]>
Commit | Line | Data |
---|---|---|
a2e6db36 | 1 | # coding: utf-8 |
919052d0 | 2 | from __future__ import unicode_literals |
a2e6db36 | 3 | |
d5822b96 PH |
4 | import re |
5 | ||
6 | from .common import InfoExtractor | |
fb47cb5b | 7 | from ..compat import compat_str |
d5822b96 | 8 | from ..utils import ( |
a5c1d955 | 9 | determine_ext, |
ec5e77c5 | 10 | float_or_none, |
fb47cb5b | 11 | int_or_none, |
34921b43 | 12 | join_nonempty, |
ec5e77c5 | 13 | merge_dicts, |
fb47cb5b S |
14 | NO_DEFAULT, |
15 | orderedSet, | |
16 | parse_codecs, | |
17 | qualities, | |
5f9aaac8 | 18 | traverse_obj, |
fb47cb5b S |
19 | try_get, |
20 | unified_timestamp, | |
21 | update_url_query, | |
3052a30d | 22 | url_or_none, |
fb47cb5b | 23 | urljoin, |
d5822b96 | 24 | ) |
0b7c2485 | 25 | |
fb47cb5b S |
26 | |
27 | class ZDFBaseIE(InfoExtractor): | |
9cf26b6e | 28 | _GEO_COUNTRIES = ['DE'] |
ec5e77c5 | 29 | _QUALITIES = ('auto', 'low', 'med', 'high', 'veryhigh', 'hd') |
a2e6db36 | 30 | |
ec5e77c5 | 31 | def _call_api(self, url, video_id, item, api_token=None, referrer=None): |
32 | headers = {} | |
33 | if api_token: | |
34 | headers['Api-Auth'] = 'Bearer %s' % api_token | |
35 | if referrer: | |
36 | headers['Referer'] = referrer | |
37 | return self._download_json( | |
38 | url, video_id, 'Downloading JSON %s' % item, headers=headers) | |
a5c1d955 | 39 | |
fb47cb5b S |
40 | @staticmethod |
41 | def _extract_subtitles(src): | |
42 | subtitles = {} | |
43 | for caption in try_get(src, lambda x: x['captions'], list) or []: | |
3052a30d S |
44 | subtitle_url = url_or_none(caption.get('uri')) |
45 | if subtitle_url: | |
fb47cb5b S |
46 | lang = caption.get('language', 'deu') |
47 | subtitles.setdefault(lang, []).append({ | |
48 | 'url': subtitle_url, | |
49 | }) | |
50 | return subtitles | |
51 | ||
52 | def _extract_format(self, video_id, formats, format_urls, meta): | |
3052a30d | 53 | format_url = url_or_none(meta.get('url')) |
600e9003 | 54 | if not format_url or format_url in format_urls: |
fb47cb5b S |
55 | return |
56 | format_urls.add(format_url) | |
600e9003 | 57 | |
58 | mime_type, ext = meta.get('mimeType'), determine_ext(format_url) | |
fb47cb5b | 59 | if mime_type == 'application/x-mpegURL' or ext == 'm3u8': |
600e9003 | 60 | new_formats = self._extract_m3u8_formats( |
fb47cb5b | 61 | format_url, video_id, 'mp4', m3u8_id='hls', |
600e9003 | 62 | entry_protocol='m3u8_native', fatal=False) |
fb47cb5b | 63 | elif mime_type == 'application/f4m+xml' or ext == 'f4m': |
600e9003 | 64 | new_formats = self._extract_f4m_formats( |
65 | update_url_query(format_url, {'hdcore': '3.7.0'}), video_id, f4m_id='hds', fatal=False) | |
fb47cb5b S |
66 | else: |
67 | f = parse_codecs(meta.get('mimeCodec')) | |
600e9003 | 68 | if not f and meta.get('type'): |
69 | data = meta['type'].split('_') | |
70 | if try_get(data, lambda x: x[2]) == ext: | |
71 | f = {'vcodec': data[0], 'acodec': data[1]} | |
fb47cb5b S |
72 | f.update({ |
73 | 'url': format_url, | |
34921b43 | 74 | 'format_id': join_nonempty('http', meta.get('type'), meta.get('quality')), |
fb47cb5b | 75 | }) |
