2 from __future__ import unicode_literals
6 from .common import InfoExtractor
7 from ..compat import compat_urlparse
19 from .periscope import PeriscopeIE
22 class TwitterBaseIE(InfoExtractor):
23 def _extract_formats_from_vmap_url(self, vmap_url, video_id):
24 vmap_data = self._download_xml(vmap_url, video_id)
25 video_url = xpath_text(vmap_data, './/MediaFile').strip()
26 if determine_ext(video_url) == 'm3u8':
27 return self._extract_m3u8_formats(
28 video_url, video_id, ext='mp4', m3u8_id='hls',
29 entry_protocol='m3u8_native')
35 def _search_dimensions_in_video_url(a_format, video_url):
36 m = re.search(r'/(?P<width>\d+)x(?P<height>\d+)/', video_url)
39 'width': int(m.group('width')),
40 'height': int(m.group('height')),
44 class TwitterCardIE(TwitterBaseIE):
45 IE_NAME = 'twitter:card'
46 _VALID_URL = r'https?://(?:www\.)?twitter\.com/i/(?:cards/tfw/v1|videos(?:/tweet)?)/(?P<id>\d+)'
49 'url': 'https://twitter.com/i/cards/tfw/v1/560070183650213889',
50 # MD5 checksums are different in different places
52 'id': '560070183650213889',
54 'title': 'Twitter Card',
55 'thumbnail': r're:^https?://.*\.jpg$',
60 'url': 'https://twitter.com/i/cards/tfw/v1/623160978427936768',
61 'md5': '7ee2a553b63d1bccba97fbed97d9e1c8',
63 'id': '623160978427936768',
65 'title': 'Twitter Card',
66 'thumbnail': r're:^https?://.*\.jpg',
71 'url': 'https://twitter.com/i/cards/tfw/v1/654001591733886977',
72 'md5': 'b6d9683dd3f48e340ded81c0e917ad46',
76 'title': 'Ubuntu 11.10 Overview',
77 'description': 'md5:a831e97fa384863d6e26ce48d1c43376',
78 'upload_date': '20111013',
79 'uploader': 'OMG! Ubuntu!',
80 'uploader_id': 'omgubuntu',
82 'add_ie': ['Youtube'],
85 'url': 'https://twitter.com/i/cards/tfw/v1/665289828897005568',
86 'md5': 'ab2745d0b0ce53319a534fccaa986439',
90 'upload_date': '20151113',
91 'uploader_id': '1189339351084113920',
92 'uploader': 'ArsenalTerje',
93 'title': 'Vine by ArsenalTerje',
94 'timestamp': 1447451307,
98 'url': 'https://twitter.com/i/videos/tweet/705235433198714880',
99 'md5': '3846d0a07109b5ab622425449b59049d',
101 'id': '705235433198714880',
103 'title': 'Twitter web player',
104 'thumbnail': r're:^https?://.*\.jpg',
107 'url': 'https://twitter.com/i/videos/752274308186120192',
108 'only_matching': True,
112 def _parse_media_info(self, media_info, video_id):
114 for media_variant in media_info.get('variants', []):
115 media_url = media_variant['url']
116 if media_url.endswith('.m3u8'):
117 formats.extend(self._extract_m3u8_formats(media_url, video_id, ext='mp4', m3u8_id='hls'))
118 elif media_url.endswith('.mpd'):
119 formats.extend(self._extract_mpd_formats(media_url, video_id, mpd_id='dash'))
121 vbr = int_or_none(dict_get(media_variant, ('bitRate', 'bitrate')), scale=1000)
124 'format_id': 'http-%d' % vbr if vbr else 'http',
127 # Reported bitRate may be zero
128 if not a_format['vbr']:
131 self._search_dimensions_in_video_url(a_format, media_url)
133 formats.append(a_format)
136 def _extract_mobile_formats(self, username, video_id):
137 webpage = self._download_webpage(
138 'https://mobile.twitter.com/%s/status/%s' % (username, video_id),
139 video_id, 'Downloading mobile webpage',
141 # A recent mobile UA is necessary for `gt` cookie
142 'User-Agent': 'Mozilla/5.0 (Android 6.0.1; Mobile; rv:54.0) Gecko/54.0 Firefox/54.0',
144 main_script_url = self._html_search_regex(
145 r'<script[^>]+src="([^"]+main\.[^"]+)"', webpage, 'main script URL')
