]>
Commit | Line | Data |
---|---|---|
331ae266 JMF |
1 | from __future__ import unicode_literals |
2 | ||
525daedd S |
3 | import re |
4 | ||
171ca612 | 5 | from .common import InfoExtractor |
83e8fce6 S |
6 | from ..compat import ( |
7 | compat_HTTPError, | |
8 | compat_str, | |
9 | ) | |
331ae266 | 10 | from ..utils import ( |
83e8fce6 S |
11 | determine_ext, |
12 | ExtractorError, | |
5c4a81d9 | 13 | fix_xml_ampersands, |
ee706f10 | 14 | int_or_none, |
5742c18b | 15 | orderedSet, |
944a3de2 | 16 | parse_duration, |
c9ea760e | 17 | qualities, |
609a61e3 | 18 | strip_jsonp, |
5c4a81d9 | 19 | unified_strdate, |
331ae266 JMF |
20 | ) |
21 | ||
22 | ||
b9b42f2e | 23 | class NPOBaseIE(InfoExtractor): |
04e0bac2 | 24 | def _get_token(self, video_id): |
aa9cc2ec S |
25 | return self._download_json( |
26 | 'http://ida.omroep.nl/app.php/auth', video_id, | |
27 | note='Downloading token')['token'] | |
04e0bac2 S |
28 | |
29 | ||
30 | class NPOIE(NPOBaseIE): | |
e118031e | 31 | IE_NAME = 'npo' |
359aa2fd | 32 | IE_DESC = 'npo.nl, ntr.nl, omroepwnl.nl, zapp.nl and npo3.nl' |
525daedd | 33 | _VALID_URL = r'''(?x) |
03f32a7e S |
34 | (?: |
35 | npo:| | |
36 | https?:// | |
37 | (?:www\.)? | |
38 | (?: | |
9d581efe S |
39 | npo\.nl/(?:[^/]+/)*| |
40 | (?:ntr|npostart)\.nl/(?:[^/]+/){2,}| | |
f264c623 | 41 | omroepwnl\.nl/video/fragment/[^/]+__| |
c4e74964 | 42 | (?:zapp|npo3)\.nl/(?:[^/]+/){2,} |
03f32a7e | 43 | ) |
525daedd S |
44 | ) |
45 | (?P<id>[^/?#]+) | |
46 | ''' | |
331ae266 | 47 | |
83e8fce6 S |
48 | _TESTS = [{ |
49 | 'url': 'http://www.npo.nl/nieuwsuur/22-06-2014/VPWON_1220719', | |
50 | 'md5': '4b3f9c429157ec4775f2c9cb7b911016', | |
51 | 'info_dict': { | |
52 | 'id': 'VPWON_1220719', | |
53 | 'ext': 'm4v', | |
54 | 'title': 'Nieuwsuur', | |
55 | 'description': 'Dagelijks tussen tien en elf: nieuws, sport en achtergronden.', | |
56 | 'upload_date': '20140622', | |
331ae266 | 57 | }, |
83e8fce6 S |
58 | }, { |
59 | 'url': 'http://www.npo.nl/de-mega-mike-mega-thomas-show/27-02-2009/VARA_101191800', | |
60 | 'md5': 'da50a5787dbfc1603c4ad80f31c5120b', | |
61 | 'info_dict': { | |
62 | 'id': 'VARA_101191800', | |
63 | 'ext': 'm4v', | |
64 | 'title': 'De Mega Mike & Mega Thomas show: The best of.', | |
65 | 'description': 'md5:3b74c97fc9d6901d5a665aac0e5400f4', | |
66 | 'upload_date': '20090227', | |
67 | 'duration': 2400, | |
944a3de2 | 68 | }, |
83e8fce6 S |
69 | }, { |
70 | 'url': 'http://www.npo.nl/tegenlicht/25-02-2013/VPWON_1169289', | |
71 | 'md5': 'f8065e4e5a7824068ed3c7e783178f2c', | |
72 | 'info_dict': { | |
73 | 'id': 'VPWON_1169289', | |
74 | 'ext': 'm4v', | |
75 | 'title': 'Tegenlicht: Zwart geld. De toekomst komt uit Afrika', | |
76 | 'description': 'md5:52cf4eefbc96fffcbdc06d024147abea', | |
77 | 'upload_date': '20130225', | |
78 | 'duration': 3000, | |
1ff30d7b | 79 | }, |
83e8fce6 S |
80 | }, { |
81 | 'url': 'http://www.npo.nl/de-nieuwe-mens-deel-1/21-07-2010/WO_VPRO_043706', | |
82 | 'info_dict': { | |
83 | 'id': 'WO_VPRO_043706', | |
84 | 'ext': 'm4v', | |
85 | 'title': 'De nieuwe mens - Deel 1', | |
86 | 'description': 'md5:518ae51ba1293ffb80d8d8ce90b74e4b', | |
