info_dict': { | |
77 | 'id': '100764', | |
b7bb0df2 | 78 | 'ext': 'flv', |
321c1e44 S |
79 | 'title': 'OPPTAK: VGTV følger EM-kvalifiseringen', |
80 | 'description': 'md5:3772d9c0dc2dff92a886b60039a7d4d3', | |
ec85ded8 | 81 | 'thumbnail': r're:^https?://.*\.jpg', |
eecd6a46 | 82 | 'duration': 9103.0, |
321c1e44 | 83 | 'timestamp': 1410113864, |
3fbeb95e S |
84 | 'upload_date': '20140907', |
85 | 'view_count': int, | |
321c1e44 S |
86 | }, |
87 | 'params': { | |
88 | # m3u8 download | |
89 | 'skip_download': True, | |
90 | }, | |
00d24327 | 91 | 'skip': 'Video is no longer available', |
321c1e44 S |
92 | }, |
93 | { | |
00d24327 | 94 | # streamType: wasLive |
b4dd9835 | 95 | 'url': 'http://www.vgtv.no/#!/live/113063/direkte-v75-fra-solvalla', |
321c1e44 | 96 | 'info_dict': { |
b4dd9835 | 97 | 'id': '113063', |
7c38af48 | 98 | 'ext': 'mp4', |
00d24327 | 99 | 'title': 'V75 fra Solvalla 30.05.15', |
b4dd9835 | 100 | 'description': 'md5:b3743425765355855f88e096acc93231', |
ec85ded8 | 101 | 'thumbnail': r're:^https?://.*\.jpg', |
00d24327 | 102 | 'duration': 25966, |
b4dd9835 S |
103 | 'timestamp': 1432975582, |
104 | 'upload_date': '20150530', | |
3fbeb95e | 105 | 'view_count': int, |
321c1e44 | 106 | }, |
7c38af48 YCH |
107 | 'params': { |
108 | # m3u8 download | |
109 | 'skip_download': True, | |
110 | }, | |
9f6b5176 | 111 | }, |
112 | { | |
00d24327 | 113 | 'url': 'http://www.aftenposten.no/webtv/#!/video/21039/trailer-sweatshop-i-can-t-take-any-more', |
3b68efdc | 114 | 'md5': 'fd828cd29774a729bf4d4425fe192972', |
00d24327 | 115 | 'info_dict': { |
116 | 'id': '21039', | |
7177fd24 | 117 | 'ext': 'mp4', |
00d24327 | 118 | 'title': 'TRAILER: «SWEATSHOP» - I can´t take any more', |
119 | 'description': 'md5:21891f2b0dd7ec2f78d84a50e54f8238', | |
120 | 'duration': 66, | |
121 | 'timestamp': 1417002452, | |
122 | 'upload_date': '20141126', | |
123 | 'view_count': int, | |
7177fd24 | 124 | }, |
125 | 'params': { | |
126 | # m3u8 download | |
127 | 'skip_download': True, | |
128 | }, | |
321c1e44 | 129 | }, |
dcbf7394 | 130 | { |
131 | 'url': 'https://tv.vg.no/video/241779/politiets-ekstremkjoering', | |
132 | 'only_matching': True, | |
133 | }, | |
34e7dc81 S |
134 | { |
135 | 'url': 'http://www.bt.no/tv/#!/video/100250/norling-dette-er-forskjellen-paa-1-divisjon-og-eliteserien', | |
136 | 'only_matching': True, | |
137 | }, | |
7177fd24 | 138 | { |
139 | 'url': 'http://ap.vgtv.no/webtv#!/video/111084/de-nye-bysyklene-lettere-bedre-gir-stoerre-hjul-og-feste-til-mobil', | |
140 | 'only_matching': True, | |
141 | }, | |
320d597c S |
142 | { |
143 | # geoblocked | |
144 | 'url': 'http://www.vgtv.no/#!/video/127205/inside-the-mind-of-favela-funk', | |
145 | 'only_matching': True, | |
146 | }, | |
1418a043 | 147 | { |
148 | 'url': 'https://tv.aftonbladet.se/video/36015/vulkanutbrott-i-rymden-nu-slapper-nasa-bilderna', | |
149 | 'only_matching': True, | |
150 | }, | |
b0582fc8 S |
151 | { |
152 | 'url': 'http://tv.aftonbladet.se/abtv/articles/36015', | |
153 | 'only_matching': True, | |
154 | }, | |
721a877d S |
155 | { |
156 | 'url': 'https://www.aftonbladet.se/tv/a/36015', | |
157 | 'only_matching': True, | |
