]>
Commit | Line | Data |
---|---|---|
b461641f S |
1 | import re |
2 | ||
3 | from .common import InfoExtractor | |
54e8f62e S |
4 | from ..compat import ( |
5 | compat_str, | |
6 | compat_urlparse, | |
7 | ) | |
1cc79574 | 8 | from ..utils import ( |
bec2c14f | 9 | determine_ext, |
50e93e03 | 10 | dict_get, |
1594a493 | 11 | ExtractorError, |
33a1ff71 | 12 | js_to_json, |
37f97295 | 13 | strip_jsonp, |
54e8f62e | 14 | try_get, |
cd7481a3 | 15 | unified_strdate, |
6869d634 | 16 | update_url_query, |
50918c4e | 17 | urlhandle_detect_ext, |
29f7c58a | 18 | url_or_none, |
b461641f S |
19 | ) |
20 | ||
21 | ||
54e8f62e | 22 | class WDRIE(InfoExtractor): |
50e93e03 | 23 | __API_URL_TPL = '//deviceids-medp.wdr.de/ondemand/%s/%s' |
c98d4df2 | 24 | _VALID_URL = r'''(?x)https?:// |
25 | (?:deviceids-medp\.wdr\.de/ondemand/\d+/| | |
26 | kinder\.wdr\.de/(?!mediathek/)[^#?]+-) | |
27 | (?P<id>\d+)\.(?:js|assetjsonp) | |
28 | ''' | |
1915662d | 29 | _GEO_COUNTRIES = ['DE'] |
50e93e03 | 30 | _TESTS = [{ |
54e8f62e S |
31 | 'url': 'http://deviceids-medp.wdr.de/ondemand/155/1557833.js', |
32 | 'info_dict': { | |
0ce39bc5 | 33 | 'id': 'mdb-1557833', |
54e8f62e | 34 | 'ext': 'mp4', |
0ce39bc5 S |
35 | 'title': 'Biathlon-Staffel verpasst Podest bei Olympia-Generalprobe', |
36 | 'upload_date': '20180112', | |
54e8f62e | 37 | }, |
50e93e03 | 38 | }] |
39 | ||
40 | def _asset_url(self, wdr_id): | |
41 | id_len = max(len(wdr_id), 5) | |
42 | return ''.join(('https:', self.__API_URL_TPL % (wdr_id[:id_len - 4], wdr_id, ), '.js')) | |
6d0d4fc2 | 43 | |
54e8f62e S |
44 | def _real_extract(self, url): |
45 | video_id = self._match_id(url) | |
6d0d4fc2 | 46 | |
50e93e03 | 47 | if url.startswith('wdr:'): |
48 | video_id = url[4:] | |
49 | url = self._asset_url(video_id) | |
50 | ||
6d0d4fc2 | 51 | metadata = self._download_json( |
54e8f62e | 52 | url, video_id, transform_source=strip_jsonp) |
6d0d4fc2 | 53 | |
54e8f62e S |
54 | is_live = metadata.get('mediaType') == 'live' |
55 | ||
56 | tracker_data = metadata['trackerData'] | |
29f7c58a | 57 | title = tracker_data['trackerClipTitle'] |
54e8f62e | 58 | media_resource = metadata['mediaResource'] |
6d0d4fc2 YCH |
59 | |
60 | formats = [] | |
c76eb41b | 61 | subtitles = {} |
6d0d4fc2 YCH |
62 | |
63 | # check if the metadata contains a direct URL to a file | |
29f7c58a | 64 | for kind, media in media_resource.items(): |
6de54568 | 65 | if kind == 'captionsHash': |
29f7c58a | 66 | for ext, url in media.items(): |
6de54568 | 67 | subtitles.setdefault('de', []).append({ |
68 | 'url': url, | |
69 | 'ext': ext, | |
70 | }) | |
71 | continue | |
72 | ||
6d0d4fc2 YCH |
73 | if kind not in ('dflt', 'alt'): |
74 | continue | |
29f7c58a | 75 | if not isinstance(media, dict): |
