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',
97 'url': 'https://twitter.com/i/videos/tweet/705235433198714880',
98 'md5': '3846d0a07109b5ab622425449b59049d',
100 'id': '705235433198714880',
102 'title': 'Twitter web player',
103 'thumbnail': r're:^https?://.*\.jpg',
106 'url': 'https://twitter.com/i/videos/752274308186120192',
107 'only_matching': True,
111 def _parse_media_info(self, media_info, video_id):
113 for media_variant in media_info.get('variants', []):
114 media_url = media_variant['url']
115 if media_url.endswith('.m3u8'):
116 formats.extend(self._extract_m3u8_formats(media_url, video_id, ext='mp4', m3u8_id='hls'))
117 elif media_url.endswith('.mpd'):
118 formats.extend(self._extract_mpd_formats(media_url, video_id, mpd_id='dash'))
120 vbr = int_or_none(dict_get(media_variant, ('bitRate', 'bitrate')), scale=1000)
123 'format_id': 'http-%d' % vbr if vbr else 'http',
126 # Reported bitRate may be zero
127 if not a_format['vbr']:
130 self._search_dimensions_in_video_url(a_format, media_url)
132 formats.append(a_format)
135 def _extract_mobile_formats(self, username, video_id):
136 webpage = self._download_webpage(
137 'https://mobile.twitter.com/%s/status/%s' % (username, video_id),
138 video_id, 'Downloading mobile webpage',
140 # A recent mobile UA is necessary for `gt` cookie
141 'User-Agent': 'Mozilla/5.0 (Android 6.0.1; Mobile; rv:54.0) Gecko/54.0 Firefox/54.0',
143 main_script_url = self._html_search_regex(
144 r'<script[^>]+src="([^"]+main\.[^"]+)"', webpage, 'main script URL')
145 main_script = self._download_webpage(
146 main_script_url, video_id, 'Downloading main script')
147 bearer_token = self._search_regex(
148 r'BEARER_TOKEN\s*:\s*"([^"]+)"',
149 main_script, 'bearer token')
150 guest_token = self._search_regex(
151 r'document\.cookie\s*=\s*decodeURIComponent\("gt=(\d+)',
152 webpage, 'guest token')
153 api_data = self._download_json(
154 'https://api.twitter.com/2/timeline/conversation/%s.json' % video_id,
155 video_id, 'Downloading mobile API data',
157 'Authorization': 'Bearer ' + bearer_token,
158 'x-guest-token': guest_token,
160 media_info = try_get(api_data, lambda o: o['globalObjects']['tweets'][video_id]
161 ['extended_entities']['media'][0]['video_info']) or {}
162 return self._parse_media_info(media_info, video_id)
164 def _real_extract(self, url):
165 video_id = self._match_id(url)
171 webpage = self._download_webpage(url, video_id)
173 iframe_url = self._html_search_regex(
174 r'<iframe[^>]+src="((?:https?:)?//(?:www.youtube.com/embed/[^"]+|(?:www\.)?vine\.co/v/\w+/card))"',
175 webpage, 'video iframe', default=None)
177 return self.url_result(iframe_url)
179 config = self._parse_json(self._html_search_regex(
180 r'data-(?:player-)?config="([^"]+)"', webpage,
181 'data player config', default='{}'),
184 if config.get('source_type') == 'vine':
185 return self.url_result(config['player_url'], 'Vine')
187 periscope_url = PeriscopeIE._extract_url(webpage)
189 return self.url_result(periscope_url, PeriscopeIE.ie_key())
191 video_url = config.get('video_url') or config.get('playlist', [{}])[0].get('source')
194 if determine_ext(video_url) == 'm3u8':
195 formats.extend(self._extract_m3u8_formats(video_url, video_id, ext='mp4', m3u8_id='hls'))
201 self._search_dimensions_in_video_url(f, video_url)
205 vmap_url = config.get('vmapUrl') or config.get('vmap_url')
208 self._extract_formats_from_vmap_url(vmap_url, video_id))
212 for entity in config.get('status', {}).get('entities', []):
213 if 'mediaInfo' in entity:
214 media_info = entity['mediaInfo']
217 formats.extend(self._parse_media_info(media_info, video_id))
218 duration = float_or_none(media_info.get('duration', {}).get('nanos'), scale=1e9)
220 username = config.get('user', {}).get('screen_name')
222 formats.extend(self._extract_mobile_formats(username, video_id))
224 self._remove_duplicate_formats(formats)
225 self._sort_formats(formats)