158 | }, | |
b0582fc8 S |
159 | { |
160 | 'url': 'abtv:140026', | |
161 | 'only_matching': True, | |
cbf85239 S |
162 | }, |
163 | { | |
164 | 'url': 'http://www.vgtv.no/video/84196/hevnen-er-soet-episode-10-abu', | |
165 | 'only_matching': True, | |
166 | }, | |
321c1e44 | 167 | ] |
78149a96 | 168 | |
321c1e44 | 169 | def _real_extract(self, url): |
5ad28e7f | 170 | mobj = self._match_valid_url(url) |
34e7dc81 S |
171 | video_id = mobj.group('id') |
172 | host = mobj.group('host') | |
41c3b34b | 173 | appname = self._HOST_TO_APPNAME[host] if host else mobj.group('appname') |
174 | vendor = self._APP_NAME_TO_VENDOR[appname] | |
34e7dc81 | 175 | |
321c1e44 | 176 | data = self._download_json( |
34e7dc81 | 177 | 'http://svp.vg.no/svp/api/v1/%s/assets/%s?appName=%s-website' |
41c3b34b | 178 | % (vendor, video_id, appname), |
321c1e44 | 179 | video_id, 'Downloading media JSON') |
78149a96 | 180 | |
4d454c5e S |
181 | if data.get('status') == 'inactive': |
182 | raise ExtractorError( | |
183 | 'Video %s is no longer available' % video_id, expected=True) | |
184 | ||
d50116b8 | 185 | info = { |
186 | 'formats': [], | |
187 | } | |
188 | if len(video_id) == 5: | |
189 | if appname == 'bttv': | |
190 | info = self._extract_video_info('btno', video_id) | |
d50116b8 | 191 | |
321c1e44 | 192 | streams = data['streamUrls'] |
b4dd9835 | 193 | stream_type = data.get('streamType') |
713afa70 | 194 | is_live = stream_type == 'live' |
321c1e44 | 195 | formats = [] |
78149a96 | 196 | |
321c1e44 S |
197 | hls_url = streams.get('hls') |
198 | if hls_url: | |
7e5edcfd | 199 | formats.extend(self._extract_m3u8_formats( |
a5c0c202 | 200 | hls_url, video_id, 'mp4', live=is_live, m3u8_id='hls', fatal=False)) |
78149a96 | 201 | |
321c1e44 | 202 | hds_url = streams.get('hds') |
97f18fac | 203 | if hds_url: |
204 | hdcore_sign = 'hdcore=3.7.0' | |
41c3b34b | 205 | f4m_formats = self._extract_f4m_formats( |
97f18fac | 206 | hds_url + '?%s' % hdcore_sign, video_id, f4m_id='hds', fatal=False) |
9f6b5176 | 207 | if f4m_formats: |
97f18fac | 208 | for entry in f4m_formats: |
209 | # URLs without the extra param induce an 404 error | |
210 | entry.update({'extra_param_to_segment_url': hdcore_sign}) | |
211 | formats.append(entry) | |
78149a96 | 212 | |
9f6b5176 | 213 | mp4_urls = streams.get('pseudostreaming') or [] |
321c1e44 S |
214 | mp4_url = streams.get('mp4') |
215 | if mp4_url: | |
9f6b5176 | 216 | mp4_urls.append(mp4_url) |
217 | for mp4_url in mp4_urls: | |
218 | format_info = { | |
219 | 'url': mp4_url, | |
9f6b5176 | 220 | } |
ec85ded8 | 221 | mobj = re.search(r'(\d+)_(\d+)_(\d+)', mp4_url) |
9f6b5176 | 222 | if mobj: |
3b68efdc | 223 | tbr = int(mobj.group(3)) |
9f6b5176 | 224 | format_info.update({ |
225 | 'width': int(mobj.group(1)), | |
226 | 'height': int(mobj.group(2)), | |
3b68efdc | 227 | 'tbr': tbr, |
228 | 'format_id': 'mp4-%s' % tbr, | |
321c1e44 | 229 | }) |
9f6b5176 | 230 | formats.append(format_info) |
231 | ||
d50116b8 | 232 | info['formats'].extend(formats) |
321c1e44 | 233 | |
320d597c S |
234 | if not info['formats']: |
235 | properties = try_get( | |
236 | data, lambda x: x['streamConfiguration']['properties'], list) | |