76 | continue | |
6d0d4fc2 | 77 | |
29f7c58a | 78 | for tag_name, medium_url in media.items(): |
6d0d4fc2 YCH |
79 | if tag_name not in ('videoURL', 'audioURL'): |
80 | continue | |
81 | ||
82 | ext = determine_ext(medium_url) | |
83 | if ext == 'm3u8': | |
84 | formats.extend(self._extract_m3u8_formats( | |
54e8f62e | 85 | medium_url, video_id, 'mp4', 'm3u8_native', |
6d0d4fc2 YCH |
86 | m3u8_id='hls')) |
87 | elif ext == 'f4m': | |
88 | manifest_url = update_url_query( | |
89 | medium_url, {'hdcore': '3.2.0', 'plugin': 'aasp-3.2.0.77.18'}) | |
90 | formats.extend(self._extract_f4m_formats( | |
54e8f62e | 91 | manifest_url, video_id, f4m_id='hds', fatal=False)) |
6d0d4fc2 YCH |
92 | elif ext == 'smil': |
93 | formats.extend(self._extract_smil_formats( | |
94 | medium_url, 'stream', fatal=False)) | |
95 | else: | |
96 | a_format = { | |
97 | 'url': medium_url | |
98 | } | |
99 | if ext == 'unknown_video': | |
100 | urlh = self._request_webpage( | |
54e8f62e | 101 | medium_url, video_id, note='Determining extension') |
6d0d4fc2 YCH |
102 | ext = urlhandle_detect_ext(urlh) |
103 | a_format['ext'] = ext | |
104 | formats.append(a_format) | |
105 | ||
29f7c58a | 106 | caption_url = media_resource.get('captionURL') |
107 | if caption_url: | |
108 | subtitles['de'] = [{ | |
109 | 'url': caption_url, | |
110 | 'ext': 'ttml', | |
111 | }] | |
112 | captions_hash = media_resource.get('captionsHash') | |
113 | if isinstance(captions_hash, dict): | |
114 | for ext, format_url in captions_hash.items(): | |
115 | format_url = url_or_none(format_url) | |
116 | if not format_url: | |
117 | continue | |
118 | subtitles.setdefault('de', []).append({ | |
119 | 'url': format_url, | |
120 | 'ext': determine_ext(format_url, None) or ext, | |
121 | }) | |
6d0d4fc2 YCH |
122 | |
123 | return { | |
54e8f62e | 124 | 'id': tracker_data.get('trackerClipId', video_id), |
39ca3b5c | 125 | 'title': title, |
54e8f62e | 126 | 'alt_title': tracker_data.get('trackerClipSubcategory'), |
6d0d4fc2 YCH |
127 | 'formats': formats, |
128 | 'subtitles': subtitles, | |
54e8f62e S |
129 | 'upload_date': unified_strdate(tracker_data.get('trackerClipAirTime')), |
130 | 'is_live': is_live, | |
6d0d4fc2 YCH |
131 | } |
132 | ||
133 | ||
6368e2e6 | 134 | class WDRPageIE(WDRIE): # XXX: Do not subclass from concrete IE |
50e93e03 | 135 | _MAUS_REGEX = r'https?://(?:www\.)wdrmaus.de/(?:[^/]+/)*?(?P<maus_id>[^/?#.]+)(?:/?|/index\.php5|\.php5)$' |
54e8f62e | 136 | _PAGE_REGEX = r'/(?:mediathek/)?(?:[^/]+/)*(?P<display_id>[^/]+)\.html' |
50e93e03 | 137 | _VALID_URL = r'https?://(?:www\d?\.)?(?:(?:kinder\.)?wdr\d?|sportschau)\.de' + _PAGE_REGEX + '|' + _MAUS_REGEX |
c0837a12 | 138 | |
b461641f S |
139 | _TESTS = [ |
140 | { | |
