]>
Commit | Line | Data |
---|---|---|
91a6adde | 1 | import base64 |
10db0d2f | 2 | import io |
ac668111 | 3 | import struct |
91a6adde JMF |
4 | |
5 | from .common import InfoExtractor | |
ac668111 | 6 | from ..compat import compat_b64decode |
91a6adde | 7 | from ..utils import ( |
ce73839f | 8 | ExtractorError, |
ac668111 | 9 | determine_ext, |
f3bff94c | 10 | float_or_none, |
10db0d2f | 11 | qualities, |
2b9faf55 | 12 | remove_end, |
3e769682 | 13 | remove_start, |
5edb8dfe | 14 | try_get, |
91a6adde JMF |
15 | ) |
16 | ||
2b9faf55 | 17 | |
91a6adde JMF |
18 | class RTVEALaCartaIE(InfoExtractor): |
19 | IE_NAME = 'rtve.es:alacarta' | |
20 | IE_DESC = 'RTVE a la carta' | |
92519402 | 21 | _VALID_URL = r'https?://(?:www\.)?rtve\.es/(m/)?(alacarta/videos|filmoteca)/[^/]+/[^/]+/(?P<id>\d+)' |
91a6adde | 22 | |
2b9faf55 | 23 | _TESTS = [{ |
91a6adde | 24 | 'url': 'http://www.rtve.es/alacarta/videos/balonmano/o-swiss-cup-masculina-final-espana-suecia/2491869/', |
9aeaf730 | 25 | 'md5': '1d49b7e1ca7a7502c56a4bf1b60f1b43', |
91a6adde JMF |
26 | 'info_dict': { |
27 | 'id': '2491869', | |
28 | 'ext': 'mp4', | |
29 | 'title': 'Balonmano - Swiss Cup masculina. Final: España-Suecia', | |
f3bff94c | 30 | 'duration': 5024.566, |
10db0d2f | 31 | 'series': 'Balonmano', |
91a6adde | 32 | }, |
10db0d2f | 33 | 'expected_warnings': ['Failed to download MPD manifest', 'Failed to download m3u8 information'], |
2b9faf55 PH |
34 | }, { |
35 | 'note': 'Live stream', | |
36 | 'url': 'http://www.rtve.es/alacarta/videos/television/24h-live/1694255/', | |
37 | 'info_dict': { | |
38 | 'id': '1694255', | |
10db0d2f | 39 | 'ext': 'mp4', |
40 | 'title': 're:^24H LIVE [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$', | |
41 | 'is_live': True, | |
42 | }, | |
43 | 'params': { | |
44 | 'skip_download': 'live stream', | |
cc57bd33 | 45 | }, |
2c53bd51 GF |
46 | }, { |
47 | 'url': 'http://www.rtve.es/alacarta/videos/servir-y-proteger/servir-proteger-capitulo-104/4236788/', | |
10db0d2f | 48 | 'md5': 'd850f3c8731ea53952ebab489cf81cbf', |
2c53bd51 GF |
49 | 'info_dict': { |
50 | 'id': '4236788', | |
51 | 'ext': 'mp4', | |
10db0d2f | 52 | 'title': 'Servir y proteger - Capítulo 104', |
2c53bd51 GF |
53 | 'duration': 3222.0, |
54 | }, | |
10db0d2f | 55 | 'expected_warnings': ['Failed to download MPD manifest', 'Failed to download m3u8 information'], |
cd596028 JMF |
56 | }, { |
57 | 'url': 'http://www.rtve.es/m/alacarta/videos/cuentame-como-paso/cuentame-como-paso-t16-ultimo-minuto-nuestra-vida-capitulo-276/2969138/?media=tve', | |
58 | 'only_matching': True, | |
4c718d3c JMF |
59 | }, { |
60 | 'url': 'http://www.rtve.es/filmoteca/no-do/not-1-introduccion-primer-noticiario-espanol/1465256/', | |
61 | 'only_matching': True, | |
2b9faf55 | 62 | }] |
91a6adde | 63 | |
e9f65f87 | 64 | def _real_initialize(self): |
8b7539d2 | 65 | user_agent_b64 = base64.b64encode(self.get_param('http_headers')['User-Agent'].encode('utf-8')).decode('utf-8') |
10db0d2f | 66 | self._manager = self._download_json( |
e9f65f87 | 67 | 'http://www.rtve.es/odin/loki/' + user_agent_b64, |
10db0d2f | 68 | None, 'Fetching manager info')['manager'] |
69 | ||
70 | @staticmethod | |
71 | def _decrypt_url(png): | |
72 | encrypted_data = io.BytesIO(compat_b64decode(png)[8:]) | |
73 | while True: | |
ac668111 | 74 | length = struct.unpack('!I', encrypted_data.read(4))[0] |
10db0d2f | 75 | chunk_type = encrypted_data.read(4) |
76 | if chunk_type == b'IEND': | |
77 | break | |
78 | data = encrypted_data.read(length) | |