237 | if properties and 'geoblocked' in properties: | |
4248dad9 S |
238 | raise self.raise_geo_restricted( |
239 | countries=[host.rpartition('.')[-1].partition('/')[0].upper()]) | |
320d597c | 240 | |
d50116b8 | 241 | info.update({ |
321c1e44 | 242 | 'id': video_id, |
39ca3b5c | 243 | 'title': data['title'], |
321c1e44 S |
244 | 'description': data['description'], |
245 | 'thumbnail': data['images']['main'] + '?t[]=900x506q80', | |
246 | 'timestamp': data['published'], | |
247 | 'duration': float_or_none(data['duration'], 1000), | |
248 | 'view_count': data['displays'], | |
713afa70 | 249 | 'is_live': is_live, |
d50116b8 | 250 | }) |
251 | return info | |
0ceab847 S |
252 | |
253 | ||
254 | class BTArticleIE(InfoExtractor): | |
fe373287 S |
255 | IE_NAME = 'bt:article' |
256 | IE_DESC = 'Bergens Tidende Articles' | |
5886b38d | 257 | _VALID_URL = r'https?://(?:www\.)?bt\.no/(?:[^/]+/)+(?P<id>[^/]+)-\d+\.html' |
0ceab847 S |
258 | _TEST = { |
259 | 'url': 'http://www.bt.no/nyheter/lokalt/Kjemper-for-internatet-1788214.html', | |
3b68efdc | 260 | 'md5': '2acbe8ad129b3469d5ae51b1158878df', |
0ceab847 S |
261 | 'info_dict': { |
262 | 'id': '23199', | |
263 | 'ext': 'mp4', | |
264 | 'title': 'Alrekstad internat', | |
265 | 'description': 'md5:dc81a9056c874fedb62fc48a300dac58', | |
ec85ded8 | 266 | 'thumbnail': r're:^https?://.*\.jpg', |
0ceab847 S |
267 | 'duration': 191, |
268 | 'timestamp': 1289991323, | |
269 | 'upload_date': '20101117', | |
270 | 'view_count': int, | |
271 | }, | |
272 | } | |
273 | ||
274 | def _real_extract(self, url): | |
275 | webpage = self._download_webpage(url, self._match_id(url)) | |
276 | video_id = self._search_regex( | |
3b68efdc | 277 | r'<video[^>]+data-id="(\d+)"', webpage, 'video id') |
41c3b34b | 278 | return self.url_result('bttv:%s' % video_id, 'VGTV') |
fe373287 S |
279 | |
280 | ||
281 | class BTVestlendingenIE(InfoExtractor): | |
282 | IE_NAME = 'bt:vestlendingen' | |
283 | IE_DESC = 'Bergens Tidende - Vestlendingen' | |
5886b38d | 284 | _VALID_URL = r'https?://(?:www\.)?bt\.no/spesial/vestlendingen/#!/(?P<id>\d+)' |
3b68efdc | 285 | _TESTS = [{ |
fe373287 S |
286 | 'url': 'http://www.bt.no/spesial/vestlendingen/#!/86588', |
287 | 'md5': 'd7d17e3337dc80de6d3a540aefbe441b', | |
288 | 'info_dict': { | |
289 | 'id': '86588', | |
290 | 'ext': 'mov', | |
291 | 'title': 'Otto Wollertsen', | |
292 | 'description': 'Vestlendingen Otto Fredrik Wollertsen', | |
293 | 'timestamp': 1430473209, | |
294 | 'upload_date': '20150501', | |
295 | }, | |
3b68efdc | 296 | 'skip': '404 Error', |
297 | }, { | |
298 | 'url': 'http://www.bt.no/spesial/vestlendingen/#!/86255', | |
299 | 'md5': 'a2893f8632e96389f4bdf36aa9463ceb', | |
300 | 'info_dict': { | |
301 | 'id': '86255', | |
302 | 'ext': 'mov', | |
303 | 'title': 'Du må tåle å fryse og være sulten', | |
304 | 'description': 'md5:b8046f4d022d5830ddab04865791d063', | |
305 | 'upload_date': '20150321', | |
306 | 'timestamp': 1426942023, | |
307 | }, | |
308 | }] | |
fe373287 S |
309 | |
310 | def _real_extract(self, url): | |
41c3b34b | 311 | return self.url_result('bttv:%s' % self._match_id(url), 'VGTV') |