600e9003 | 76 | new_formats = [f] |
77 | formats.extend(merge_dicts(f, { | |
34921b43 | 78 | 'format_note': join_nonempty('quality', 'class', from_dict=meta, delim=', '), |
600e9003 | 79 | 'language': meta.get('language'), |
80 | 'language_preference': 10 if meta.get('class') == 'main' else -10 if meta.get('class') == 'ad' else -1, | |
81 | 'quality': qualities(self._QUALITIES)(meta.get('quality')), | |
82 | }) for f in new_formats) | |
fb47cb5b | 83 | |
ec5e77c5 | 84 | def _extract_ptmd(self, ptmd_url, video_id, api_token, referrer): |
50de3dba | 85 | ptmd = self._call_api( |
ec5e77c5 | 86 | ptmd_url, video_id, 'metadata', api_token, referrer) |
87 | ||
88 | content_id = ptmd.get('basename') or ptmd_url.split('/')[-1] | |
fb47cb5b | 89 | |
a5c1d955 | 90 | formats = [] |
fb47cb5b S |
91 | track_uris = set() |
92 | for p in ptmd['priorityList']: | |
93 | formitaeten = p.get('formitaeten') | |
94 | if not isinstance(formitaeten, list): | |
a5c1d955 | 95 | continue |
fb47cb5b S |
96 | for f in formitaeten: |
97 | f_qualities = f.get('qualities') | |
98 | if not isinstance(f_qualities, list): | |
99 | continue | |
100 | for quality in f_qualities: | |
101 | tracks = try_get(quality, lambda x: x['audio']['tracks'], list) | |
102 | if not tracks: | |
103 | continue | |
104 | for track in tracks: | |
105 | self._extract_format( | |
ec5e77c5 | 106 | content_id, formats, track_uris, { |
fb47cb5b S |
107 | 'url': track.get('uri'), |
108 | 'type': f.get('type'), | |
109 | 'mimeType': f.get('mimeType'), | |
110 | 'quality': quality.get('quality'), | |
600e9003 | 111 | 'class': track.get('class'), |
fb47cb5b S |
112 | 'language': track.get('language'), |
113 | }) | |
600e9003 | 114 | self._sort_formats(formats, ('hasaud', 'res', 'quality', 'language_preference')) |
fb47cb5b | 115 | |
ec5e77c5 | 116 | duration = float_or_none(try_get( |
117 | ptmd, lambda x: x['attributes']['duration']['value']), scale=1000) | |
118 | ||
119 | return { | |
120 | 'extractor_key': ZDFIE.ie_key(), | |
121 | 'id': content_id, | |
122 | 'duration': duration, | |
123 | 'formats': formats, | |
124 | 'subtitles': self._extract_subtitles(ptmd), | |
125 | } | |
126 | ||
127 | def _extract_player(self, webpage, video_id, fatal=True): | |
128 | return self._parse_json( | |
129 | self._search_regex( | |
130 | r'(?s)data-zdfplayer-jsb=(["\'])(?P<json>{.+?})\1', webpage, | |
131 | 'player JSON', default='{}' if not fatal else NO_DEFAULT, | |
132 | group='json'), | |
133 | video_id) | |
134 | ||
135 | ||
136 | class ZDFIE(ZDFBaseIE): | |
137 | _VALID_URL = r'https?://www\.zdf\.de/(?:[^/]+/)*(?P<id>[^/?#&]+)\.html' | |
138 | _TESTS = [{ | |
50e93e03 | 139 | # Same as https://www.phoenix.de/sendungen/ereignisse/corona-nachgehakt/wohin-fuehrt-der-protest-in-der-pandemie-a-2050630.html |
140 | 'url': 'https://www.zdf.de/politik/phoenix-sendungen/wohin-fuehrt-der-protest-in-der-pandemie-100.html', | |
141 | 'md5': '34ec321e7eb34231fd88616c65c92db0', | |
142 | 'info_dict': { | |
143 | 'id': '210222_phx_nachgehakt_corona_protest', | |
144 | 'ext': 'mp4', | |