79 | if chunk_type == b'tEXt': | |
80 | alphabet_data, text = data.split(b'\0') | |
81 | quality, url_data = text.split(b'%%') | |
82 | alphabet = [] | |
83 | e = 0 | |
84 | d = 0 | |
19a03940 | 85 | for l in alphabet_data.decode('iso-8859-1'): |
10db0d2f | 86 | if d == 0: |
87 | alphabet.append(l) | |
88 | d = e = (e + 1) % 4 | |
89 | else: | |
90 | d -= 1 | |
91 | url = '' | |
92 | f = 0 | |
93 | e = 3 | |
94 | b = 1 | |
19a03940 | 95 | for letter in url_data.decode('iso-8859-1'): |
10db0d2f | 96 | if f == 0: |
97 | l = int(letter) * 10 | |
98 | f = 1 | |
99 | else: | |
100 | if e == 0: | |
101 | l += int(letter) | |
102 | url += alphabet[l] | |
103 | e = (b + 3) % 4 | |
104 | f = 0 | |
105 | b += 1 | |
106 | else: | |
107 | e -= 1 | |
108 | ||
109 | yield quality.decode(), url | |
110 | encrypted_data.read(4) # CRC | |
111 | ||
112 | def _extract_png_formats(self, video_id): | |
113 | png = self._download_webpage( | |
114 | 'http://www.rtve.es/ztnr/movil/thumbnail/%s/videos/%s.png' % (self._manager, video_id), | |
115 | video_id, 'Downloading url information', query={'q': 'v2'}) | |
116 | q = qualities(['Media', 'Alta', 'HQ', 'HD_READY', 'HD_FULL']) | |
117 | formats = [] | |
118 | for quality, video_url in self._decrypt_url(png): | |
119 | ext = determine_ext(video_url) | |
120 | if ext == 'm3u8': | |
121 | formats.extend(self._extract_m3u8_formats( | |
122 | video_url, video_id, 'mp4', 'm3u8_native', | |
123 | m3u8_id='hls', fatal=False)) | |
124 | elif ext == 'mpd': | |
125 | formats.extend(self._extract_mpd_formats( | |
126 | video_url, video_id, 'dash', fatal=False)) | |
127 | else: | |
128 | formats.append({ | |
129 | 'format_id': quality, | |
130 | 'quality': q(quality), | |
131 | 'url': video_url, | |
132 | }) | |
133 | self._sort_formats(formats) | |
134 | return formats | |
e9f65f87 | 135 | |
91a6adde | 136 | def _real_extract(self, url): |
10db0d2f | 137 | video_id = self._match_id(url) |
91a6adde JMF |
138 | info = self._download_json( |
139 | 'http://www.rtve.es/api/videos/%s/config/alacarta_videos.json' % video_id, | |
140 | video_id)['page']['items'][0] | |
ce73839f JMF |
141 | if info['state'] == 'DESPU': |
142 | raise ExtractorError('The video is no longer available', expected=True) | |
10db0d2f | 143 | title = info['title'].strip() |
144 | formats = self._extract_png_formats(video_id) | |
2c53bd51 | 145 | |
25ac63ed | 146 | subtitles = None |
10db0d2f | 147 | sbt_file = info.get('sbtFile') |
148 | if sbt_file: | |
149 | subtitles = self.extract_subtitles(video_id, sbt_file) | |
150 | ||
151 | is_live = info.get('live') is True | |
25ac63ed | 152 | |
91a6adde JMF |
153 | return { |
154 | 'id': video_id, | |
39ca3b5c | 155 | 'title': title, |
2c53bd51 | 156 | 'formats': formats, |
2b9faf55 | 157 | 'thumbnail': info.get('image'), |
25ac63ed | 158 | 'subtitles': subtitles, |
10db0d2f | 159 | 'duration': float_or_none(info.get('duration'), 1000), |
160 | 'is_live': is_live, | |
161 | 'series': info.get('programTitle'), | |
2b9faf55 PH |
162 | } |
163 | ||
25ac63ed JMF |
164 | def _get_subtitles(self, video_id, sub_file): |
165 | subs = self._download_json( | |
166 | sub_file + '.json', video_id, | |
167 | 'Downloading subtitles info')['page']['items'] | |
9c665ab7 PH |
168 | return dict( |
169 | (s['lang'], [{'ext': 'vtt', 'url': s['src']}]) | |
25ac63ed JMF |
170 | for s in subs) |
171 | ||
b68eedba | 172 | |
5edb8dfe | 173 | class RTVEAudioIE(RTVEALaCartaIE): |
174 | IE_NAME = 'rtve.es:audio' | |
175 | IE_DESC = 'RTVE audio' | |
