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 _get_vmap_video_url(self, vmap_url, video_id):
24 vmap_data = self._download_xml(vmap_url, video_id)
25 return xpath_text(vmap_data, './/MediaFile').strip()
28 def _search_dimensions_in_video_url(a_format, video_url):
29 m = re.search(r'/(?P<width>\d+)x(?P<height>\d+)/', video_url)
32 'width': int(m.group('width')),
33 'height': int(m.group('height')),
37 class TwitterCardIE(TwitterBaseIE):
38 IE_NAME = 'twitter:card'
39 _VALID_URL = r'https?://(?:www\.)?twitter\.com/i/(?:cards/tfw/v1|videos(?:/tweet)?)/(?P<id>\d+)'
42 'url': 'https://twitter.com/i/cards/tfw/v1/560070183650213889',
43 # MD5 checksums are different in different places
45 'id': '560070183650213889',
47 'title': 'Twitter Card',
48 'thumbnail': r're:^https?://.*\.jpg$',
53 'url': 'https://twitter.com/i/cards/tfw/v1/623160978427936768',
54 'md5': '7ee2a553b63d1bccba97fbed97d9e1c8',
56 'id': '623160978427936768',
58 'title': 'Twitter Card',
59 'thumbnail': r're:^https?://.*\.jpg',
64 'url': 'https://twitter.com/i/cards/tfw/v1/654001591733886977',
65 'md5': 'b6d9683dd3f48e340ded81c0e917ad46',
69 'title': 'Ubuntu 11.10 Overview',
70 'description': 'md5:a831e97fa384863d6e26ce48d1c43376',
71 'upload_date': '20111013',
72 'uploader': 'OMG! Ubuntu!',
73 'uploader_id': 'omgubuntu',
75 'add_ie': ['Youtube'],
78 'url': 'https://twitter.com/i/cards/tfw/v1/665289828897005568',
79 'md5': 'ab2745d0b0ce53319a534fccaa986439',
83 'upload_date': '20151113',
84 'uploader_id': '1189339351084113920',
85 'uploader': 'ArsenalTerje',
86 'title': 'Vine by ArsenalTerje',
90 'url': 'https://twitter.com/i/videos/tweet/705235433198714880',
91 'md5': '3846d0a07109b5ab622425449b59049d',
93 'id': '705235433198714880',
95 'title': 'Twitter web player',
96 'thumbnail': r're:^https?://.*\.jpg',
99 'url': 'https://twitter.com/i/videos/752274308186120192',
100 'only_matching': True,
104 def _parse_media_info(self, media_info, video_id):
106 for media_variant in media_info.get('variants', []):
107 media_url = media_variant['url']
108 if media_url.endswith('.m3u8'):
109 formats.extend(self._extract_m3u8_formats(media_url, video_id, ext='mp4', m3u8_id='hls'))
110 elif media_url.endswith('.mpd'):
111 formats.extend(self._extract_mpd_formats(media_url, video_id, mpd_id='dash'))
113 vbr = int_or_none(dict_get(media_variant, ('bitRate', 'bitrate')), scale=1000)
116 'format_id': 'http-%d' % vbr if vbr else 'http',
119 # Reported bitRate may be zero
120 if not a_format['vbr']:
123 self._search_dimensions_in_video_url(a_format, media_url)
125 formats.append(a_format)
128 def _extract_mobile_formats(self, username, video_id):
129 webpage = self._download_webpage(
130 'https://mobile.twitter.com/%s/status/%s' % (username, video_id),
131 video_id, 'Downloading mobile webpage',
133 # A recent mobile UA is necessary for `gt` cookie
134 'User-Agent': 'Mozilla/5.0 (Android 6.0.1; Mobile; rv:54.0) Gecko/54.0 Firefox/54.0',
136 main_script_url = self._html_search_regex(
137 r'<script[^>]+src="([^"]+main\.[^"]+)"', webpage, 'main script URL')
138 main_script = self._download_webpage(
139 main_script_url, video_id, 'Downloading main script')
140 bearer_token = self._search_regex(
141 r'BEARER_TOKEN\s*:\s*"([^"]+)"',
142 main_script, 'bearer token')
143 guest_token = self._search_regex(