145 | 'title': 'Wohin führt der Protest in der Pandemie?', | |
146 | 'description': 'md5:7d643fe7f565e53a24aac036b2122fbd', | |
147 | 'duration': 1691, | |
148 | 'timestamp': 1613948400, | |
149 | 'upload_date': '20210221', | |
150 | }, | |
151 | 'skip': 'No longer available: "Diese Seite wurde leider nicht gefunden"', | |
152 | }, { | |
153 | # Same as https://www.3sat.de/film/ab-18/10-wochen-sommer-108.html | |
154 | 'url': 'https://www.zdf.de/dokumentation/ab-18/10-wochen-sommer-102.html', | |
155 | 'md5': '0aff3e7bc72c8813f5e0fae333316a1d', | |
156 | 'info_dict': { | |
157 | 'id': '141007_ab18_10wochensommer_film', | |
158 | 'ext': 'mp4', | |
159 | 'title': 'Ab 18! - 10 Wochen Sommer', | |
160 | 'description': 'md5:8253f41dc99ce2c3ff892dac2d65fe26', | |
161 | 'duration': 2660, | |
162 | 'timestamp': 1608604200, | |
163 | 'upload_date': '20201222', | |
164 | }, | |
165 | 'skip': 'No longer available: "Diese Seite wurde leider nicht gefunden"', | |
166 | }, { | |
f5c2c2c9 | 167 | 'url': 'https://www.zdf.de/nachrichten/heute-journal/heute-journal-vom-30-12-2021-100.html', |
ec5e77c5 | 168 | 'info_dict': { |
f5c2c2c9 | 169 | 'id': '211230_sendung_hjo', |
ec5e77c5 | 170 | 'ext': 'mp4', |
f5c2c2c9 | 171 | 'description': 'md5:47dff85977bde9fb8cba9e9c9b929839', |
172 | 'duration': 1890.0, | |
173 | 'upload_date': '20211230', | |
174 | 'chapters': list, | |
175 | 'thumbnail': 'md5:e65f459f741be5455c952cd820eb188e', | |
176 | 'title': 'heute journal vom 30.12.2021', | |
177 | 'timestamp': 1640897100, | |
178 | } | |
ec5e77c5 | 179 | }, { |
180 | 'url': 'https://www.zdf.de/dokumentation/terra-x/die-magie-der-farben-von-koenigspurpur-und-jeansblau-100.html', | |
181 | 'info_dict': { | |
182 | 'id': '151025_magie_farben2_tex', | |
183 | 'ext': 'mp4', | |
184 | 'title': 'Die Magie der Farben (2/2)', | |
185 | 'description': 'md5:a89da10c928c6235401066b60a6d5c1a', | |
186 | 'duration': 2615, | |
187 | 'timestamp': 1465021200, | |
188 | 'upload_date': '20160604', | |
f5c2c2c9 | 189 | 'thumbnail': 'https://www.zdf.de/assets/mauve-im-labor-100~768x432?cb=1464909117806', |
ec5e77c5 | 190 | }, |
5f9aaac8 | 191 | }, { |
192 | 'url': 'https://www.zdf.de/funk/druck-11790/funk-alles-ist-verzaubert-102.html', | |
193 | 'md5': '3d6f1049e9682178a11c54b91f3dd065', | |
194 | 'info_dict': { | |
195 | 'ext': 'mp4', | |
196 | 'id': 'video_funk_1770473', | |
197 | 'duration': 1278, | |
198 | 'description': 'Die Neue an der Schule verdreht Ismail den Kopf.', | |
199 | 'title': 'Alles ist verzaubert', | |
200 | 'timestamp': 1635520560, | |
f5c2c2c9 | 201 | 'upload_date': '20211029', |
202 | 'thumbnail': 'https://www.zdf.de/assets/teaser-funk-alles-ist-verzaubert-100~1920x1080?cb=1636466431799', | |
5f9aaac8 | 203 | }, |
ec5e77c5 | 204 | }, { |
205 | # Same as https://www.phoenix.de/sendungen/dokumentationen/gesten-der-maechtigen-i-a-89468.html?ref=suche | |
206 | 'url': 'https://www.zdf.de/politik/phoenix-sendungen/die-gesten-der-maechtigen-100.html', | |
207 | 'only_matching': True, | |
208 | }, { | |
209 | # Same as https://www.3sat.de/film/spielfilm/der-hauptmann-100.html | |