176 | _VALID_URL = r'https?://(?:www\.)?rtve\.es/(alacarta|play)/audios/[^/]+/[^/]+/(?P<id>[0-9]+)' | |
177 | ||
178 | _TESTS = [{ | |
179 | 'url': 'https://www.rtve.es/alacarta/audios/a-hombros-de-gigantes/palabra-ingeniero-codigos-informaticos-27-04-21/5889192/', | |
180 | 'md5': 'ae06d27bff945c4e87a50f89f6ce48ce', | |
181 | 'info_dict': { | |
182 | 'id': '5889192', | |
183 | 'ext': 'mp3', | |
184 | 'title': 'Códigos informáticos', | |
185 | 'thumbnail': r're:https?://.+/1598856591583.jpg', | |
186 | 'duration': 349.440, | |
187 | 'series': 'A hombros de gigantes', | |
188 | }, | |
189 | }, { | |
190 | 'url': 'https://www.rtve.es/play/audios/en-radio-3/ignatius-farray/5791165/', | |
191 | 'md5': '072855ab89a9450e0ba314c717fa5ebc', | |
192 | 'info_dict': { | |
193 | 'id': '5791165', | |
194 | 'ext': 'mp3', | |
195 | 'title': 'Ignatius Farray', | |
196 | 'thumbnail': r're:https?://.+/1613243011863.jpg', | |
197 | 'duration': 3559.559, | |
198 | 'series': 'En Radio 3' | |
199 | }, | |
200 | }, { | |
201 | 'url': 'https://www.rtve.es/play/audios/frankenstein-o-el-moderno-prometeo/capitulo-26-ultimo-muerte-victor-juan-jose-plans-mary-shelley/6082623/', | |
202 | 'md5': '0eadab248cc8dd193fa5765712e84d5c', | |
203 | 'info_dict': { | |
204 | 'id': '6082623', | |
205 | 'ext': 'mp3', | |
206 | 'title': 'Capítulo 26 y último: La muerte de Victor', | |
207 | 'thumbnail': r're:https?://.+/1632147445707.jpg', | |
208 | 'duration': 3174.086, | |
209 | 'series': 'Frankenstein o el moderno Prometeo' | |
210 | }, | |
211 | }] | |
212 | ||
213 | def _extract_png_formats(self, audio_id): | |
214 | """ | |
215 | This function retrieves media related png thumbnail which obfuscate | |
216 | valuable information about the media. This information is decrypted | |
217 | via base class _decrypt_url function providing media quality and | |
218 | media url | |
219 | """ | |
220 | png = self._download_webpage( | |
221 | 'http://www.rtve.es/ztnr/movil/thumbnail/%s/audios/%s.png' % | |
222 | (self._manager, audio_id), | |
223 | audio_id, 'Downloading url information', query={'q': 'v2'}) | |
224 | q = qualities(['Media', 'Alta', 'HQ', 'HD_READY', 'HD_FULL']) | |
225 | formats = [] | |
226 | for quality, audio_url in self._decrypt_url(png): | |
227 | ext = determine_ext(audio_url) | |
228 | if ext == 'm3u8': | |
229 | formats.extend(self._extract_m3u8_formats( | |
230 | audio_url, audio_id, 'mp4', 'm3u8_native', | |
231 | m3u8_id='hls', fatal=False)) | |
232 | elif ext == 'mpd': | |
233 | formats.extend(self._extract_mpd_formats( | |
234 | audio_url, audio_id, 'dash', fatal=False)) | |
235 | else: | |
236 | formats.append({ | |
237 | 'format_id': quality, | |
238 | 'quality': q(quality), | |
239 | 'url': audio_url, | |
240 | }) | |
241 | self._sort_formats(formats) | |
242 | return formats | |
243 | ||
244 | def _real_extract(self, url): | |
245 | audio_id = self._match_id(url) | |
246 | info = self._download_json( | |
247 | 'https://www.rtve.es/api/audios/%s.json' % audio_id, | |
248 | audio_id)['page']['items'][0] | |
249 | ||
250 | return { | |
251 | 'id': audio_id, | |
252 | 'title': info['title'].strip(), | |
253 | 'thumbnail': info.get('thumbnail'), | |
254 | 'duration': float_or_none(info.get('duration'), 1000), | |
255 | 'series': try_get(info, lambda x: x['programInfo']['title']), | |
256 | 'formats': self._extract_png_formats(audio_id), | |
257 | } | |
258 | ||
259 | ||
10db0d2f | 260 | class RTVEInfantilIE(RTVEALaCartaIE): |
b68eedba JMF |