227 title = self._search_regex(r'<title>([^<]+)</title>', webpage, 'title')
228 thumbnail = config.get('posterImageUrl') or config.get('image_src')
229 duration = float_or_none(config.get('duration')) or duration
234 'thumbnail': thumbnail,
235 'duration': duration,
240 class TwitterIE(InfoExtractor):
242 _VALID_URL = r'https?://(?:www\.|m\.|mobile\.)?twitter\.com/(?P<user_id>[^/]+)/status/(?P<id>\d+)'
243 _TEMPLATE_URL = 'https://twitter.com/%s/status/%s'
246 'url': 'https://twitter.com/freethenipple/status/643211948184596480',
248 'id': '643211948184596480',
250 'title': 'FREE THE NIPPLE - FTN supporters on Hollywood Blvd today!',
251 'thumbnail': r're:^https?://.*\.jpg',
252 'description': 'FREE THE NIPPLE on Twitter: "FTN supporters on Hollywood Blvd today! http://t.co/c7jHH749xJ"',
253 'uploader': 'FREE THE NIPPLE',
254 'uploader_id': 'freethenipple',
257 'skip_download': True, # requires ffmpeg
260 'url': 'https://twitter.com/giphz/status/657991469417025536/photo/1',
261 'md5': 'f36dcd5fb92bf7057f155e7d927eeb42',
263 'id': '657991469417025536',
265 'title': 'Gifs - tu vai cai tu vai cai tu nao eh capaz disso tu vai cai',
266 'description': 'Gifs on Twitter: "tu vai cai tu vai cai tu nao eh capaz disso tu vai cai https://t.co/tM46VHFlO5"',
267 'thumbnail': r're:^https?://.*\.png',
269 'uploader_id': 'giphz',
271 'expected_warnings': ['height', 'width'],
272 'skip': 'Account suspended',
274 'url': 'https://twitter.com/starwars/status/665052190608723968',
275 'md5': '39b7199856dee6cd4432e72c74bc69d4',
277 'id': '665052190608723968',
279 'title': 'Star Wars - A new beginning is coming December 18. Watch the official 60 second #TV spot for #StarWars: #TheForceAwakens.',
280 'description': 'Star Wars on Twitter: "A new beginning is coming December 18. Watch the official 60 second #TV spot for #StarWars: #TheForceAwakens."',
281 'uploader_id': 'starwars',
282 'uploader': 'Star Wars',
285 'url': 'https://twitter.com/BTNBrentYarina/status/705235433198714880',
287 'id': '705235433198714880',
289 'title': 'Brent Yarina - Khalil Iverson\'s missed highlight dunk. And made highlight dunk. In one highlight.',
290 'description': 'Brent Yarina on Twitter: "Khalil Iverson\'s missed highlight dunk. And made highlight dunk. In one highlight."',
291 'uploader_id': 'BTNBrentYarina',
292 'uploader': 'Brent Yarina',
295 # The same video as https://twitter.com/i/videos/tweet/705235433198714880
296 # Test case of TwitterCardIE
297 'skip_download': True,
300 'url': 'https://twitter.com/jaydingeer/status/700207533655363584',
303 'id': '700207533655363584',
305 'title': 'JG - BEAT PROD: @suhmeduh #Damndaniel',
306 'description': 'JG on Twitter: "BEAT PROD: @suhmeduh https://t.co/HBrQ4AfpvZ #Damndaniel https://t.co/byBooq2ejZ"',
307 'thumbnail': r're:^https?://.*\.jpg',
309 'uploader_id': 'jaydingeer',
312 'skip_download': True, # requires ffmpeg
315 'url': 'https://twitter.com/Filmdrunk/status/713801302971588609',
316 'md5': '89a15ed345d13b86e9a5a5e051fa308a',
320 'title': 'Dr.Pepperの飲み方 #japanese #バカ #ドクペ #電動ガン',
321 'uploader': 'TAKUMA',
322 'uploader_id': '1004126642786242560',
323 'upload_date': '20140615',
327 'url': 'https://twitter.com/captainamerica/status/719944021058060289',
329 'id': '719944021058060289',
331 'title': 'Captain America - @King0fNerd Are you sure you made the right choice? Find out in theaters.',
332 'description': 'Captain America on Twitter: "@King0fNerd Are you sure you made the right choice? Find out in theaters. https://t.co/GpgYi9xMJI"',
333 'uploader_id': 'captainamerica',
334 'uploader': 'Captain America',
337 'skip_download': True, # requires ffmpeg
340 'url': 'https://twitter.com/OPP_HSD/status/779210622571536384',
342 'id': '1zqKVVlkqLaKB',
344 'title': 'Sgt Kerry Schmidt - LIVE on #Periscope: Road rage, mischief, assault, rollover and fire in one occurrence',