146 main_script = self._download_webpage(
147 main_script_url, video_id, 'Downloading main script')
148 bearer_token = self._search_regex(
149 r'BEARER_TOKEN\s*:\s*"([^"]+)"',
150 main_script, 'bearer token')
151 guest_token = self._search_regex(
152 r'document\.cookie\s*=\s*decodeURIComponent\("gt=(\d+)',
153 webpage, 'guest token')
154 api_data = self._download_json(
155 'https://api.twitter.com/2/timeline/conversation/%s.json' % video_id,
156 video_id, 'Downloading mobile API data',
158 'Authorization': 'Bearer ' + bearer_token,
159 'x-guest-token': guest_token,
161 media_info = try_get(api_data, lambda o: o['globalObjects']['tweets'][video_id]
162 ['extended_entities']['media'][0]['video_info']) or {}
163 return self._parse_media_info(media_info, video_id)
165 def _real_extract(self, url):
166 video_id = self._match_id(url)
172 webpage = self._download_webpage(url, video_id)
174 iframe_url = self._html_search_regex(
175 r'<iframe[^>]+src="((?:https?:)?//(?:www.youtube.com/embed/[^"]+|(?:www\.)?vine\.co/v/\w+/card))"',
176 webpage, 'video iframe', default=None)
178 return self.url_result(iframe_url)
180 config = self._parse_json(self._html_search_regex(
181 r'data-(?:player-)?config="([^"]+)"', webpage,
182 'data player config', default='{}'),
185 if config.get('source_type') == 'vine':
186 return self.url_result(config['player_url'], 'Vine')
188 periscope_url = PeriscopeIE._extract_url(webpage)
190 return self.url_result(periscope_url, PeriscopeIE.ie_key())
192 video_url = config.get('video_url') or config.get('playlist', [{}])[0].get('source')
195 if determine_ext(video_url) == 'm3u8':
196 formats.extend(self._extract_m3u8_formats(video_url, video_id, ext='mp4', m3u8_id='hls'))
202 self._search_dimensions_in_video_url(f, video_url)
206 vmap_url = config.get('vmapUrl') or config.get('vmap_url')
209 self._extract_formats_from_vmap_url(vmap_url, video_id))
213 for entity in config.get('status', {}).get('entities', []):
214 if 'mediaInfo' in entity:
215 media_info = entity['mediaInfo']
218 formats.extend(self._parse_media_info(media_info, video_id))
219 duration = float_or_none(media_info.get('duration', {}).get('nanos'), scale=1e9)
221 username = config.get('user', {}).get('screen_name')
223 formats.extend(self._extract_mobile_formats(username, video_id))
225 self._remove_duplicate_formats(formats)
226 self._sort_formats(formats)
228 title = self._search_regex(r'<title>([^<]+)</title>', webpage, 'title')
229 thumbnail = config.get('posterImageUrl') or config.get('image_src')
230 duration = float_or_none(config.get('duration')) or duration
235 'thumbnail': thumbnail,
236 'duration': duration,
241 class TwitterIE(InfoExtractor):
243 _VALID_URL = r'https?://(?:www\.|m\.|mobile\.)?twitter\.com/(?P<user_id>[^/]+)/status/(?P<id>\d+)'
244 _TEMPLATE_URL = 'https://twitter.com/%s/status/%s'
247 'url': 'https://twitter.com/freethenipple/status/643211948184596480',
249 'id': '643211948184596480',
251 'title': 'FREE THE NIPPLE - FTN supporters on Hollywood Blvd today!',
252 'thumbnail': r're:^https?://.*\.jpg',
253 'description': 'FREE THE NIPPLE on Twitter: "FTN supporters on Hollywood Blvd today! http://t.co/c7jHH749xJ"',
254 'uploader': 'FREE THE NIPPLE',
255 'uploader_id': 'freethenipple',
258 'skip_download': True, # requires ffmpeg
261 'url': 'https://twitter.com/giphz/status/657991469417025536/photo/1',
262 'md5': 'f36dcd5fb92bf7057f155e7d927eeb42',
264 'id': '657991469417025536',
