2 from __future__ import unicode_literals
6 from .common import InfoExtractor
10 compat_urllib_parse_unquote,
11 compat_urllib_parse_urlparse,
25 from .periscope import (
31 class TwitterBaseIE(InfoExtractor):
32 _API_BASE = 'https://api.twitter.com/1.1/'
33 _BASE_REGEX = r'https?://(?:(?:www|m(?:obile)?)\.)?twitter\.com/'
36 def _extract_variant_formats(self, variant, video_id):
37 variant_url = variant.get('url')
40 elif '.m3u8' in variant_url:
41 return self._extract_m3u8_formats(
42 variant_url, video_id, 'mp4', 'm3u8_native',
43 m3u8_id='hls', fatal=False)
45 tbr = int_or_none(dict_get(variant, ('bitrate', 'bit_rate')), 1000) or None
48 'format_id': 'http' + ('-%d' % tbr if tbr else ''),
51 self._search_dimensions_in_video_url(f, variant_url)
54 def _extract_formats_from_vmap_url(self, vmap_url, video_id):
55 vmap_data = self._download_xml(vmap_url, video_id)
58 for video_variant in vmap_data.findall('.//{http://twitter.com/schema/videoVMapV2.xsd}videoVariant'):
59 video_variant.attrib['url'] = compat_urllib_parse_unquote(
60 video_variant.attrib['url'])
61 urls.append(video_variant.attrib['url'])
62 formats.extend(self._extract_variant_formats(
63 video_variant.attrib, video_id))
64 video_url = strip_or_none(xpath_text(vmap_data, './/MediaFile'))
65 if video_url not in urls:
66 formats.extend(self._extract_variant_formats({'url': video_url}, video_id))
70 def _search_dimensions_in_video_url(a_format, video_url):
71 m = re.search(r'/(?P<width>\d+)x(?P<height>\d+)/', video_url)
74 'width': int(m.group('width')),
75 'height': int(m.group('height')),
78 def _call_api(self, path, video_id, query={}):
80 'Authorization': 'Bearer AAAAAAAAAAAAAAAAAAAAAPYXBAAAAAAACLXUNDekMxqa8h%2F40K4moUkGsoc%3DTYfbDKbT3jJPCEVnMYqilB28NHfOPqkca3qaAxGfsyKCs0wRbw',
82 if not self._GUEST_TOKEN:
83 self._GUEST_TOKEN = self._download_json(
84 self._API_BASE + 'guest/activate.json', video_id,
85 'Downloading guest token', data=b'',
86 headers=headers)['guest_token']
87 headers['x-guest-token'] = self._GUEST_TOKEN
89 return self._download_json(
90 self._API_BASE + path, video_id, headers=headers, query=query)
91 except ExtractorError as e:
92 if isinstance(e.cause, compat_HTTPError) and e.cause.code == 403:
93 raise ExtractorError(self._parse_json(
94 e.cause.read().decode(),
95 video_id)['errors'][0]['message'], expected=True)
99 class TwitterCardIE(InfoExtractor):
100 IE_NAME = 'twitter:card'
101 _VALID_URL = TwitterBaseIE._BASE_REGEX + r'i/(?:cards/tfw/v1|videos(?:/tweet)?)/(?P<id>\d+)'
104 'url': 'https://twitter.com/i/cards/tfw/v1/560070183650213889',
105 # MD5 checksums are different in different places
107 'id': '560070183650213889',
109 'title': "Twitter - You can now shoot, edit and share video on Twitter. Capture life's most moving moments from your perspective.",
110 'description': 'md5:18d3e24bb4f6e5007487dd546e53bd96',
111 'uploader': 'Twitter',
112 'uploader_id': 'Twitter',
113 'thumbnail': r're:^https?://.*\.jpg',
115 'timestamp': 1422366112,
116 'upload_date': '20150127',
120 'url': 'https://twitter.com/i/cards/tfw/v1/623160978427936768',
121 'md5': '7137eca597f72b9abbe61e5ae0161399',
123 'id': '623160978427936768',
125 'title': "NASA - Fly over Pluto's icy Norgay Mountains and Sputnik Plain in this @NASANewHorizons #PlutoFlyby video.",
126 'description': "Fly over Pluto's icy Norgay Mountains and Sputnik Plain in this @NASANewHorizons #PlutoFlyby video. https://t.co/BJYgOjSeGA",
128 'uploader_id': 'NASA',
129 'timestamp': 1437408129,
130 'upload_date': '20150720',
134 'url': 'https://twitter.com/i/cards/tfw/v1/654001591733886977',
135 'md5': 'b6d9683dd3f48e340ded81c0e917ad46',
139 'title': 'Ubuntu 11.10 Overview',
140 'description': 'md5:a831e97fa384863d6e26ce48d1c43376',
141 'upload_date': '20111013',
142 'uploader': 'OMG! UBUNTU!',