210 | 'url': 'https://www.zdf.de/filme/filme-sonstige/der-hauptmann-112.html', | |
211 | 'only_matching': True, | |
212 | }, { | |
213 | # Same as https://www.3sat.de/wissen/nano/nano-21-mai-2019-102.html, equal media ids | |
214 | 'url': 'https://www.zdf.de/wissen/nano/nano-21-mai-2019-102.html', | |
215 | 'only_matching': True, | |
216 | }, { | |
217 | 'url': 'https://www.zdf.de/service-und-hilfe/die-neue-zdf-mediathek/zdfmediathek-trailer-100.html', | |
218 | 'only_matching': True, | |
219 | }, { | |
220 | 'url': 'https://www.zdf.de/filme/taunuskrimi/die-lebenden-und-die-toten-1---ein-taunuskrimi-100.html', | |
221 | 'only_matching': True, | |
222 | }, { | |
223 | 'url': 'https://www.zdf.de/dokumentation/planet-e/planet-e-uebersichtsseite-weitere-dokumentationen-von-planet-e-100.html', | |
224 | 'only_matching': True, | |
5f9aaac8 | 225 | }, { |
50e93e03 | 226 | 'url': 'https://www.zdf.de/arte/todliche-flucht/page-video-artede-toedliche-flucht-16-100.html', |
227 | 'info_dict': { | |
228 | 'id': 'video_artede_083871-001-A', | |
229 | 'ext': 'mp4', | |
230 | 'title': 'Tödliche Flucht (1/6)', | |
231 | 'description': 'md5:e34f96a9a5f8abd839ccfcebad3d5315', | |
232 | 'duration': 3193.0, | |
233 | 'timestamp': 1641355200, | |
234 | 'upload_date': '20220105', | |
235 | }, | |
ec5e77c5 | 236 | }] |
237 | ||
238 | def _extract_entry(self, url, player, content, video_id): | |
239 | title = content.get('title') or content['teaserHeadline'] | |
240 | ||
241 | t = content['mainVideoContent']['http://zdf.de/rels/target'] | |
242 | ||
243 | ptmd_path = t.get('http://zdf.de/rels/streams/ptmd') | |
244 | ||
245 | if not ptmd_path: | |
5f9aaac8 | 246 | ptmd_path = traverse_obj( |
247 | t, ('streams', 'default', 'http://zdf.de/rels/streams/ptmd-template'), | |
248 | 'http://zdf.de/rels/streams/ptmd-template').replace( | |
ec5e77c5 | 249 | '{playerId}', 'ngplayer_2_4') |
250 | ||
251 | info = self._extract_ptmd( | |
252 | urljoin(url, ptmd_path), video_id, player['apiToken'], url) | |
253 | ||
fb47cb5b S |
254 | thumbnails = [] |
255 | layouts = try_get( | |
256 | content, lambda x: x['teaserImageRef']['layouts'], dict) | |
257 | if layouts: | |
258 | for layout_key, layout_url in layouts.items(): | |
3052a30d S |
259 | layout_url = url_or_none(layout_url) |
260 | if not layout_url: | |
fb47cb5b S |
261 | continue |
262 | thumbnail = { | |
263 | 'url': layout_url, | |
264 | 'format_id': layout_key, | |
265 | } | |
266 | mobj = re.search(r'(?P<width>\d+)x(?P<height>\d+)', layout_key) | |
267 | if mobj: | |
268 | thumbnail.update({ | |
269 | 'width': int(mobj.group('width')), | |
270 | 'height': int(mobj.group('height')), | |
271 | }) | |
272 | thumbnails.append(thumbnail) | |
a5c1d955 | 273 | |
f5c2c2c9 | 274 | chapter_marks = t.get('streamAnchorTag') or [] |
275 | chapter_marks.append({'anchorOffset': int_or_none(t.get('duration'))}) | |
276 | chapters = [{ | |
277 | 'start_time': chap.get('anchorOffset'), | |
278 | 'end_time': next_chap.get('anchorOffset'), | |
279 | 'title': chap.get('anchorLabel') | |
280 | } for chap, next_chap in zip(chapter_marks, chapter_marks[1:])] | |