144 r'document\.cookie\s*=\s*decodeURIComponent\("gt=(\d+)',
145 webpage, 'guest token')
146 api_data = self._download_json(
147 'https://api.twitter.com/2/timeline/conversation/%s.json' % video_id,
148 video_id, 'Downloading mobile API data',
150 'Authorization': 'Bearer ' + bearer_token,
151 'x-guest-token': guest_token,
153 media_info = try_get(api_data, lambda o: o['globalObjects']['tweets'][video_id]
154 ['extended_entities']['media'][0]['video_info']) or {}
155 return self._parse_media_info(media_info, video_id)
157 def _real_extract(self, url):
158 video_id = self._match_id(url)
164 webpage = self._download_webpage(url, video_id)
166 iframe_url = self._html_search_regex(
167 r'<iframe[^>]+src="((?:https?:)?//(?:www.youtube.com/embed/[^"]+|(?:www\.)?vine\.co/v/\w+/card))"',
168 webpage, 'video iframe', default=None)
170 return self.url_result(iframe_url)
172 config = self._parse_json(self._html_search_regex(
173 r'data-(?:player-)?config="([^"]+)"', webpage,
174 'data player config', default='{}'),
177 if config.get('source_type') == 'vine':
178 return self.url_result(config['player_url'], 'Vine')
180 periscope_url = PeriscopeIE._extract_url(webpage)
182 return self.url_result(periscope_url, PeriscopeIE.ie_key())
184 video_url = config.get('video_url') or config.get('playlist', [{}])[0].get('source')
187 if determine_ext(video_url) == 'm3u8':
188 formats.extend(self._extract_m3u8_formats(video_url, video_id, ext='mp4', m3u8_id='hls'))
194 self._search_dimensions_in_video_url(f, video_url)
198 vmap_url = config.get('vmapUrl') or config.get('vmap_url')
201 'url': self._get_vmap_video_url(vmap_url, video_id),
206 for entity in config.get('status', {}).get('entities', []):
207 if 'mediaInfo' in entity:
208 media_info = entity['mediaInfo']
211 formats.extend(self._parse_media_info(media_info, video_id))
212 duration = float_or_none(media_info.get('duration', {}).get('nanos'), scale=1e9)
214 username = config.get('user', {}).get('screen_name')
216 formats.extend(self._extract_mobile_formats(username, video_id))
218 self._remove_duplicate_formats(formats)
219 self._sort_formats(formats)
221 title = self._search_regex(r'<title>([^<]+)</title>', webpage, 'title')
222 thumbnail = config.get('posterImageUrl') or config.get('image_src')
223 duration = float_or_none(config.get('duration')) or duration
228 'thumbnail': thumbnail,
229 'duration': duration,
234 class TwitterIE(InfoExtractor):
236 _VALID_URL = r'https?://(?:www\.|m\.|mobile\.)?twitter\.com/(?P<user_id>[^/]+)/status/(?P<id>\d+)'
237 _TEMPLATE_URL = 'https://twitter.com/%s/status/%s'
240 'url': 'https://twitter.com/freethenipple/status/643211948184596480',
242 'id': '643211948184596480',
244 'title': 'FREE THE NIPPLE - FTN supporters on Hollywood Blvd today!',
245 'thumbnail': r're:^https?://.*\.jpg',
246 'description': 'FREE THE NIPPLE on Twitter: "FTN supporters on Hollywood Blvd today! http://t.co/c7jHH749xJ"',
247 'uploader': 'FREE THE NIPPLE',
248 'uploader_id': 'freethenipple',
251 'skip_download': True, # requires ffmpeg
254 'url': 'https://twitter.com/giphz/status/657991469417025536/photo/1',
255 'md5': 'f36dcd5fb92bf7057f155e7d927eeb42',
257 'id': '657991469417025536',
259 'title': 'Gifs - tu vai cai tu vai cai tu nao eh capaz disso tu vai cai',