87 | 'duration': 4680, | |
1ff30d7b | 88 | }, |
83e8fce6 S |
89 | 'params': { |
90 | 'skip_download': True, | |
91 | } | |
92 | }, { | |
c85f3683 | 93 | # non asf in streams |
83e8fce6 S |
94 | 'url': 'http://www.npo.nl/hoe-gaat-europa-verder-na-parijs/10-01-2015/WO_NOS_762771', |
95 | 'info_dict': { | |
96 | 'id': 'WO_NOS_762771', | |
97 | 'ext': 'mp4', | |
98 | 'title': 'Hoe gaat Europa verder na Parijs?', | |
f264c623 | 99 | }, |
83e8fce6 S |
100 | 'params': { |
101 | 'skip_download': True, | |
102 | } | |
103 | }, { | |
104 | 'url': 'http://www.ntr.nl/Aap-Poot-Pies/27/detail/Aap-poot-pies/VPWON_1233944#content', | |
105 | 'info_dict': { | |
106 | 'id': 'VPWON_1233944', | |
107 | 'ext': 'm4v', | |
108 | 'title': 'Aap, poot, pies', | |
109 | 'description': 'md5:c9c8005d1869ae65b858e82c01a91fde', | |
110 | 'upload_date': '20150508', | |
111 | 'duration': 599, | |
f264c623 | 112 | }, |
83e8fce6 S |
113 | 'params': { |
114 | 'skip_download': True, | |
115 | } | |
116 | }, { | |
117 | 'url': 'http://www.omroepwnl.nl/video/fragment/vandaag-de-dag-verkiezingen__POMS_WNL_853698', | |
118 | 'info_dict': { | |
119 | 'id': 'POW_00996502', | |
120 | 'ext': 'm4v', | |
121 | 'title': '''"Dit is wel een 'landslide'..."''', | |
122 | 'description': 'md5:f8d66d537dfb641380226e31ca57b8e8', | |
123 | 'upload_date': '20150508', | |
124 | 'duration': 462, | |
18abb743 | 125 | }, |
83e8fce6 S |
126 | 'params': { |
127 | 'skip_download': True, | |
e118031e | 128 | } |
4b8a984c S |
129 | }, { |
130 | # audio | |
131 | 'url': 'http://www.npo.nl/jouw-stad-rotterdam/29-01-2017/RBX_FUNX_6683215/RBX_FUNX_7601437', | |
132 | 'info_dict': { | |
133 | 'id': 'RBX_FUNX_6683215', | |
134 | 'ext': 'mp3', | |
135 | 'title': 'Jouw Stad Rotterdam', | |
136 | 'description': 'md5:db251505244f097717ec59fabc372d9f', | |
137 | }, | |
138 | 'params': { | |
139 | 'skip_download': True, | |
140 | } | |
83e8fce6 S |
141 | }, { |
142 | 'url': 'http://www.zapp.nl/de-bzt-show/gemist/KN_1687547', | |
143 | 'only_matching': True, | |
144 | }, { | |
145 | 'url': 'http://www.zapp.nl/de-bzt-show/filmpjes/POMS_KN_7315118', | |
146 | 'only_matching': True, | |
147 | }, { | |
148 | 'url': 'http://www.zapp.nl/beste-vrienden-quiz/extra-video-s/WO_NTR_1067990', | |
149 | 'only_matching': True, | |
359aa2fd | 150 | }, { |
151 | 'url': 'https://www.npo3.nl/3onderzoekt/16-09-2015/VPWON_1239870', | |
152 | 'only_matching': True, | |
83e8fce6 S |
153 | }, { |
154 | # live stream | |
155 | 'url': 'npo:LI_NL1_4188102', | |
156 | 'only_matching': True, | |
7dd5415c S |
157 | }, { |
158 | 'url': 'http://www.npo.nl/radio-gaga/13-06-2017/BNN_101383373', | |
159 | 'only_matching': True, | |
c4e74964 S |
160 | }, { |
161 | 'url': 'https://www.zapp.nl/1803-skelterlab/instructie-video-s/740-instructievideo-s/POMS_AT_11736927', | |
162 | 'only_matching': True, | |
9d581efe S |
163 | }, { |
164 | 'url': 'https://www.npostart.nl/broodje-gezond-ei/28-05-2018/KN_1698996', | |
165 | 'only_matching': True, | |
166 | }, { | |
167 | 'url': 'https://npo.nl/KN_1698996', | |
168 | 'only_matching': True, | |
83e8fce6 | 169 | }] |
331ae266 | 170 | |
9d581efe S |