281 | ||
ec5e77c5 | 282 | return merge_dicts(info, { |
fb47cb5b S |
283 | 'title': title, |
284 | 'description': content.get('leadParagraph') or content.get('teasertext'), | |
285 | 'duration': int_or_none(t.get('duration')), | |
286 | 'timestamp': unified_timestamp(content.get('editorialDate')), | |
287 | 'thumbnails': thumbnails, | |
f5c2c2c9 | 288 | 'chapters': chapters or None |
ec5e77c5 | 289 | }) |
d5822b96 | 290 | |
fb47cb5b | 291 | def _extract_regular(self, url, player, video_id): |
50de3dba | 292 | content = self._call_api( |
ec5e77c5 | 293 | player['content'], video_id, 'content', player['apiToken'], url) |
50de3dba | 294 | return self._extract_entry(player['content'], player, content, video_id) |
b6de53ea | 295 | |
fb47cb5b | 296 | def _extract_mobile(self, video_id): |
ec5e77c5 | 297 | video = self._download_json( |
fb47cb5b | 298 | 'https://zdf-cdn.live.cellular.de/mediathekV2/document/%s' % video_id, |
ec5e77c5 | 299 | video_id) |
300 | ||
301 | document = video['document'] | |
b6de53ea | 302 | |
fb47cb5b | 303 | title = document['titel'] |
ec5e77c5 | 304 | content_id = document['basename'] |
b6de53ea | 305 | |
fb47cb5b S |
306 | formats = [] |
307 | format_urls = set() | |
308 | for f in document['formitaeten']: | |
ec5e77c5 | 309 | self._extract_format(content_id, formats, format_urls, f) |
fb47cb5b S |
310 | self._sort_formats(formats) |
311 | ||
312 | thumbnails = [] | |
313 | teaser_bild = document.get('teaserBild') | |
314 | if isinstance(teaser_bild, dict): | |
315 | for thumbnail_key, thumbnail in teaser_bild.items(): | |
316 | thumbnail_url = try_get( | |
317 | thumbnail, lambda x: x['url'], compat_str) | |
318 | if thumbnail_url: | |
319 | thumbnails.append({ | |
320 | 'url': thumbnail_url, | |
321 | 'id': thumbnail_key, | |
322 | 'width': int_or_none(thumbnail.get('width')), | |
323 | 'height': int_or_none(thumbnail.get('height')), | |
324 | }) | |
b6de53ea | 325 | |
fb47cb5b | 326 | return { |
ec5e77c5 | 327 | 'id': content_id, |
fb47cb5b S |
328 | 'title': title, |
329 | 'description': document.get('beschreibung'), | |
330 | 'duration': int_or_none(document.get('length')), | |
ec5e77c5 | 331 | 'timestamp': unified_timestamp(document.get('date')) or unified_timestamp( |
332 | try_get(video, lambda x: x['meta']['editorialDate'], compat_str)), | |
fb47cb5b S |
333 | 'thumbnails': thumbnails, |
334 | 'subtitles': self._extract_subtitles(document), | |
335 | 'formats': formats, | |
336 | } | |
b6de53ea | 337 | |
fb47cb5b S |
338 | def _real_extract(self, url): |
339 | video_id = self._match_id(url) | |
b6de53ea | 340 | |
fb47cb5b S |
341 | webpage = self._download_webpage(url, video_id, fatal=False) |
342 | if webpage: | |
343 | player = self._extract_player(webpage, url, fatal=False) | |
344 | if player: | |
345 | return self._extract_regular(url, player, video_id) | |
b6de53ea | 346 | |
fb47cb5b | 347 | return self._extract_mobile(video_id) |
b6de53ea | 348 | |
b6de53ea | 349 | |
fb47cb5b S |
350 | class ZDFChannelIE(ZDFBaseIE): |
351 | _VALID_URL = r'https?://www\.zdf\.de/(?:[^/]+/)*(?P<id>[^/?#&]+)' | |