c0837a12 | 141 | 'url': 'http://www1.wdr.de/mediathek/video/sendungen/doku-am-freitag/video-geheimnis-aachener-dom-100.html', |
a4a8201c | 142 | # HDS download, MD5 is unstable |
b461641f | 143 | 'info_dict': { |
c0837a12 | 144 | 'id': 'mdb-1058683', |
b461641f | 145 | 'ext': 'flv', |
c0837a12 BW |
146 | 'display_id': 'doku-am-freitag/video-geheimnis-aachener-dom-100', |
147 | 'title': 'Geheimnis Aachener Dom', | |
148 | 'alt_title': 'Doku am Freitag', | |
149 | 'upload_date': '20160304', | |
150 | 'description': 'md5:87be8ff14d8dfd7a7ee46f0299b52318', | |
151 | 'is_live': False, | |
152 | 'subtitles': {'de': [{ | |
bc7e7adf YCH |
153 | 'url': 'http://ondemand-ww.wdr.de/medp/fsk0/105/1058683/1058683_12220974.xml', |
154 | 'ext': 'ttml', | |
c0837a12 | 155 | }]}, |
b461641f | 156 | }, |
54e8f62e | 157 | 'skip': 'HTTP Error 404: Not Found', |
b461641f S |
158 | }, |
159 | { | |
c0837a12 BW |
160 | 'url': 'http://www1.wdr.de/mediathek/audio/wdr3/wdr3-gespraech-am-samstag/audio-schriftstellerin-juli-zeh-100.html', |
161 | 'md5': 'f4c1f96d01cf285240f53ea4309663d8', | |
b461641f | 162 | 'info_dict': { |
c0837a12 | 163 | 'id': 'mdb-1072000', |
b461641f | 164 | 'ext': 'mp3', |
c0837a12 BW |
165 | 'display_id': 'wdr3-gespraech-am-samstag/audio-schriftstellerin-juli-zeh-100', |
166 | 'title': 'Schriftstellerin Juli Zeh', | |
167 | 'alt_title': 'WDR 3 Gespräch am Samstag', | |
168 | 'upload_date': '20160312', | |
169 | 'description': 'md5:e127d320bc2b1f149be697ce044a3dd7', | |
170 | 'is_live': False, | |
171 | 'subtitles': {} | |
b461641f | 172 | }, |
54e8f62e | 173 | 'skip': 'HTTP Error 404: Not Found', |
b461641f | 174 | }, |
176cf9e0 | 175 | { |
5d039549 | 176 | # FIXME: Asset JSON is directly embedded in webpage |
c0837a12 | 177 | 'url': 'http://www1.wdr.de/mediathek/video/live/index.html', |
dd8982f1 | 178 | 'info_dict': { |
50e93e03 | 179 | 'id': 'mdb-2296252', |
bec2c14f | 180 | 'ext': 'mp4', |
50e93e03 | 181 | 'title': r're:^WDR Fernsehen im Livestream (?:\(nur in Deutschland erreichbar\) )?[0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$', |
c0837a12 | 182 | 'alt_title': 'WDR Fernsehen Live', |
50e93e03 | 183 | 'upload_date': '20201112', |
c0837a12 | 184 | 'is_live': True, |
bec2c14f BW |
185 | }, |
186 | 'params': { | |
187 | 'skip_download': True, # m3u8 download | |
188 | }, | |
b8988b63 AA |
189 | }, |
190 | { | |
c0837a12 | 191 | 'url': 'http://www1.wdr.de/mediathek/video/sendungen/aktuelle-stunde/aktuelle-stunde-120.html', |
50e93e03 | 192 | 'playlist_mincount': 6, |
b8988b63 | 193 | 'info_dict': { |
54e8f62e | 194 | 'id': 'aktuelle-stunde-120', |
b8988b63 | 195 | }, |
14f7a2b8 BW |
196 | }, |
197 | { | |
198 | 'url': 'http://www.wdrmaus.de/aktuelle-sendung/index.php5', | |
199 | 'info_dict': { | |