266 'title': 'Gifs - tu vai cai tu vai cai tu nao eh capaz disso tu vai cai',
267 'description': 'Gifs on Twitter: "tu vai cai tu vai cai tu nao eh capaz disso tu vai cai https://t.co/tM46VHFlO5"',
268 'thumbnail': r're:^https?://.*\.png',
270 'uploader_id': 'giphz',
272 'expected_warnings': ['height', 'width'],
273 'skip': 'Account suspended',
275 'url': 'https://twitter.com/starwars/status/665052190608723968',
276 'md5': '39b7199856dee6cd4432e72c74bc69d4',
278 'id': '665052190608723968',
280 'title': 'Star Wars - A new beginning is coming December 18. Watch the official 60 second #TV spot for #StarWars: #TheForceAwakens.',
281 'description': 'Star Wars on Twitter: "A new beginning is coming December 18. Watch the official 60 second #TV spot for #StarWars: #TheForceAwakens."',
282 'uploader_id': 'starwars',
283 'uploader': 'Star Wars',
286 'url': 'https://twitter.com/BTNBrentYarina/status/705235433198714880',
288 'id': '705235433198714880',
290 'title': 'Brent Yarina - Khalil Iverson\'s missed highlight dunk. And made highlight dunk. In one highlight.',
291 'description': 'Brent Yarina on Twitter: "Khalil Iverson\'s missed highlight dunk. And made highlight dunk. In one highlight."',
292 'uploader_id': 'BTNBrentYarina',
293 'uploader': 'Brent Yarina',
296 # The same video as https://twitter.com/i/videos/tweet/705235433198714880
297 # Test case of TwitterCardIE
298 'skip_download': True,
301 'url': 'https://twitter.com/jaydingeer/status/700207533655363584',
304 'id': '700207533655363584',
306 'title': 'JG - BEAT PROD: @suhmeduh #Damndaniel',
307 'description': 'JG on Twitter: "BEAT PROD: @suhmeduh https://t.co/HBrQ4AfpvZ #Damndaniel https://t.co/byBooq2ejZ"',
308 'thumbnail': r're:^https?://.*\.jpg',
310 'uploader_id': 'jaydingeer',
313 'skip_download': True, # requires ffmpeg
316 'url': 'https://twitter.com/Filmdrunk/status/713801302971588609',
317 'md5': '89a15ed345d13b86e9a5a5e051fa308a',
321 'title': 'Dr.Pepperの飲み方 #japanese #バカ #ドクペ #電動ガン',
322 'uploader': 'TAKUMA',
323 'uploader_id': '1004126642786242560',
324 'upload_date': '20140615',
328 'url': 'https://twitter.com/captainamerica/status/719944021058060289',
330 'id': '719944021058060289',
332 'title': 'Captain America - @King0fNerd Are you sure you made the right choice? Find out in theaters.',
333 'description': 'Captain America on Twitter: "@King0fNerd Are you sure you made the right choice? Find out in theaters. https://t.co/GpgYi9xMJI"',
334 'uploader_id': 'captainamerica',
335 'uploader': 'Captain America',
338 'skip_download': True, # requires ffmpeg
341 'url': 'https://twitter.com/OPP_HSD/status/779210622571536384',
343 'id': '1zqKVVlkqLaKB',
345 'title': 'Sgt Kerry Schmidt - LIVE on #Periscope: Road rage, mischief, assault, rollover and fire in one occurrence',
346 'description': 'Sgt Kerry Schmidt on Twitter: "LIVE on #Periscope: Road rage, mischief, assault, rollover and fire in one occurrence https://t.co/EKrVgIXF3s"',
347 'upload_date': '20160923',
348 'uploader_id': 'OPP_HSD',
349 'uploader': 'Sgt Kerry Schmidt',
350 'timestamp': 1474613214,
352 'add_ie': ['Periscope'],
354 # has mp4 formats via mobile API
355 'url': 'https://twitter.com/news_al3alm/status/852138619213144067',
357 'id': '852138619213144067',
359 'title': 'عالم الأخبار - كلمة تاريخية بجلسة الجناسي التاريخية.. النائب خالد مؤنس العتيبي للمعارضين : اتقوا الله .. الظلم ظلمات يوم القيامة',