260 'description': 'Gifs on Twitter: "tu vai cai tu vai cai tu nao eh capaz disso tu vai cai https://t.co/tM46VHFlO5"',
261 'thumbnail': r're:^https?://.*\.png',
263 'uploader_id': 'giphz',
265 'expected_warnings': ['height', 'width'],
266 'skip': 'Account suspended',
268 'url': 'https://twitter.com/starwars/status/665052190608723968',
269 'md5': '39b7199856dee6cd4432e72c74bc69d4',
271 'id': '665052190608723968',
273 'title': 'Star Wars - A new beginning is coming December 18. Watch the official 60 second #TV spot for #StarWars: #TheForceAwakens.',
274 'description': 'Star Wars on Twitter: "A new beginning is coming December 18. Watch the official 60 second #TV spot for #StarWars: #TheForceAwakens."',
275 'uploader_id': 'starwars',
276 'uploader': 'Star Wars',
279 'url': 'https://twitter.com/BTNBrentYarina/status/705235433198714880',
281 'id': '705235433198714880',
283 'title': 'Brent Yarina - Khalil Iverson\'s missed highlight dunk. And made highlight dunk. In one highlight.',
284 'description': 'Brent Yarina on Twitter: "Khalil Iverson\'s missed highlight dunk. And made highlight dunk. In one highlight."',
285 'uploader_id': 'BTNBrentYarina',
286 'uploader': 'Brent Yarina',
289 # The same video as https://twitter.com/i/videos/tweet/705235433198714880
290 # Test case of TwitterCardIE
291 'skip_download': True,
294 'url': 'https://twitter.com/jaydingeer/status/700207533655363584',
297 'id': '700207533655363584',
299 'title': 'JG - BEAT PROD: @suhmeduh #Damndaniel',
300 'description': 'JG on Twitter: "BEAT PROD: @suhmeduh https://t.co/HBrQ4AfpvZ #Damndaniel https://t.co/byBooq2ejZ"',
301 'thumbnail': r're:^https?://.*\.jpg',
303 'uploader_id': 'jaydingeer',
306 'skip_download': True, # requires ffmpeg
309 'url': 'https://twitter.com/Filmdrunk/status/713801302971588609',
310 'md5': '89a15ed345d13b86e9a5a5e051fa308a',
314 'title': 'Dr.Pepperの飲み方 #japanese #バカ #ドクペ #電動ガン',
315 'uploader': 'TAKUMA',
316 'uploader_id': '1004126642786242560',
317 'upload_date': '20140615',
321 'url': 'https://twitter.com/captainamerica/status/719944021058060289',
323 'id': '719944021058060289',
325 'title': 'Captain America - @King0fNerd Are you sure you made the right choice? Find out in theaters.',
326 'description': 'Captain America on Twitter: "@King0fNerd Are you sure you made the right choice? Find out in theaters. https://t.co/GpgYi9xMJI"',
327 'uploader_id': 'captainamerica',
328 'uploader': 'Captain America',
331 'skip_download': True, # requires ffmpeg
334 'url': 'https://twitter.com/OPP_HSD/status/779210622571536384',
336 'id': '1zqKVVlkqLaKB',
338 'title': 'Sgt Kerry Schmidt - LIVE on #Periscope: Road rage, mischief, assault, rollover and fire in one occurrence',
339 'description': 'Sgt Kerry Schmidt on Twitter: "LIVE on #Periscope: Road rage, mischief, assault, rollover and fire in one occurrence https://t.co/EKrVgIXF3s"',
340 'upload_date': '20160923',
341 'uploader_id': 'OPP_HSD',
342 'uploader': 'Sgt Kerry Schmidt',
343 'timestamp': 1474613214,
345 'add_ie': ['Periscope'],
347 # has mp4 formats via mobile API
348 'url': 'https://twitter.com/news_al3alm/status/852138619213144067',
350 'id': '852138619213144067',
352 'title': 'عالم الأخبار - كلمة تاريخية بجلسة الجناسي التاريخية.. النائب خالد مؤنس العتيبي للمعارضين : اتقوا الله .. الظلم ظلمات يوم القيامة',