50e93e03 | 200 | 'id': 'mdb-2627637', |
f802c486 | 201 | 'ext': 'mp4', |
14f7a2b8 | 202 | 'upload_date': 're:^[0-9]{8}$', |
50e93e03 | 203 | 'title': 're:^Die Sendung (?:mit der Maus )?vom [0-9.]{10}$', |
14f7a2b8 BW |
204 | }, |
205 | 'skip': 'The id changes from week to week because of the new episode' | |
206 | }, | |
949fc42e | 207 | { |
f802c486 | 208 | 'url': 'http://www.wdrmaus.de/filme/sachgeschichten/achterbahn.php5', |
6d0d4fc2 | 209 | 'md5': '803138901f6368ee497b4d195bb164f2', |
949fc42e BW |
210 | 'info_dict': { |
211 | 'id': 'mdb-186083', | |
6d0d4fc2 | 212 | 'ext': 'mp4', |
949fc42e BW |
213 | 'upload_date': '20130919', |
214 | 'title': 'Sachgeschichte - Achterbahn ', | |
949fc42e | 215 | }, |
50e93e03 | 216 | 'skip': 'HTTP Error 404: Not Found', |
949fc42e | 217 | }, |
50918c4e YCH |
218 | { |
219 | 'url': 'http://www1.wdr.de/radio/player/radioplayer116~_layout-popupVersion.html', | |
6869d634 | 220 | # Live stream, MD5 unstable |
50918c4e YCH |
221 | 'info_dict': { |
222 | 'id': 'mdb-869971', | |
54e8f62e S |
223 | 'ext': 'mp4', |
224 | 'title': r're:^COSMO Livestream [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$', | |
5d039549 SG |
225 | 'alt_title': 'COSMO Livestream', |
226 | 'live_status': 'is_live', | |
50918c4e YCH |
227 | 'upload_date': '20160101', |
228 | }, | |
54e8f62e S |
229 | 'params': { |
230 | 'skip_download': True, # m3u8 download | |
231 | } | |
232 | }, | |
233 | { | |
234 | 'url': 'http://www.sportschau.de/handballem2018/handball-nationalmannschaft-em-stolperstein-vorrunde-100.html', | |
235 | 'info_dict': { | |
236 | 'id': 'mdb-1556012', | |
237 | 'ext': 'mp4', | |
238 | 'title': 'DHB-Vizepräsident Bob Hanning - "Die Weltspitze ist extrem breit"', | |
239 | 'upload_date': '20180111', | |
240 | }, | |
241 | 'params': { | |
242 | 'skip_download': True, | |
243 | }, | |
50e93e03 | 244 | 'skip': 'HTTP Error 404: Not Found', |
54e8f62e S |
245 | }, |
246 | { | |
247 | 'url': 'http://www.sportschau.de/handballem2018/audio-vorschau---die-handball-em-startet-mit-grossem-favoritenfeld-100.html', | |
248 | 'only_matching': True, | |
29f7c58a | 249 | }, |
250 | { | |
251 | 'url': 'https://kinder.wdr.de/tv/die-sendung-mit-dem-elefanten/av/video-folge---astronaut-100.html', | |
252 | 'only_matching': True, | |
253 | }, | |
5d039549 SG |
254 | { |
255 | 'url': 'https://www1.wdr.de/mediathek/video/sendungen/rockpalast/video-baroness---freak-valley-festival--100.html', | |
256 | 'info_dict': { | |
257 | 'id': 'mdb-2741028', | |
258 | 'ext': 'mp4', | |
259 | 'title': 'Baroness - Freak Valley Festival 2022', | |
260 | 'alt_title': 'Rockpalast', | |
261 | 'upload_date': '20220725', | |
262 | }, | |
263 | } | |
b461641f S |
264 | ] |
265 | ||
266 | def _real_extract(self, url): | |