143 'uploader_id': 'omgubuntu',
145 'add_ie': ['Youtube'],
148 'url': 'https://twitter.com/i/cards/tfw/v1/665289828897005568',
149 'md5': '6dabeaca9e68cbb71c99c322a4b42a11',
153 'upload_date': '20151113',
154 'uploader_id': '1189339351084113920',
155 'uploader': 'ArsenalTerje',
156 'title': 'Vine by ArsenalTerje',
157 'timestamp': 1447451307,
161 'url': 'https://twitter.com/i/videos/tweet/705235433198714880',
162 'md5': '884812a2adc8aaf6fe52b15ccbfa3b88',
164 'id': '705235433198714880',
166 'title': "Brent Yarina - Khalil Iverson's missed highlight dunk. And made highlight dunk. In one highlight.",
167 'description': "Khalil Iverson's missed highlight dunk. And made highlight dunk. In one highlight. https://t.co/OrxcJ28Bns",
168 'uploader': 'Brent Yarina',
169 'uploader_id': 'BTNBrentYarina',
170 'timestamp': 1456976204,
171 'upload_date': '20160303',
173 'skip': 'This content is no longer available.',
175 'url': 'https://twitter.com/i/videos/752274308186120192',
176 'only_matching': True,
180 def _real_extract(self, url):
181 status_id = self._match_id(url)
182 return self.url_result(
183 'https://twitter.com/statuses/' + status_id,
184 TwitterIE.ie_key(), status_id)
187 class TwitterIE(TwitterBaseIE):
189 _VALID_URL = TwitterBaseIE._BASE_REGEX + r'(?:(?:i/web|[^/]+)/status|statuses)/(?P<id>\d+)'
192 'url': 'https://twitter.com/freethenipple/status/643211948184596480',
194 'id': '643211948184596480',
196 'title': 'FREE THE NIPPLE - FTN supporters on Hollywood Blvd today!',
197 'thumbnail': r're:^https?://.*\.jpg',
198 'description': 'FTN supporters on Hollywood Blvd today! http://t.co/c7jHH749xJ',
199 'uploader': 'FREE THE NIPPLE',
200 'uploader_id': 'freethenipple',
202 'timestamp': 1442188653,
203 'upload_date': '20150913',
207 'url': 'https://twitter.com/giphz/status/657991469417025536/photo/1',
208 'md5': 'f36dcd5fb92bf7057f155e7d927eeb42',
210 'id': '657991469417025536',
212 'title': 'Gifs - tu vai cai tu vai cai tu nao eh capaz disso tu vai cai',
213 'description': 'Gifs on Twitter: "tu vai cai tu vai cai tu nao eh capaz disso tu vai cai https://t.co/tM46VHFlO5"',
214 'thumbnail': r're:^https?://.*\.png',
216 'uploader_id': 'giphz',
218 'expected_warnings': ['height', 'width'],
219 'skip': 'Account suspended',
221 'url': 'https://twitter.com/starwars/status/665052190608723968',
223 'id': '665052190608723968',
225 'title': 'Star Wars - A new beginning is coming December 18. Watch the official 60 second #TV spot for #StarWars: #TheForceAwakens.',
226 'description': 'A new beginning is coming December 18. Watch the official 60 second #TV spot for #StarWars: #TheForceAwakens. https://t.co/OkSqT2fjWJ',
227 'uploader_id': 'starwars',
228 'uploader': 'Star Wars',
229 'timestamp': 1447395772,
230 'upload_date': '20151113',
233 'url': 'https://twitter.com/BTNBrentYarina/status/705235433198714880',
235 'id': '705235433198714880',
237 'title': "Brent Yarina - Khalil Iverson's missed highlight dunk. And made highlight dunk. In one highlight.",
238 'description': "Khalil Iverson's missed highlight dunk. And made highlight dunk. In one highlight. https://t.co/OrxcJ28Bns",
239 'uploader_id': 'BTNBrentYarina',
240 'uploader': 'Brent Yarina',
241 'timestamp': 1456976204,
242 'upload_date': '20160303',
245 # The same video as https://twitter.com/i/videos/tweet/705235433198714880
246 # Test case of TwitterCardIE
247 'skip_download': True,
250 'url': 'https://twitter.com/jaydingeer/status/700207533655363584',
252 'id': '700207533655363584',
254 'title': 'simon vetugo - BEAT PROD: @suhmeduh #Damndaniel',
255 'description': 'BEAT PROD: @suhmeduh https://t.co/HBrQ4AfpvZ #Damndaniel https://t.co/byBooq2ejZ',
256 'thumbnail': r're:^https?://.*\.jpg',
257 'uploader': 'simon vetugo',
258 'uploader_id': 'simonvertugo',
260 'timestamp': 1455777459,