c2404463 | 352 | _TESTS = [{ |
fb47cb5b | 353 | 'url': 'https://www.zdf.de/sport/das-aktuelle-sportstudio', |
8560c618 | 354 | 'info_dict': { |
fb47cb5b S |
355 | 'id': 'das-aktuelle-sportstudio', |
356 | 'title': 'das aktuelle sportstudio | ZDF', | |
8560c618 | 357 | }, |
b4cbdbd4 | 358 | 'playlist_mincount': 23, |
c2404463 | 359 | }, { |
fb47cb5b S |
360 | 'url': 'https://www.zdf.de/dokumentation/planet-e', |
361 | 'info_dict': { | |
362 | 'id': 'planet-e', | |
363 | 'title': 'planet e.', | |
364 | }, | |
b4cbdbd4 | 365 | 'playlist_mincount': 50, |
c2404463 | 366 | }, { |
fb47cb5b | 367 | 'url': 'https://www.zdf.de/filme/taunuskrimi/', |
c2404463 S |
368 | 'only_matching': True, |
369 | }] | |
fb47cb5b S |
370 | |
371 | @classmethod | |
372 | def suitable(cls, url): | |
373 | return False if ZDFIE.suitable(url) else super(ZDFChannelIE, cls).suitable(url) | |
9abd500a PH |
374 | |
375 | def _real_extract(self, url): | |
376 | channel_id = self._match_id(url) | |
8560c618 | 377 | |
fb47cb5b S |
378 | webpage = self._download_webpage(url, channel_id) |
379 | ||
380 | entries = [ | |
381 | self.url_result(item_url, ie=ZDFIE.ie_key()) | |
382 | for item_url in orderedSet(re.findall( | |
383 | r'data-plusbar-url=["\'](http.+?\.html)', webpage))] | |
384 | ||
385 | return self.playlist_result( | |
386 | entries, channel_id, self._og_search_title(webpage, fatal=False)) | |
387 | ||
ec85ded8 | 388 | r""" |
fb47cb5b S |
389 | player = self._extract_player(webpage, channel_id) |
390 | ||
391 | channel_id = self._search_regex( | |
392 | r'docId\s*:\s*(["\'])(?P<id>(?!\1).+?)\1', webpage, | |
393 | 'channel id', group='id') | |
394 | ||
395 | channel = self._call_api( | |
396 | 'https://api.zdf.de/content/documents/%s.json' % channel_id, | |
397 | player, url, channel_id) | |
398 | ||
399 | items = [] | |
400 | for module in channel['module']: | |
401 | for teaser in try_get(module, lambda x: x['teaser'], list) or []: | |
402 | t = try_get( | |
403 | teaser, lambda x: x['http://zdf.de/rels/target'], dict) | |
404 | if not t: | |
405 | continue | |
406 | items.extend(try_get( | |
407 | t, | |
408 | lambda x: x['resultsWithVideo']['http://zdf.de/rels/search/results'], | |
409 | list) or []) | |
410 | items.extend(try_get( | |
411 | module, | |
412 | lambda x: x['filterRef']['resultsWithVideo']['http://zdf.de/rels/search/results'], | |
413 | list) or []) | |
414 | ||
415 | entries = [] | |
416 | entry_urls = set() | |
417 | for item in items: | |
418 | t = try_get(item, lambda x: x['http://zdf.de/rels/target'], dict) | |
419 | if not t: | |
420 | continue | |
421 | sharing_url = t.get('http://zdf.de/rels/sharing-url') | |
422 | if not sharing_url or not isinstance(sharing_url, compat_str): | |
423 | continue | |
424 | if sharing_url in entry_urls: | |
425 | continue | |
426 | entry_urls.add(sharing_url) | |
427 | entries.append(self.url_result( | |
428 | sharing_url, ie=ZDFIE.ie_key(), video_id=t.get('id'))) | |
429 | ||
430 | return self.playlist_result(entries, channel_id, channel.get('title')) | |
431 | """ |