261 | IE_NAME = 'rtve.es:infantil' |
262 | IE_DESC = 'RTVE infantil' | |
10db0d2f | 263 | _VALID_URL = r'https?://(?:www\.)?rtve\.es/infantil/serie/[^/]+/video/[^/]+/(?P<id>[0-9]+)/' |
d5b55939 EF |
264 | |
265 | _TESTS = [{ | |
266 | 'url': 'http://www.rtve.es/infantil/serie/cleo/video/maneras-vivir/3040283/', | |
10db0d2f | 267 | 'md5': '5747454717aedf9f9fdf212d1bcfc48d', |
d5b55939 EF |
268 | 'info_dict': { |
269 | 'id': '3040283', | |
270 | 'ext': 'mp4', | |
271 | 'title': 'Maneras de vivir', | |
10db0d2f | 272 | 'thumbnail': r're:https?://.+/1426182947956\.JPG', |
d5b55939 EF |
273 | 'duration': 357.958, |
274 | }, | |
10db0d2f | 275 | 'expected_warnings': ['Failed to download MPD manifest', 'Failed to download m3u8 information'], |
b68eedba | 276 | }] |
d5b55939 | 277 | |
d5b55939 | 278 | |
10db0d2f | 279 | class RTVELiveIE(RTVEALaCartaIE): |
2b9faf55 PH |
280 | IE_NAME = 'rtve.es:live' |
281 | IE_DESC = 'RTVE.es live streams' | |
92519402 | 282 | _VALID_URL = r'https?://(?:www\.)?rtve\.es/directo/(?P<id>[a-zA-Z0-9-]+)' |
2b9faf55 PH |
283 | |
284 | _TESTS = [{ | |
3e769682 | 285 | 'url': 'http://www.rtve.es/directo/la-1/', |
2b9faf55 | 286 | 'info_dict': { |
3e769682 JMF |
287 | 'id': 'la-1', |
288 | 'ext': 'mp4', | |
10db0d2f | 289 | 'title': 're:^La 1 [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$', |
2b9faf55 PH |
290 | }, |
291 | 'params': { | |
292 | 'skip_download': 'live stream', | |
293 | } | |
294 | }] | |
295 | ||
296 | def _real_extract(self, url): | |
5ad28e7f | 297 | mobj = self._match_valid_url(url) |
2b9faf55 PH |
298 | video_id = mobj.group('id') |
299 | ||
300 | webpage = self._download_webpage(url, video_id) | |
3e769682 JMF |
301 | title = remove_end(self._og_search_title(webpage), ' en directo en RTVE.es') |
302 | title = remove_start(title, 'Estoy viendo ') | |
2b9faf55 PH |
303 | |
304 | vidplayer_id = self._search_regex( | |
b63005f5 S |
305 | (r'playerId=player([0-9]+)', |
306 | r'class=["\'].*?\blive_mod\b.*?["\'][^>]+data-assetid=["\'](\d+)', | |
307 | r'data-id=["\'](\d+)'), | |
308 | webpage, 'internal video ID') | |
2b9faf55 | 309 | |
2b9faf55 PH |
310 | return { |
311 | 'id': video_id, | |
39ca3b5c | 312 | 'title': title, |
10db0d2f | 313 | 'formats': self._extract_png_formats(vidplayer_id), |
3e769682 | 314 | 'is_live': True, |
91a6adde | 315 | } |
233b58de JMF |
316 | |
317 | ||
318 | class RTVETelevisionIE(InfoExtractor): | |
319 | IE_NAME = 'rtve.es:television' | |
92519402 | 320 | _VALID_URL = r'https?://(?:www\.)?rtve\.es/television/[^/]+/[^/]+/(?P<id>\d+).shtml' |
233b58de JMF |
321 | |
322 | _TEST = { | |
323 | 'url': 'http://www.rtve.es/television/20160628/revolucion-del-movil/1364141.shtml', | |
324 | 'info_dict': { | |
325 | 'id': '3069778', | |
326 | 'ext': 'mp4', | |
327 | 'title': 'Documentos TV - La revolución del móvil', | |
328 | 'duration': 3496.948, | |
329 | }, | |
330 | 'params': { | |
331 | 'skip_download': True, | |
332 | }, | |
333 | } | |
334 | ||
335 | def _real_extract(self, url): | |
336 | page_id = self._match_id(url) | |
337 | webpage = self._download_webpage(url, page_id) | |
338 | ||
339 | alacarta_url = self._search_regex( | |
340 | r'data-location="alacarta_videos"[^<]+url":"(http://www\.rtve\.es/alacarta.+?)&', | |
341 | webpage, 'alacarta url', default=None) | |
342 | if alacarta_url is None: | |
343 | raise ExtractorError( | |
344 | 'The webpage doesn\'t contain any video', expected=True) | |
345 | ||
346 | return self.url_result(alacarta_url, ie=RTVEALaCartaIE.ie_key()) |