5ad28e7f | 267 | mobj = self._match_valid_url(url) |
50e93e03 | 268 | display_id = dict_get(mobj.groupdict(), ('display_id', 'maus_id'), 'wdrmaus') |
c0837a12 | 269 | webpage = self._download_webpage(url, display_id) |
b461641f | 270 | |
54e8f62e | 271 | entries = [] |
b8988b63 | 272 | |
54e8f62e | 273 | # Article with several videos |
b461641f | 274 | |
5d039549 | 275 | # for wdr.de the data-extension-ard is in a tag with the class "mediaLink" |
54e8f62e S |
276 | # for wdr.de radio players, in a tag with the class "wdrrPlayerPlayBtn" |
277 | # for wdrmaus, in a tag with the class "videoButton" (previously a link | |
278 | # to the page in a multiline "videoLink"-tag) | |
279 | for mobj in re.finditer( | |
280 | r'''(?sx)class= | |
281 | (?: | |
282 | (["\'])(?:mediaLink|wdrrPlayerPlayBtn|videoButton)\b.*?\1[^>]+| | |
283 | (["\'])videoLink\b.*?\2[\s]*>\n[^\n]* | |
5d039549 | 284 | )data-extension(?:-ard)?=(["\'])(?P<data>(?:(?!\3).)+)\3 |
54e8f62e S |
285 | ''', webpage): |
286 | media_link_obj = self._parse_json( | |
287 | mobj.group('data'), display_id, transform_source=js_to_json, | |
288 | fatal=False) | |
289 | if not media_link_obj: | |
290 | continue | |
291 | jsonp_url = try_get( | |
292 | media_link_obj, lambda x: x['mediaObj']['url'], compat_str) | |
293 | if jsonp_url: | |
50e93e03 | 294 | # metadata, or player JS with ['ref'] giving WDR id, or just media, perhaps |
295 | clip_id = media_link_obj['mediaObj'].get('ref') | |
296 | if jsonp_url.endswith('.assetjsonp'): | |
297 | asset = self._download_json( | |
298 | jsonp_url, display_id, fatal=False, transform_source=strip_jsonp) | |
299 | clip_id = try_get(asset, lambda x: x['trackerData']['trackerClipId'], compat_str) | |
300 | if clip_id: | |
301 | jsonp_url = self._asset_url(clip_id[4:]) | |
54e8f62e | 302 | entries.append(self.url_result(jsonp_url, ie=WDRIE.ie_key())) |
6d0d4fc2 | 303 | |
54e8f62e S |
304 | # Playlist (e.g. https://www1.wdr.de/mediathek/video/sendungen/aktuelle-stunde/aktuelle-stunde-120.html) |
305 | if not entries: | |
306 | entries = [ | |
307 | self.url_result( | |
308 | compat_urlparse.urljoin(url, mobj.group('href')), | |
309 | ie=WDRPageIE.ie_key()) | |
310 | for mobj in re.finditer( | |
5d039549 | 311 | r'<a[^>]+\bhref=(["\'])(?P<href>(?:(?!\1).)+)\1[^>]+\bdata-extension(?:-ard)?=', |
54e8f62e S |
312 | webpage) if re.match(self._PAGE_REGEX, mobj.group('href')) |
313 | ] | |
cd7481a3 | 314 | |
54e8f62e | 315 | return self.playlist_result(entries, playlist_id=display_id) |
cd7481a3 | 316 | |
2d8bb80c | 317 | |
54e8f62e S |
318 | class WDRElefantIE(InfoExtractor): |
319 | _VALID_URL = r'https?://(?:www\.)wdrmaus\.de/elefantenseite/#(?P<id>.+)' | |
320 | _TEST = { | |