171 | @classmethod |
172 | def suitable(cls, url): | |
173 | return (False if any(ie.suitable(url) | |
174 | for ie in (NPOLiveIE, NPORadioIE, NPORadioFragmentIE)) | |
175 | else super(NPOIE, cls).suitable(url)) | |
176 | ||
331ae266 | 177 | def _real_extract(self, url): |
04e0bac2 | 178 | video_id = self._match_id(url) |
d0df9292 | 179 | return self._get_info(video_id) |
331ae266 | 180 | |
d0df9292 | 181 | def _get_info(self, video_id): |
331ae266 | 182 | metadata = self._download_json( |
8d06a624 | 183 | 'http://e.omroep.nl/metadata/%s' % video_id, |
331ae266 JMF |
184 | video_id, |
185 | # We have to remove the javascript callback | |
609a61e3 | 186 | transform_source=strip_jsonp, |
331ae266 | 187 | ) |
04e0bac2 | 188 | |
b8adcec4 S |
189 | error = metadata.get('error') |
190 | if error: | |
191 | raise ExtractorError(error, expected=True) | |
192 | ||
525daedd S |
193 | # For some videos actual video id (prid) is different (e.g. for |
194 | # http://www.omroepwnl.nl/video/fragment/vandaag-de-dag-verkiezingen__POMS_WNL_853698 | |
195 | # video id is POMS_WNL_853698 but prid is POW_00996502) | |
196 | video_id = metadata.get('prid') or video_id | |
197 | ||
23fc384f S |
198 | # titel is too generic in some cases so utilize aflevering_titel as well |
199 | # when available (e.g. http://tegenlicht.vpro.nl/afleveringen/2014-2015/access-to-africa.html) | |
200 | title = metadata['titel'] | |
201 | sub_title = metadata.get('aflevering_titel') | |
202 | if sub_title and sub_title != title: | |
203 | title += ': %s' % sub_title | |
204 | ||
04e0bac2 | 205 | token = self._get_token(video_id) |
331ae266 | 206 | |
c9ea760e | 207 | formats = [] |
83e8fce6 S |
208 | urls = set() |
209 | ||
d1b61870 S |
210 | def is_legal_url(format_url): |
211 | return format_url and format_url not in urls and re.match( | |
212 | r'^(?:https?:)?//', format_url) | |
213 | ||
ee706f10 S |
214 | QUALITY_LABELS = ('Laag', 'Normaal', 'Hoog') |
215 | QUALITY_FORMATS = ('adaptive', 'wmv_sb', 'h264_sb', 'wmv_bb', 'h264_bb', 'wvc1_std', 'h264_std') | |
216 | ||
217 | quality_from_label = qualities(QUALITY_LABELS) | |
218 | quality_from_format_id = qualities(QUALITY_FORMATS) | |
83e8fce6 S |
219 | items = self._download_json( |
220 | 'http://ida.omroep.nl/app.php/%s' % video_id, video_id, | |
221 | 'Downloading formats JSON', query={ | |
222 | 'adaptive': 'yes', | |
223 | 'token': token, | |
224 | })['items'][0] | |
225 | for num, item in enumerate(items): | |
226 | item_url = item.get('url') | |
d1b61870 | 227 | if not is_legal_url(item_url): |
83e8fce6 S |
228 | continue |
229 | urls.add(item_url) | |
230 | format_id = self._search_regex( | |
231 | r'video/ida/([^/]+)', item_url, 'format id', | |
232 | default=None) | |
233 | ||
ee706f10 S |
234 | item_label = item.get('label') |
235 | ||
83e8fce6 | 236 | def add_format_url(format_url): |
ee706f10 S |
237 | width = int_or_none(self._search_regex( |
238 | r'(\d+)[xX]\d+', format_url, 'width', default=None)) | |
239 | height = int_or_none(self._search_regex( | |
240 | r'\d+[xX](\d+)', format_url, 'height', default=None)) | |
241 | if item_label in QUALITY_LABELS: | |