261 'upload_date': '20160218',
264 'url': 'https://twitter.com/Filmdrunk/status/713801302971588609',
265 'md5': '89a15ed345d13b86e9a5a5e051fa308a',
269 'title': 'Dr.Pepperの飲み方 #japanese #バカ #ドクペ #電動ガン',
270 'uploader': 'TAKUMA',
271 'uploader_id': '1004126642786242560',
272 'timestamp': 1402826626,
273 'upload_date': '20140615',
277 'url': 'https://twitter.com/captainamerica/status/719944021058060289',
279 'id': '719944021058060289',
281 'title': 'Captain America - @King0fNerd Are you sure you made the right choice? Find out in theaters.',
282 'description': '@King0fNerd Are you sure you made the right choice? Find out in theaters. https://t.co/GpgYi9xMJI',
283 'uploader_id': 'CaptainAmerica',
284 'uploader': 'Captain America',
286 'timestamp': 1460483005,
287 'upload_date': '20160412',
290 'url': 'https://twitter.com/OPP_HSD/status/779210622571536384',
292 'id': '1zqKVVlkqLaKB',
294 'title': 'Sgt Kerry Schmidt - Ontario Provincial Police - Road rage, mischief, assault, rollover and fire in one occurrence',
295 'upload_date': '20160923',
296 'uploader_id': '1PmKqpJdOJQoY',
297 'uploader': 'Sgt Kerry Schmidt - Ontario Provincial Police',
298 'timestamp': 1474613214,
300 'add_ie': ['Periscope'],
302 # has mp4 formats via mobile API
303 'url': 'https://twitter.com/news_al3alm/status/852138619213144067',
305 'id': '852138619213144067',
307 'title': 'عالم الأخبار - كلمة تاريخية بجلسة الجناسي التاريخية.. النائب خالد مؤنس العتيبي للمعارضين : اتقوا الله .. الظلم ظلمات يوم القيامة',
308 'description': 'كلمة تاريخية بجلسة الجناسي التاريخية.. النائب خالد مؤنس العتيبي للمعارضين : اتقوا الله .. الظلم ظلمات يوم القيامة https://t.co/xg6OhpyKfN',
309 'uploader': 'عالم الأخبار',
310 'uploader_id': 'news_al3alm',
312 'timestamp': 1492000653,
313 'upload_date': '20170412',
316 'url': 'https://twitter.com/i/web/status/910031516746514432',
318 'id': '910031516746514432',
320 'title': 'Préfet de Guadeloupe - [Direct] #Maria Le centre se trouve actuellement au sud de Basse-Terre. Restez confinés. Réfugiez-vous dans la pièce la + sûre.',
321 'thumbnail': r're:^https?://.*\.jpg',
322 'description': '[Direct] #Maria Le centre se trouve actuellement au sud de Basse-Terre. Restez confinés. Réfugiez-vous dans la pièce la + sûre. https://t.co/mwx01Rs4lo',
323 'uploader': 'Préfet de Guadeloupe',
324 'uploader_id': 'Prefet971',
326 'timestamp': 1505803395,
327 'upload_date': '20170919',
330 'skip_download': True, # requires ffmpeg
333 # card via api.twitter.com/1.1/videos/tweet/config
334 'url': 'https://twitter.com/LisPower1/status/1001551623938805763',
336 'id': '1001551623938805763',
338 'title': 're:.*?Shep is on a roll today.*?',
339 'thumbnail': r're:^https?://.*\.jpg',
340 'description': 'md5:37b9f2ff31720cef23b2bd42ee8a0f09',
341 'uploader': 'Lis Power',
342 'uploader_id': 'LisPower1',
344 'timestamp': 1527623489,
345 'upload_date': '20180529',
348 'skip_download': True, # requires ffmpeg
351 'url': 'https://twitter.com/foobar/status/1087791357756956680',
353 'id': '1087791357756956680',
355 'title': 'Twitter - A new is coming. Some of you got an opt-in to try it now. Check out the emoji button, quick keyboard shortcuts, upgraded trends, advanced search, and more. Let us know your thoughts!',
356 'thumbnail': r're:^https?://.*\.jpg',
357 'description': 'md5:6dfd341a3310fb97d80d2bf7145df976',
358 'uploader': 'Twitter',
359 'uploader_id': 'Twitter',
361 'timestamp': 1548184644,
362 'upload_date': '20190122',
365 # not available in Periscope
366 'url': 'https://twitter.com/ViviEducation/status/1136534865145286656',
368 'id': '1vOGwqejwoWxB',
370 'title': 'Vivi - Vivi founder @lior_rauchy announcing our new student feedback tool live at @EduTECH_AU #EduTECH2019',
372 'uploader_id': '1eVjYOLGkGrQL',
374 'add_ie': ['TwitterBroadcast'],
377 'url': 'https://twitter.com/GunB1g/status/1163218564784017422',