353 '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"',
354 'uploader': 'عالم الأخبار',
355 'uploader_id': 'news_al3alm',
358 'format': 'best[format_id^=http-]',
362 def _real_extract(self, url):
363 mobj = re.match(self._VALID_URL, url)
364 user_id = mobj.group('user_id')
365 twid = mobj.group('id')
367 webpage, urlh = self._download_webpage_handle(
368 self._TEMPLATE_URL % (user_id, twid), twid)
370 if 'twitter.com/account/suspended' in urlh.geturl():
371 raise ExtractorError('Account suspended by Twitter.', expected=True)
373 username = remove_end(self._og_search_title(webpage), ' on Twitter')
375 title = description = self._og_search_description(webpage).strip('').replace('\n', ' ').strip('“”')
377 # strip 'https -_t.co_BJYgOjSeGA' junk from filenames
378 title = re.sub(r'\s+(https?://[^ ]+)', '', title)
381 'uploader_id': user_id,
382 'uploader': username,
384 'description': '%s on Twitter: "%s"' % (username, description),
385 'title': username + ' - ' + title,
388 mobj = re.search(r'''(?x)
389 <video[^>]+class="animated-gif"(?P<more_info>[^>]+)>\s*
390 <source[^>]+video-src="(?P<url>[^"]+)"
394 more_info = mobj.group('more_info')
395 height = int_or_none(self._search_regex(
396 r'data-height="(\d+)"', more_info, 'height', fatal=False))
397 width = int_or_none(self._search_regex(
398 r'data-width="(\d+)"', more_info, 'width', fatal=False))
399 thumbnail = self._search_regex(
400 r'poster="([^"]+)"', more_info, 'poster', fatal=False)
403 'url': mobj.group('url'),
406 'thumbnail': thumbnail,
410 twitter_card_url = None
411 if 'class="PlayableMedia' in webpage:
412 twitter_card_url = '%s//twitter.com/i/videos/tweet/%s' % (self.http_scheme(), twid)
414 twitter_card_iframe_url = self._search_regex(
415 r'data-full-card-iframe-url=([\'"])(?P<url>(?:(?!\1).)+)\1',
416 webpage, 'Twitter card iframe URL', default=None, group='url')
417 if twitter_card_iframe_url:
418 twitter_card_url = compat_urlparse.urljoin(url, twitter_card_iframe_url)
422 '_type': 'url_transparent',
423 'ie_key': 'TwitterCard',
424 'url': twitter_card_url,
428 raise ExtractorError('There\'s no video in this tweet.')
431 class TwitterAmplifyIE(TwitterBaseIE):
432 IE_NAME = 'twitter:amplify'
433 _VALID_URL = r'https?://amp\.twimg\.com/v/(?P<id>[0-9a-f\-]{36})'
436 'url': 'https://amp.twimg.com/v/0ba0c3c7-0af3-4c0a-bed5-7efd1ffa2951',
437 'md5': '7df102d0b9fd7066b86f3159f8e81bf6',
439 'id': '0ba0c3c7-0af3-4c0a-bed5-7efd1ffa2951',
441 'title': 'Twitter Video',
442 'thumbnail': 're:^https?://.*',
446 def _real_extract(self, url):
447 video_id = self._match_id(url)
448 webpage = self._download_webpage(url, video_id)
450 vmap_url = self._html_search_meta(
451 'twitter:amplify:vmap', webpage, 'vmap url')
452 video_url = self._get_vmap_video_url(vmap_url, video_id)
455 thumbnail = self._html_search_meta(
456 'twitter:image:src', webpage, 'thumbnail', fatal=False)
458 def _find_dimension(target):
459 w = int_or_none(self._html_search_meta(
460 'twitter:%s:width' % target, webpage, fatal=False))
461 h = int_or_none(self._html_search_meta(
462 'twitter:%s:height' % target, webpage, fatal=False))
466 thumbnail_w, thumbnail_h = _find_dimension('image')
469 'width': thumbnail_w,
470 'height': thumbnail_h,
473 video_w, video_h = _find_dimension('player')
482 'title': 'Twitter Video',
484 'thumbnails': thumbnails,