242 | quality = quality_from_label(item_label) | |
243 | f_id = item_label | |
244 | elif item_label in QUALITY_FORMATS: | |
245 | quality = quality_from_format_id(format_id) | |
246 | f_id = format_id | |
247 | else: | |
5616caf8 | 248 | quality, f_id = [None] * 2 |
83e8fce6 S |
249 | formats.append({ |
250 | 'url': format_url, | |
ee706f10 S |
251 | 'format_id': f_id, |
252 | 'width': width, | |
253 | 'height': height, | |
254 | 'quality': quality, | |
83e8fce6 | 255 | }) |
1ff30d7b | 256 | |
83e8fce6 | 257 | # Example: http://www.npo.nl/de-nieuwe-mens-deel-1/21-07-2010/WO_VPRO_043706 |
4b8a984c | 258 | if item.get('contentType') in ('url', 'audio'): |
83e8fce6 S |
259 | add_format_url(item_url) |
260 | continue | |
261 | ||
262 | try: | |
263 | stream_info = self._download_json( | |
264 | item_url + '&type=json', video_id, | |
265 | 'Downloading %s stream JSON' | |
ee706f10 | 266 | % item_label or item.get('format') or format_id or num) |
83e8fce6 S |
267 | except ExtractorError as ee: |
268 | if isinstance(ee.cause, compat_HTTPError) and ee.cause.code == 404: | |
269 | error = (self._parse_json( | |
270 | ee.cause.read().decode(), video_id, | |
271 | fatal=False) or {}).get('errorstring') | |
272 | if error: | |
273 | raise ExtractorError(error, expected=True) | |
274 | raise | |
275 | # Stream URL instead of JSON, example: npo:LI_NL1_4188102 | |
276 | if isinstance(stream_info, compat_str): | |
277 | if not stream_info.startswith('http'): | |
aa9cc2ec | 278 | continue |
83e8fce6 S |
279 | video_url = stream_info |
280 | # JSON | |
281 | else: | |
aa9cc2ec | 282 | video_url = stream_info.get('url') |
83e8fce6 S |
283 | if not video_url or video_url in urls: |
284 | continue | |
285 | urls.add(item_url) | |
286 | if determine_ext(video_url) == 'm3u8': | |
287 | formats.extend(self._extract_m3u8_formats( | |
288 | video_url, video_id, ext='mp4', | |
289 | entry_protocol='m3u8_native', m3u8_id='hls', fatal=False)) | |
290 | else: | |
291 | add_format_url(video_url) | |
292 | ||
293 | is_live = metadata.get('medium') == 'live' | |
294 | ||
295 | if not is_live: | |
296 | for num, stream in enumerate(metadata.get('streams', [])): | |
297 | stream_url = stream.get('url') | |
d1b61870 | 298 | if not is_legal_url(stream_url): |
1ff30d7b | 299 | continue |
83e8fce6 S |
300 | urls.add(stream_url) |
301 | # smooth streaming is not supported | |
302 | stream_type = stream.get('type', '').lower() | |
303 | if stream_type in ['ss', 'ms']: | |
304 | continue | |
305 | if stream_type == 'hds': | |
306 | f4m_formats = self._extract_f4m_formats( | |
307 | stream_url, video_id, fatal=False) | |
308 | # f4m downloader downloads only piece of live stream | |
309 | for f4m_format in f4m_formats: | |
310 | f4m_format['preference'] = -1 | |
311 | formats.extend(f4m_formats) | |
312 | elif stream_type == 'hls': | |
aa9cc2ec | 313 | formats.extend(self._extract_m3u8_formats( |
83e8fce6 S |
314 | stream_url, video_id, ext='mp4', fatal=False)) |
315 | # Example: http://www.npo.nl/de-nieuwe-mens-deel-1/21-07-2010/WO_VPRO_043706 | |
316 | elif '.asf' in stream_url: | |
317 | asx = self._download_xml( | |