50e93e03 | 321 | 'url': 'http://www.wdrmaus.de/elefantenseite/#elefantenkino_wippe', |
322 | # adaptive stream: unstable file MD5 | |
54e8f62e | 323 | 'info_dict': { |
50e93e03 | 324 | 'title': 'Wippe', |
325 | 'id': 'mdb-1198320', | |
54e8f62e S |
326 | 'ext': 'mp4', |
327 | 'age_limit': None, | |
50e93e03 | 328 | 'upload_date': '20071003' |
54e8f62e S |
329 | }, |
330 | } | |
2d8bb80c SL |
331 | |
332 | def _real_extract(self, url): | |
54e8f62e | 333 | display_id = self._match_id(url) |
2d8bb80c SL |
334 | |
335 | # Table of Contents seems to always be at this address, so fetch it directly. | |
336 | # The website fetches configurationJS.php5, which links to tableOfContentsJS.php5. | |
337 | table_of_contents = self._download_json( | |
54e8f62e S |
338 | 'https://www.wdrmaus.de/elefantenseite/data/tableOfContentsJS.php5', |
339 | display_id) | |
2d8bb80c SL |
340 | if display_id not in table_of_contents: |
341 | raise ExtractorError( | |
342 | 'No entry in site\'s table of contents for this URL. ' | |
343 | 'Is the fragment part of the URL (after the #) correct?', | |
344 | expected=True) | |
345 | xml_metadata_path = table_of_contents[display_id]['xmlPath'] | |
346 | xml_metadata = self._download_xml( | |
54e8f62e S |
347 | 'https://www.wdrmaus.de/elefantenseite/' + xml_metadata_path, |
348 | display_id) | |
2d8bb80c SL |
349 | zmdb_url_element = xml_metadata.find('./movie/zmdb_url') |
350 | if zmdb_url_element is None: | |
351 | raise ExtractorError( | |
54e8f62e S |
352 | '%s is not a video' % display_id, expected=True) |
353 | return self.url_result(zmdb_url_element.text, ie=WDRIE.ie_key()) | |
2d8bb80c SL |
354 | |
355 | ||
e4cbb5f3 PH |
356 | class WDRMobileIE(InfoExtractor): |
357 | _VALID_URL = r'''(?x) | |
358 | https?://mobile-ondemand\.wdr\.de/ | |
359 | .*?/fsk(?P<age_limit>[0-9]+) | |
360 | /[0-9]+/[0-9]+/ | |
361 | (?P<id>[0-9]+)_(?P<title>[0-9]+)''' | |
362 | IE_NAME = 'wdr:mobile' | |
50e93e03 | 363 | _WORKING = False # no such domain |
e4cbb5f3 PH |
364 | _TEST = { |
365 | 'url': 'http://mobile-ondemand.wdr.de/CMS2010/mdb/ondemand/weltweit/fsk0/42/421735/421735_4283021.mp4', | |
366 | 'info_dict': { | |
367 | 'title': '4283021', | |
368 | 'id': '421735', | |
7807ee66 | 369 | 'ext': 'mp4', |
e4cbb5f3 PH |
370 | 'age_limit': 0, |
371 | }, | |
7807ee66 | 372 | 'skip': 'Problems with loading data.' |
e4cbb5f3 PH |
373 | } |
374 | ||
375 | def _real_extract(self, url): | |
5ad28e7f | 376 | mobj = self._match_valid_url(url) |
e4cbb5f3 PH |
377 | return { |
378 | 'id': mobj.group('id'), | |
379 | 'title': mobj.group('title'), | |
380 | 'age_limit': int(mobj.group('age_limit')), | |
381 | 'url': url, | |
e1554a40 JMF |
382 | 'http_headers': { |
383 | 'User-Agent': 'mobile', | |
384 | }, | |
e4cbb5f3 | 385 | } |