345 'description': 'Sgt Kerry Schmidt on Twitter: "LIVE on #Periscope: Road rage, mischief, assault, rollover and fire in one occurrence https://t.co/EKrVgIXF3s"',
346 'upload_date': '20160923',
347 'uploader_id': 'OPP_HSD',
348 'uploader': 'Sgt Kerry Schmidt',
349 'timestamp': 1474613214,
351 'add_ie': ['Periscope'],
353 # has mp4 formats via mobile API
354 'url': 'https://twitter.com/news_al3alm/status/852138619213144067',
356 'id': '852138619213144067',
358 'title': 'عالم الأخبار - كلمة تاريخية بجلسة الجناسي التاريخية.. النائب خالد مؤنس العتيبي للمعارضين : اتقوا الله .. الظلم ظلمات يوم القيامة',
359 '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"',
360 'uploader': 'عالم الأخبار',
361 'uploader_id': 'news_al3alm',
364 'format': 'best[format_id^=http-]',
368 def _real_extract(self, url):
369 mobj = re.match(self._VALID_URL, url)
370 user_id = mobj.group('user_id')
371 twid = mobj.group('id')
373 webpage, urlh = self._download_webpage_handle(
374 self._TEMPLATE_URL % (user_id, twid), twid)
376 if 'twitter.com/account/suspended' in urlh.geturl():
377 raise ExtractorError('Account suspended by Twitter.', expected=True)
379 username = remove_end(self._og_search_title(webpage), ' on Twitter')
381 title = description = self._og_search_description(webpage).strip('').replace('\n', ' ').strip('“”')
383 # strip 'https -_t.co_BJYgOjSeGA' junk from filenames
384 title = re.sub(r'\s+(https?://[^ ]+)', '', title)
387 'uploader_id': user_id,
388 'uploader': username,
390 'description': '%s on Twitter: "%s"' % (username, description),
391 'title': username + ' - ' + title,
394 mobj = re.search(r'''(?x)
395 <video[^>]+class="animated-gif"(?P<more_info>[^>]+)>\s*
396 <source[^>]+video-src="(?P<url>[^"]+)"
400 more_info = mobj.group('more_info')
401 height = int_or_none(self._search_regex(
402 r'data-height="(\d+)"', more_info, 'height', fatal=False))
403 width = int_or_none(self._search_regex(
404 r'data-width="(\d+)"', more_info, 'width', fatal=False))
405 thumbnail = self._search_regex(
406 r'poster="([^"]+)"', more_info, 'poster', fatal=False)
409 'url': mobj.group('url'),
412 'thumbnail': thumbnail,
416 twitter_card_url = None
417 if 'class="PlayableMedia' in webpage:
418 twitter_card_url = '%s//twitter.com/i/videos/tweet/%s' % (self.http_scheme(), twid)
420 twitter_card_iframe_url = self._search_regex(
421 r'data-full-card-iframe-url=([\'"])(?P<url>(?:(?!\1).)+)\1',
422 webpage, 'Twitter card iframe URL', default=None, group='url')
423 if twitter_card_iframe_url:
424 twitter_card_url = compat_urlparse.urljoin(url, twitter_card_iframe_url)
428 '_type': 'url_transparent',
429 'ie_key': 'TwitterCard',
430 'url': twitter_card_url,
434 raise ExtractorError('There\'s no video in this tweet.')
437 class TwitterAmplifyIE(TwitterBaseIE):
438 IE_NAME = 'twitter:amplify'
439 _VALID_URL = r'https?://amp\.twimg\.com/v/(?P<id>[0-9a-f\-]{36})'
442 'url': 'https://amp.twimg.com/v/0ba0c3c7-0af3-4c0a-bed5-7efd1ffa2951',
443 'md5': '7df102d0b9fd7066b86f3159f8e81bf6',
445 'id': '0ba0c3c7-0af3-4c0a-bed5-7efd1ffa2951',
447 'title': 'Twitter Video',
448 'thumbnail': 're:^https?://.*',
452 def _real_extract(self, url):
453 video_id = self._match_id(url)
454 webpage = self._download_webpage(url, video_id)
456 vmap_url = self._html_search_meta(
457 'twitter:amplify:vmap', webpage, 'vmap url')
458 formats = self._extract_formats_from_vmap_url(vmap_url, video_id)
461 thumbnail = self._html_search_meta(
462 'twitter:image:src', webpage, 'thumbnail', fatal=False)
464 def _find_dimension(target):
465 w = int_or_none(self._html_search_meta(
466 'twitter:%s:width' % target, webpage, fatal=False))
467 h = int_or_none(self._html_search_meta(
468 'twitter:%s:height' % target, webpage, fatal=False))
472 thumbnail_w, thumbnail_h = _find_dimension('image')
475 'width': thumbnail_w,
476 'height': thumbnail_h,
479 video_w, video_h = _find_dimension('player')
487 'title': 'Twitter Video',
489 'thumbnails': thumbnails,