318 | stream_url, video_id, | |
319 | 'Downloading stream %d ASX playlist' % num, | |
320 | transform_source=fix_xml_ampersands, fatal=False) | |
321 | if not asx: | |
322 | continue | |
323 | ref = asx.find('./ENTRY/Ref') | |
324 | if ref is None: | |
325 | continue | |
326 | video_url = ref.get('href') | |
327 | if not video_url or video_url in urls: | |
328 | continue | |
329 | urls.add(video_url) | |
1ff30d7b S |
330 | formats.append({ |
331 | 'url': video_url, | |
83e8fce6 S |
332 | 'ext': stream.get('formaat', 'asf'), |
333 | 'quality': stream.get('kwaliteit'), | |
334 | 'preference': -10, | |
1ff30d7b | 335 | }) |
83e8fce6 | 336 | else: |
a0977064 S |
337 | formats.append({ |
338 | 'url': stream_url, | |
339 | 'quality': stream.get('kwaliteit'), | |
340 | }) | |
1ff30d7b | 341 | |
9e91449c | 342 | self._sort_formats(formats) |
25e5ebf3 | 343 | |
9e91449c S |
344 | subtitles = {} |
345 | if metadata.get('tt888') == 'ja': | |
b9b42f2e JMF |
346 | subtitles['nl'] = [{ |
347 | 'ext': 'vtt', | |
56fc078d | 348 | 'url': 'http://tt888.omroep.nl/tt888/%s' % video_id, |
b9b42f2e | 349 | }] |
331ae266 JMF |
350 | |
351 | return { | |
352 | 'id': video_id, | |
83e8fce6 | 353 | 'title': self._live_title(title) if is_live else title, |
536b0700 | 354 | 'description': metadata.get('info'), |
944a3de2 S |
355 | 'thumbnail': metadata.get('images', [{'url': None}])[-1]['url'], |
356 | 'upload_date': unified_strdate(metadata.get('gidsdatum')), | |
357 | 'duration': parse_duration(metadata.get('tijdsduur')), | |
c9ea760e | 358 | 'formats': formats, |
25e5ebf3 | 359 | 'subtitles': subtitles, |
83e8fce6 | 360 | 'is_live': is_live, |
331ae266 | 361 | } |
d0df9292 JMF |
362 | |
363 | ||
04e0bac2 S |
364 | class NPOLiveIE(NPOBaseIE): |
365 | IE_NAME = 'npo.nl:live' | |
04e431cf | 366 | _VALID_URL = r'https?://(?:www\.)?npo\.nl/live(?:/(?P<id>[^/?#&]+))?' |
04e0bac2 | 367 | |
1591ba25 | 368 | _TESTS = [{ |
04e0bac2 S |
369 | 'url': 'http://www.npo.nl/live/npo-1', |
370 | 'info_dict': { | |
83e8fce6 | 371 | 'id': 'LI_NL1_4188102', |
04e0bac2 S |
372 | 'display_id': 'npo-1', |
373 | 'ext': 'mp4', | |
83e8fce6 | 374 | 'title': 're:^NPO 1 [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$', |
04e0bac2 S |
375 | 'is_live': True, |
376 | }, | |
377 | 'params': { | |
378 | 'skip_download': True, | |
379 | } | |
1591ba25 AG |
380 | }, { |
381 | 'url': 'http://www.npo.nl/live', | |
04e431cf | 382 | 'only_matching': True, |
1591ba25 | 383 | }] |
04e0bac2 S |
384 | |
385 | def _real_extract(self, url): | |
1591ba25 | 386 | display_id = self._match_id(url) or 'npo-1' |
04e0bac2 S |
387 | |
388 | webpage = self._download_webpage(url, display_id) | |
389 | ||
390 | live_id = self._search_regex( | |
7a577309 | 391 | [r'media-id="([^"]+)"', r'data-prid="([^"]+)"'], webpage, 'live id') |
04e0bac2 | 392 | |
04e0bac2 | 393 | return { |
83e8fce6 S |
394 | '_type': 'url_transparent', |
395 | 'url': 'npo:%s' % live_id, | |
396 | 'ie_key': NPOIE.ie_key(), | |
04e0bac2 S |
397 | 'id': live_id, |
398 | 'display_id': display_id, | |
04e0bac2 S |
399 | } |
400 | ||
401 | ||
171ca612 S |
402 | class NPORadioIE(InfoExtractor): |