360 'description': 'عالم الأخبار on Twitter: "كلمة تاريخية بجلسة الجناسي التاريخية.. النائب خالد مؤنس العتيبي للمعارضين : اتقوا الله .. الظلم ظلمات يوم القيامة https://t.co/xg6OhpyKfN"',
361 'uploader': 'عالم الأخبار',
362 'uploader_id': 'news_al3alm',
365 'format': 'best[format_id^=http-]',
369 def _real_extract(self, url):
370 mobj = re.match(self._VALID_URL, url)
371 user_id = mobj.group('user_id')
372 twid = mobj.group('id')
374 webpage, urlh = self._download_webpage_handle(
375 self._TEMPLATE_URL % (user_id, twid), twid)
377 if 'twitter.com/account/suspended' in urlh.geturl():
378 raise ExtractorError('Account suspended by Twitter.', expected=True)
380 username = remove_end(self._og_search_title(webpage), ' on Twitter')
382 title = description = self._og_search_description(webpage).strip('').replace('\n', ' ').strip('“”')
384 # strip 'https -_t.co_BJYgOjSeGA' junk from filenames
385 title = re.sub(r'\s+(https?://[^ ]+)', '', title)
388 'uploader_id': user_id,
389 'uploader': username,
391 'description': '%s on Twitter: "%s"' % (username, description),
392 'title': username + ' - ' + title,
395 mobj = re.search(r'''(?x)
396 <video[^>]+class="animated-gif"(?P<more_info>[^>]+)>\s*
397 <source[^>]+video-src="(?P<url>[^"]+)"
401 more_info = mobj.group('more_info')
402 height = int_or_none(self._search_regex(
403 r'data-height="(\d+)"', more_info, 'height', fatal=False))
404 width = int_or_none(self._search_regex(
405 r'data-width="(\d+)"', more_info, 'width', fatal=False))
406 thumbnail = self._search_regex(
407 r'poster="([^"]+)"', more_info, 'poster', fatal=False)
410 'url': mobj.group('url'),
413 'thumbnail': thumbnail,
417 twitter_card_url = None
418 if 'class="PlayableMedia' in webpage:
419 twitter_card_url = '%s//twitter.com/i/videos/tweet/%s' % (self.http_scheme(), twid)
421 twitter_card_iframe_url = self._search_regex(
422 r'data-full-card-iframe-url=([\'"])(?P<url>(?:(?!\1).)+)\1',
423 webpage, 'Twitter card iframe URL', default=None, group='url')
424 if twitter_card_iframe_url:
425 twitter_card_url = compat_urlparse.urljoin(url, twitter_card_iframe_url)
429 '_type': 'url_transparent',
430 'ie_key': 'TwitterCard',
431 'url': twitter_card_url,
435 raise ExtractorError('There\'s no video in this tweet.')
438 class TwitterAmplifyIE(TwitterBaseIE):
439 IE_NAME = 'twitter:amplify'
440 _VALID_URL = r'https?://amp\.twimg\.com/v/(?P<id>[0-9a-f\-]{36})'
443 'url': 'https://amp.twimg.com/v/0ba0c3c7-0af3-4c0a-bed5-7efd1ffa2951',
444 'md5': '7df102d0b9fd7066b86f3159f8e81bf6',
446 'id': '0ba0c3c7-0af3-4c0a-bed5-7efd1ffa2951',
448 'title': 'Twitter Video',
449 'thumbnail': 're:^https?://.*',
453 def _real_extract(self, url):
454 video_id = self._match_id(url)
455 webpage = self._download_webpage(url, video_id)
457 vmap_url = self._html_search_meta(
458 'twitter:amplify:vmap', webpage, 'vmap url')
459 formats = self._extract_formats_from_vmap_url(vmap_url, video_id)
462 thumbnail = self._html_search_meta(
463 'twitter:image:src', webpage, 'thumbnail', fatal=False)
465 def _find_dimension(target):
466 w = int_or_none(self._html_search_meta(
467 'twitter:%s:width' % target, webpage, fatal=False))
468 h = int_or_none(self._html_search_meta(
469 'twitter:%s:height' % target, webpage, fatal=False))
473 thumbnail_w, thumbnail_h = _find_dimension('image')
476 'width': thumbnail_w,
477 'height': thumbnail_h,
480 video_w, video_h = _find_dimension('player')
488 'title': 'Twitter Video',
490 'thumbnails': thumbnails,