378 'only_matching': True,
380 # promo_video_website card
381 'url': 'https://twitter.com/GunB1g/status/1163218564784017422',
382 'only_matching': True,
385 def _real_extract(self, url):
386 twid = self._match_id(url)
387 status = self._call_api(
388 'statuses/show/%s.json' % twid, twid, {
389 'cards_platform': 'Web-12',
391 'include_reply_count': 1,
392 'include_user_entities': 0,
393 'tweet_mode': 'extended',
396 title = description = status['full_text'].replace('\n', ' ')
397 # strip 'https -_t.co_BJYgOjSeGA' junk from filenames
398 title = re.sub(r'\s+(https?://[^ ]+)', '', title)
399 user = status.get('user') or {}
400 uploader = user.get('name')
402 title = '%s - %s' % (uploader, title)
403 uploader_id = user.get('screen_name')
406 for hashtag in (try_get(status, lambda x: x['entities']['hashtags'], list) or []):
407 hashtag_text = hashtag.get('text')
410 tags.append(hashtag_text)
415 'description': description,
416 'uploader': uploader,
417 'timestamp': unified_timestamp(status.get('created_at')),
418 'uploader_id': uploader_id,
419 'uploader_url': 'https://twitter.com/' + uploader_id if uploader_id else None,
420 'like_count': int_or_none(status.get('favorite_count')),
421 'repost_count': int_or_none(status.get('retweet_count')),
422 'comment_count': int_or_none(status.get('reply_count')),
423 'age_limit': 18 if status.get('possibly_sensitive') else 0,
427 media = try_get(status, lambda x: x['extended_entities']['media'][0])
428 if media and media.get('type') != 'photo':
429 video_info = media.get('video_info') or {}
432 for variant in video_info.get('variants', []):
433 formats.extend(self._extract_variant_formats(variant, twid))
434 self._sort_formats(formats)
437 media_url = media.get('media_url_https') or media.get('media_url')
439 def add_thumbnail(name, size):
442 'url': update_url_query(media_url, {'name': name}),
443 'width': int_or_none(size.get('w') or size.get('width')),
444 'height': int_or_none(size.get('h') or size.get('height')),
446 for name, size in media.get('sizes', {}).items():
447 add_thumbnail(name, size)
448 add_thumbnail('orig', media.get('original_info') or {})
452 'thumbnails': thumbnails,
453 'duration': float_or_none(video_info.get('duration_millis'), 1000),
456 card = status.get('card')
458 binding_values = card['binding_values']
460 def get_binding_value(k):
461 o = binding_values.get(k) or {}
462 return try_get(o, lambda x: x[x['type'].lower() + '_value'])
464 card_name = card['name'].split(':')[-1]
465 if card_name in ('amplify', 'promo_video_website'):
466 is_amplify = card_name == 'amplify'
467 vmap_url = get_binding_value('amplify_url_vmap') if is_amplify else get_binding_value('player_stream_url')
468 content_id = get_binding_value('%s_content_id' % (card_name if is_amplify else 'player'))
469 formats = self._extract_formats_from_vmap_url(vmap_url, content_id or twid)
470 self._sort_formats(formats)
473 for suffix in ('_small', '', '_large', '_x_large', '_original'):
474 image = get_binding_value('player_image' + suffix) or {}
475 image_url = image.get('url')
476 if not image_url or '/player-placeholder' in image_url:
479 'id': suffix[1:] if suffix else 'medium',
481 'width': int_or_none(image.get('width')),
482 'height': int_or_none(image.get('height')),
487 'thumbnails': thumbnails,
488 'duration': int_or_none(get_binding_value(
489 'content_duration_seconds')),
491 elif card_name == 'player':
494 'url': get_binding_value('player_url'),
496 elif card_name == 'periscope_broadcast':
499 'url': get_binding_value('url') or get_binding_value('player_url'),
500 'ie_key': PeriscopeIE.ie_key(),
502 elif card_name == 'broadcast':
505 'url': get_binding_value('broadcast_url'),
506 'ie_key': TwitterBroadcastIE.ie_key(),
509 raise ExtractorError('Unsupported Twitter Card.')