403 | IE_NAME = 'npo.nl:radio' | |
9d581efe | 404 | _VALID_URL = r'https?://(?:www\.)?npo\.nl/radio/(?P<id>[^/]+)' |
171ca612 S |
405 | |
406 | _TEST = { | |
407 | 'url': 'http://www.npo.nl/radio/radio-1', | |
408 | 'info_dict': { | |
409 | 'id': 'radio-1', | |
410 | 'ext': 'mp3', | |
411 | 'title': 're:^NPO Radio 1 [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$', | |
412 | 'is_live': True, | |
413 | }, | |
414 | 'params': { | |
415 | 'skip_download': True, | |
416 | } | |
417 | } | |
418 | ||
9d581efe S |
419 | @classmethod |
420 | def suitable(cls, url): | |
421 | return False if NPORadioFragmentIE.suitable(url) else super(NPORadioIE, cls).suitable(url) | |
422 | ||
171ca612 S |
423 | @staticmethod |
424 | def _html_get_attribute_regex(attribute): | |
425 | return r'{0}\s*=\s*\'([^\']+)\''.format(attribute) | |
426 | ||
427 | def _real_extract(self, url): | |
428 | video_id = self._match_id(url) | |
429 | ||
430 | webpage = self._download_webpage(url, video_id) | |
431 | ||
432 | title = self._html_search_regex( | |
433 | self._html_get_attribute_regex('data-channel'), webpage, 'title') | |
434 | ||
435 | stream = self._parse_json( | |
436 | self._html_search_regex(self._html_get_attribute_regex('data-streams'), webpage, 'data-streams'), | |
437 | video_id) | |
438 | ||
439 | codec = stream.get('codec') | |
440 | ||
441 | return { | |
442 | 'id': video_id, | |
443 | 'url': stream['url'], | |
444 | 'title': self._live_title(title), | |
445 | 'acodec': codec, | |
446 | 'ext': codec, | |
447 | 'is_live': True, | |
448 | } | |
449 | ||
450 | ||
451 | class NPORadioFragmentIE(InfoExtractor): | |
452 | IE_NAME = 'npo.nl:radio:fragment' | |
453 | _VALID_URL = r'https?://(?:www\.)?npo\.nl/radio/[^/]+/fragment/(?P<id>\d+)' | |
454 | ||
455 | _TEST = { | |
456 | 'url': 'http://www.npo.nl/radio/radio-5/fragment/174356', | |
457 | 'md5': 'dd8cc470dad764d0fdc70a9a1e2d18c2', | |
458 | 'info_dict': { | |
459 | 'id': '174356', | |
460 | 'ext': 'mp3', | |
461 | 'title': 'Jubileumconcert Willeke Alberti', | |
462 | }, | |
463 | } | |
464 | ||
465 | def _real_extract(self, url): | |
466 | audio_id = self._match_id(url) | |
467 | ||
468 | webpage = self._download_webpage(url, audio_id) | |
469 | ||
470 | title = self._html_search_regex( | |
471 | r'href="/radio/[^/]+/fragment/%s" title="([^"]+)"' % audio_id, | |
472 | webpage, 'title') | |
473 | ||
474 | audio_url = self._search_regex( | |
475 | r"data-streams='([^']+)'", webpage, 'audio url') | |
476 | ||
477 | return { | |
478 | 'id': audio_id, | |
479 | 'url': audio_url, | |
480 | 'title': title, | |
481 | } | |
482 | ||
483 | ||
0dc5a86a S |
484 | class NPODataMidEmbedIE(InfoExtractor): |
485 | def _real_extract(self, url): | |
486 | display_id = self._match_id(url) | |
487 | webpage = self._download_webpage(url, display_id) | |
488 | video_id = self._search_regex( | |
489 | r'data-mid=(["\'])(?P<id>(?:(?!\1).)+)\1', webpage, 'video_id', group='id') | |
490 | return { | |
491 | '_type': 'url_transparent', | |
492 | 'ie_key': 'NPO', | |
493 | 'url': 'npo:%s' % video_id, | |
494 | 'display_id': display_id | |
495 | } | |
496 | ||
497 | ||
