2 from __future__ import unicode_literals
6 from .common import InfoExtractor
17 class TwitterBaseIE(InfoExtractor):
18 def _get_vmap_video_url(self, vmap_url, video_id):
19 vmap_data = self._download_xml(vmap_url, video_id)
20 return xpath_text(vmap_data, './/MediaFile').strip()
23 class TwitterCardIE(TwitterBaseIE):
24 IE_NAME = 'twitter:card'
25 _VALID_URL = r'https?://(?:www\.)?twitter\.com/i/(?:cards/tfw/v1|videos/tweet)/(?P<id>\d+)'
28 'url': 'https://twitter.com/i/cards/tfw/v1/560070183650213889',
29 # MD5 checksums are different in different places
31 'id': '560070183650213889',
33 'title': 'Twitter Card',
34 'thumbnail': 're:^https?://.*\.jpg$',
39 'url': 'https://twitter.com/i/cards/tfw/v1/623160978427936768',
40 'md5': '7ee2a553b63d1bccba97fbed97d9e1c8',
42 'id': '623160978427936768',
44 'title': 'Twitter Card',
45 'thumbnail': 're:^https?://.*\.jpg',
50 'url': 'https://twitter.com/i/cards/tfw/v1/654001591733886977',
51 'md5': 'd4724ffe6d2437886d004fa5de1043b3',
55 'title': 'Ubuntu 11.10 Overview',
56 'description': 'Take a quick peek at what\'s new and improved in Ubuntu 11.10.\n\nOnce installed take a look at 10 Things to Do After Installing: http://www.omgubuntu.co.uk/2011/10/10-things-to-do-after-installing-ubuntu-11-10/',
57 'upload_date': '20111013',
58 'uploader': 'OMG! Ubuntu!',
59 'uploader_id': 'omgubuntu',
61 'add_ie': ['Youtube'],
64 'url': 'https://twitter.com/i/cards/tfw/v1/665289828897005568',
65 'md5': 'ab2745d0b0ce53319a534fccaa986439',
69 'upload_date': '20151113',
70 'uploader_id': '1189339351084113920',
71 'uploader': 'ArsenalTerje',
72 'title': 'Vine by ArsenalTerje',
76 'url': 'https://twitter.com/i/videos/tweet/705235433198714880',
77 'md5': '3846d0a07109b5ab622425449b59049d',
79 'id': '705235433198714880',
81 'title': 'Twitter web player',
82 'thumbnail': 're:^https?://.*\.jpg',
87 def _real_extract(self, url):
88 video_id = self._match_id(url)
90 # Different formats served for different User-Agents
92 'Mozilla/5.0 (X11; Linux x86_64; rv:10.0) Gecko/20150101 Firefox/20.0 (Chrome)', # mp4
93 'Mozilla/5.0 (Windows NT 5.2; WOW64; rv:38.0) Gecko/20100101 Firefox/38.0', # webm
98 for user_agent in USER_AGENTS:
99 request = sanitized_Request(url)
100 request.add_header('User-Agent', user_agent)
101 webpage = self._download_webpage(request, video_id)
103 iframe_url = self._html_search_regex(
104 r'<iframe[^>]+src="((?:https?:)?//(?:www.youtube.com/embed/[^"]+|(?:www\.)?vine\.co/v/\w+/card))"',
105 webpage, 'video iframe', default=None)
107 return self.url_result(iframe_url)
109 config = self._parse_json(self._html_search_regex(
110 r'data-(?:player-)?config="([^"]+)"', webpage, 'data player config'),
112 if 'playlist' not in config:
113 vmap_url = config.get('vmapUrl') or config.get('vmap_url')
116 'url': self._get_vmap_video_url(vmap_url, video_id),
118 break # same video regardless of UA
121 video_url = config['playlist'][0]['source']
127 m = re.search(r'/(?P<width>\d+)x(?P<height>\d+)/', video_url)
130 'width': int(m.group('width')),
131 'height': int(m.group('height')),
134 self._sort_formats(formats)
136 title = self._search_regex(r'<title>([^<]+)</title>', webpage, 'title')
137 thumbnail = config.get('posterImageUrl') or config.get('image_src')
138 duration = float_or_none(config.get('duration'))
143 'thumbnail': thumbnail,
144 'duration': duration,
149 class TwitterIE(InfoExtractor):
151 _VALID_URL = r'https?://(?:www\.|m\.|mobile\.)?twitter\.com/(?P<user_id>[^/]+)/status/(?P<id>\d+)'
152 _TEMPLATE_URL = 'https://twitter.com/%s/status/%s'
155 'url': 'https://twitter.com/freethenipple/status/643211948184596480',
156 # MD5 checksums are different in different places
158 'id': '643211948184596480',
160 'title': 'FREE THE NIPPLE - FTN supporters on Hollywood Blvd today!',
161 'thumbnail': 're:^https?://.*\.jpg',
163 'description': 'FREE THE NIPPLE on Twitter: "FTN supporters on Hollywood Blvd today! http://t.co/c7jHH749xJ"',
164 'uploader': 'FREE THE NIPPLE',
165 'uploader_id': 'freethenipple',
168 'url': 'https://twitter.com/giphz/status/657991469417025536/photo/1',
169 'md5': 'f36dcd5fb92bf7057f155e7d927eeb42',
171 'id': '657991469417025536',
173 'title': 'Gifs - tu vai cai tu vai cai tu nao eh capaz disso tu vai cai',