511 expanded_url = try_get(status, lambda x: x['entities']['urls'][0]['expanded_url'])
513 raise ExtractorError("There's no video in this tweet.")
521 class TwitterAmplifyIE(TwitterBaseIE):
522 IE_NAME = 'twitter:amplify'
523 _VALID_URL = r'https?://amp\.twimg\.com/v/(?P<id>[0-9a-f\-]{36})'
526 'url': 'https://amp.twimg.com/v/0ba0c3c7-0af3-4c0a-bed5-7efd1ffa2951',
527 'md5': '7df102d0b9fd7066b86f3159f8e81bf6',
529 'id': '0ba0c3c7-0af3-4c0a-bed5-7efd1ffa2951',
531 'title': 'Twitter Video',
532 'thumbnail': 're:^https?://.*',
536 def _real_extract(self, url):
537 video_id = self._match_id(url)
538 webpage = self._download_webpage(url, video_id)
540 vmap_url = self._html_search_meta(
541 'twitter:amplify:vmap', webpage, 'vmap url')
542 formats = self._extract_formats_from_vmap_url(vmap_url, video_id)
545 thumbnail = self._html_search_meta(
546 'twitter:image:src', webpage, 'thumbnail', fatal=False)
548 def _find_dimension(target):
549 w = int_or_none(self._html_search_meta(
550 'twitter:%s:width' % target, webpage, fatal=False))
551 h = int_or_none(self._html_search_meta(
552 'twitter:%s:height' % target, webpage, fatal=False))
556 thumbnail_w, thumbnail_h = _find_dimension('image')
559 'width': thumbnail_w,
560 'height': thumbnail_h,
563 video_w, video_h = _find_dimension('player')
571 'title': 'Twitter Video',
573 'thumbnails': thumbnails,
577 class TwitterBroadcastIE(TwitterBaseIE, PeriscopeBaseIE):
578 IE_NAME = 'twitter:broadcast'
579 _VALID_URL = TwitterBaseIE._BASE_REGEX + r'i/broadcasts/(?P<id>[0-9a-zA-Z]{13})'
581 def _real_extract(self, url):
582 broadcast_id = self._match_id(url)
583 broadcast = self._call_api(
584 'broadcasts/show.json', broadcast_id,
585 {'ids': broadcast_id})['broadcasts'][broadcast_id]
586 info = self._parse_broadcast_data(broadcast, broadcast_id)
587 media_key = broadcast['media_key']
588 source = self._call_api(
589 'live_video_stream/status/' + media_key, media_key)['source']
590 m3u8_url = source.get('noRedirectPlaybackUrl') or source['location']
591 if '/live_video_stream/geoblocked/' in m3u8_url:
592 self.raise_geo_restricted()
593 m3u8_id = compat_parse_qs(compat_urllib_parse_urlparse(
594 m3u8_url).query).get('type', [None])[0]
595 state, width, height = self._extract_common_format_info(broadcast)
596 info['formats'] = self._extract_pscp_m3u8_formats(
597 m3u8_url, broadcast_id, m3u8_id, state, width, height)