498 | class SchoolTVIE(NPODataMidEmbedIE): | |
89f26028 | 499 | IE_NAME = 'schooltv' |
500 | _VALID_URL = r'https?://(?:www\.)?schooltv\.nl/video/(?P<id>[^/?#&]+)' | |
501 | ||
502 | _TEST = { | |
503 | 'url': 'http://www.schooltv.nl/video/ademhaling-de-hele-dag-haal-je-adem-maar-wat-gebeurt-er-dan-eigenlijk-in-je-lichaam/', | |
504 | 'info_dict': { | |
505 | 'id': 'WO_NTR_429477', | |
506 | 'display_id': 'ademhaling-de-hele-dag-haal-je-adem-maar-wat-gebeurt-er-dan-eigenlijk-in-je-lichaam', | |
507 | 'title': 'Ademhaling: De hele dag haal je adem. Maar wat gebeurt er dan eigenlijk in je lichaam?', | |
508 | 'ext': 'mp4', | |
509 | 'description': 'md5:abfa0ff690adb73fd0297fd033aaa631' | |
510 | }, | |
511 | 'params': { | |
512 | # Skip because of m3u8 download | |
513 | 'skip_download': True | |
514 | } | |
515 | } | |
516 | ||
0dc5a86a S |
517 | |
518 | class HetKlokhuisIE(NPODataMidEmbedIE): | |
dbc01fdb | 519 | IE_NAME = 'hetklokhuis' |
197224b7 | 520 | _VALID_URL = r'https?://(?:www\.)?hetklokhuis\.nl/[^/]+/\d+/(?P<id>[^/?#&]+)' |
0dc5a86a S |
521 | |
522 | _TEST = { | |
523 | 'url': 'http://hetklokhuis.nl/tv-uitzending/3471/Zwaartekrachtsgolven', | |
524 | 'info_dict': { | |
525 | 'id': 'VPWON_1260528', | |
526 | 'display_id': 'Zwaartekrachtsgolven', | |
527 | 'ext': 'm4v', | |
528 | 'title': 'Het Klokhuis: Zwaartekrachtsgolven', | |
529 | 'description': 'md5:c94f31fb930d76c2efa4a4a71651dd48', | |
530 | 'upload_date': '20170223', | |
531 | }, | |
532 | 'params': { | |
533 | 'skip_download': True | |
89f26028 | 534 | } |
0dc5a86a | 535 | } |
89f26028 | 536 | |
537 | ||
a3d8b381 S |
538 | class NPOPlaylistBaseIE(NPOIE): |
539 | def _real_extract(self, url): | |
540 | playlist_id = self._match_id(url) | |
541 | ||
542 | webpage = self._download_webpage(url, playlist_id) | |
543 | ||
544 | entries = [ | |
545 | self.url_result('npo:%s' % video_id if not video_id.startswith('http') else video_id) | |
5742c18b | 546 | for video_id in orderedSet(re.findall(self._PLAYLIST_ENTRY_RE, webpage)) |
a3d8b381 S |
547 | ] |
548 | ||
549 | playlist_title = self._html_search_regex( | |
550 | self._PLAYLIST_TITLE_RE, webpage, 'playlist title', | |
551 | default=None) or self._og_search_title(webpage) | |
552 | ||
553 | return self.playlist_result(entries, playlist_id, playlist_title) | |
554 | ||
555 | ||
556 | class VPROIE(NPOPlaylistBaseIE): | |
464e7924 | 557 | IE_NAME = 'vpro' |
b1d79888 S |
558 | _VALID_URL = r'https?://(?:www\.)?(?:(?:tegenlicht\.)?vpro|2doc)\.nl/(?:[^/]+/)*(?P<id>[^/]+)\.html' |
559 | _PLAYLIST_TITLE_RE = (r'<h1[^>]+class=["\'].*?\bmedia-platform-title\b.*?["\'][^>]*>([^<]+)', | |
560 | r'<h5[^>]+class=["\'].*?\bmedia-platform-subtitle\b.*?["\'][^>]*>([^<]+)') | |
a3d8b381 | 561 | _PLAYLIST_ENTRY_RE = r'data-media-id="([^"]+)"' |
d0df9292 JMF |
562 | |
563 | _TESTS = [ | |
564 | { | |
565 | 'url': 'http://tegenlicht.vpro.nl/afleveringen/2012-2013/de-toekomst-komt-uit-afrika.html', | |
566 | 'md5': 'f8065e4e5a7824068ed3c7e783178f2c', | |
567 | 'info_dict': { | |
568 | 'id': 'VPWON_1169289', | |
569 | 'ext': 'm4v', | |
5ba761eb | 570 | 'title': 'De toekomst komt uit Afrika', |