174 'description': 'Gifs on Twitter: "tu vai cai tu vai cai tu nao eh capaz disso tu vai cai https://t.co/tM46VHFlO5"',
175 'thumbnail': 're:^https?://.*\.png',
177 'uploader_id': 'giphz',
179 'expected_warnings': ['height', 'width'],
181 'url': 'https://twitter.com/starwars/status/665052190608723968',
182 'md5': '39b7199856dee6cd4432e72c74bc69d4',
184 'id': '665052190608723968',
186 'title': 'Star Wars - A new beginning is coming December 18. Watch the official 60 second #TV spot for #StarWars: #TheForceAwakens.',
187 'description': 'Star Wars on Twitter: "A new beginning is coming December 18. Watch the official 60 second #TV spot for #StarWars: #TheForceAwakens."',
188 'uploader_id': 'starwars',
189 'uploader': 'Star Wars',
192 'url': 'https://twitter.com/BTNBrentYarina/status/705235433198714880',
194 'id': '705235433198714880',
196 'title': 'Brent Yarina - Khalil Iverson\'s missed highlight dunk. And made highlight dunk. In one highlight.',
197 'description': 'Brent Yarina on Twitter: "Khalil Iverson\'s missed highlight dunk. And made highlight dunk. In one highlight."',
198 'uploader_id': 'BTNBrentYarina',
199 'uploader': 'Brent Yarina',
202 # The same video as https://twitter.com/i/videos/tweet/705235433198714880
203 # Test case of TwitterCardIE
204 'skip_download': True,
208 def _real_extract(self, url):
209 mobj = re.match(self._VALID_URL, url)
210 user_id = mobj.group('user_id')
211 twid = mobj.group('id')
213 webpage = self._download_webpage(self._TEMPLATE_URL % (user_id, twid), twid)
215 username = remove_end(self._og_search_title(webpage), ' on Twitter')
217 title = description = self._og_search_description(webpage).strip('').replace('\n', ' ').strip('“”')
219 # strip 'https -_t.co_BJYgOjSeGA' junk from filenames
220 title = re.sub(r'\s+(https?://[^ ]+)', '', title)
223 'uploader_id': user_id,
224 'uploader': username,
226 'description': '%s on Twitter: "%s"' % (username, description),
227 'title': username + ' - ' + title,
230 card_id = self._search_regex(
231 r'["\']/i/cards/tfw/v1/(\d+)', webpage, 'twitter card url', default=None)
233 card_url = 'https://twitter.com/i/cards/tfw/v1/' + card_id
235 '_type': 'url_transparent',
236 'ie_key': 'TwitterCard',
241 mobj = re.search(r'''(?x)
242 <video[^>]+class="animated-gif"(?P<more_info>[^>]+)>\s*
243 <source[^>]+video-src="(?P<url>[^"]+)"
247 more_info = mobj.group('more_info')
248 height = int_or_none(self._search_regex(
249 r'data-height="(\d+)"', more_info, 'height', fatal=False))
250 width = int_or_none(self._search_regex(
251 r'data-width="(\d+)"', more_info, 'width', fatal=False))
252 thumbnail = self._search_regex(
253 r'poster="([^"]+)"', more_info, 'poster', fatal=False)
256 'url': mobj.group('url'),
259 'thumbnail': thumbnail,
263 if 'class="PlayableMedia' in webpage:
265 '_type': 'url_transparent',
266 'ie_key': 'TwitterCard',
267 'url': '%s//twitter.com/i/videos/tweet/%s' % (self.http_scheme(), twid),
272 raise ExtractorError('There\'s no video in this tweet.')
275 class TwitterAmplifyIE(TwitterBaseIE):
276 IE_NAME = 'twitter:amplify'
277 _VALID_URL = 'https?://amp\.twimg\.com/v/(?P<id>[0-9a-f\-]{36})'
280 'url': 'https://amp.twimg.com/v/0ba0c3c7-0af3-4c0a-bed5-7efd1ffa2951',
281 'md5': '7df102d0b9fd7066b86f3159f8e81bf6',
283 'id': '0ba0c3c7-0af3-4c0a-bed5-7efd1ffa2951',
285 'title': 'Twitter Video',
286 'thumbnail': 're:^https?://.*',
290 def _real_extract(self, url):
291 video_id = self._match_id(url)
292 webpage = self._download_webpage(url, video_id)
294 vmap_url = self._html_search_meta(
295 'twitter:amplify:vmap', webpage, 'vmap url')
296 video_url = self._get_vmap_video_url(vmap_url, video_id)
299 thumbnail = self._html_search_meta(
300 'twitter:image:src', webpage, 'thumbnail', fatal=False)
302 def _find_dimension(target):
303 w = int_or_none(self._html_search_meta(
304 'twitter:%s:width' % target, webpage, fatal=False))
305 h = int_or_none(self._html_search_meta(
306 'twitter:%s:height' % target, webpage, fatal=False))
310 thumbnail_w, thumbnail_h = _find_dimension('image')
313 'width': thumbnail_w,
314 'height': thumbnail_h,
317 video_w, video_h = _find_dimension('player')
326 'title': 'Twitter Video',
328 'thumbnails': thumbnails,