611ac379 | 571 | 'description': 'md5:52cf4eefbc96fffcbdc06d024147abea', |
d0df9292 JMF |
572 | 'upload_date': '20130225', |
573 | }, | |
ddb19772 | 574 | 'skip': 'Video gone', |
d0df9292 | 575 | }, |
611ac379 S |
576 | { |
577 | 'url': 'http://www.vpro.nl/programmas/2doc/2015/sergio-herman.html', | |
578 | 'info_dict': { | |
579 | 'id': 'sergio-herman', | |
ddb19772 | 580 | 'title': 'sergio herman: fucking perfect', |
611ac379 S |
581 | }, |
582 | 'playlist_count': 2, | |
574f42d7 S |
583 | }, |
584 | { | |
585 | # playlist with youtube embed | |
586 | 'url': 'http://www.vpro.nl/programmas/2doc/2015/education-education.html', | |
587 | 'info_dict': { | |
588 | 'id': 'education-education', | |
ddb19772 | 589 | 'title': 'education education', |
574f42d7 S |
590 | }, |
591 | 'playlist_count': 2, | |
b1d79888 S |
592 | }, |
593 | { | |
594 | 'url': 'http://www.2doc.nl/documentaires/series/2doc/2015/oktober/de-tegenprestatie.html', | |
595 | 'info_dict': { | |
596 | 'id': 'de-tegenprestatie', | |
597 | 'title': 'De Tegenprestatie', | |
598 | }, | |
599 | 'playlist_count': 2, | |
600 | }, { | |
601 | 'url': 'http://www.2doc.nl/speel~VARA_101375237~mh17-het-verdriet-van-nederland~.html', | |
602 | 'info_dict': { | |
603 | 'id': 'VARA_101375237', | |
604 | 'ext': 'm4v', | |
605 | 'title': 'MH17: Het verdriet van Nederland', | |
606 | 'description': 'md5:09e1a37c1fdb144621e22479691a9f18', | |
607 | 'upload_date': '20150716', | |
608 | }, | |
609 | 'params': { | |
610 | # Skip because of m3u8 download | |
611 | 'skip_download': True | |
612 | }, | |
611ac379 | 613 | } |
d0df9292 JMF |
614 | ] |
615 | ||
03f32a7e | 616 | |
a3d8b381 | 617 | class WNLIE(NPOPlaylistBaseIE): |
f9dd86a1 | 618 | IE_NAME = 'wnl' |
03f32a7e | 619 | _VALID_URL = r'https?://(?:www\.)?omroepwnl\.nl/video/detail/(?P<id>[^/]+)__\d+' |
a3d8b381 S |
620 | _PLAYLIST_TITLE_RE = r'(?s)<h1[^>]+class="subject"[^>]*>(.+?)</h1>' |
621 | _PLAYLIST_ENTRY_RE = r'<a[^>]+href="([^"]+)"[^>]+class="js-mid"[^>]*>Deel \d+' | |
03f32a7e | 622 | |
a3d8b381 | 623 | _TESTS = [{ |
03f32a7e S |
624 | 'url': 'http://www.omroepwnl.nl/video/detail/vandaag-de-dag-6-mei__060515', |
625 | 'info_dict': { | |
626 | 'id': 'vandaag-de-dag-6-mei', | |
627 | 'title': 'Vandaag de Dag 6 mei', | |
628 | }, | |
629 | 'playlist_count': 4, | |
a3d8b381 | 630 | }] |
5742c18b S |
631 | |
632 | ||
633 | class AndereTijdenIE(NPOPlaylistBaseIE): | |
f9dd86a1 | 634 | IE_NAME = 'anderetijden' |
5742c18b S |
635 | _VALID_URL = r'https?://(?:www\.)?anderetijden\.nl/programma/(?:[^/]+/)+(?P<id>[^/?#&]+)' |
636 | _PLAYLIST_TITLE_RE = r'(?s)<h1[^>]+class=["\'].*?\bpage-title\b.*?["\'][^>]*>(.+?)</h1>' | |
637 | _PLAYLIST_ENTRY_RE = r'<figure[^>]+class=["\']episode-container episode-page["\'][^>]+data-prid=["\'](.+?)["\']' | |
638 | ||
639 | _TESTS = [{ | |
640 | 'url': 'http://anderetijden.nl/programma/1/Andere-Tijden/aflevering/676/Duitse-soldaten-over-de-Slag-bij-Arnhem', | |
641 | 'info_dict': { | |
642 | 'id': 'Duitse-soldaten-over-de-Slag-bij-Arnhem', | |
643 | 'title': 'Duitse soldaten over de Slag bij Arnhem', | |
644 | }, | |
645 | 'playlist_count': 3, | |
646 | }] |