2 from __future__ import unicode_literals
6 from .common import InfoExtractor
7 from ..compat import compat_urllib_request
17 class TwitterCardIE(InfoExtractor):
18 IE_NAME = 'twitter:card'
19 _VALID_URL = r'https?://(?:www\.)?twitter\.com/i/cards/tfw/v1/(?P<id>\d+)'
22 'url': 'https://twitter.com/i/cards/tfw/v1/560070183650213889',
23 'md5': '4fa26a35f9d1bf4b646590ba8e84be19',
25 'id': '560070183650213889',
27 'title': 'TwitterCard',
28 'thumbnail': 're:^https?://.*\.jpg$',
33 'url': 'https://twitter.com/i/cards/tfw/v1/623160978427936768',
34 'md5': '7ee2a553b63d1bccba97fbed97d9e1c8',
36 'id': '623160978427936768',
38 'title': 'TwitterCard',
39 'thumbnail': 're:^https?://.*\.jpg',
44 'url': 'https://twitter.com/i/cards/tfw/v1/654001591733886977',
45 'md5': 'b6f35e8b08a0bec6c8af77a2f4b3a814',
49 'title': 'Ubuntu 11.10 Overview',
50 '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/',
51 'upload_date': '20111013',
52 'uploader': 'OMG! Ubuntu!',
53 'uploader_id': 'omgubuntu',
57 'url': 'https://twitter.com/i/cards/tfw/v1/665289828897005568',
58 'md5': 'ab2745d0b0ce53319a534fccaa986439',
62 'upload_date': '20151113',
63 'uploader_id': '1189339351084113920',
64 'uploader': '@ArsenalTerje',
65 'title': 'Vine by @ArsenalTerje',
71 def _real_extract(self, url):
72 video_id = self._match_id(url)
74 # Different formats served for different User-Agents
76 'Mozilla/5.0 (X11; Linux x86_64; rv:10.0) Gecko/20150101 Firefox/20.0 (Chrome)', # mp4
77 'Mozilla/5.0 (Windows NT 5.2; WOW64; rv:38.0) Gecko/20100101 Firefox/38.0', # webm
82 for user_agent in USER_AGENTS:
83 request = compat_urllib_request.Request(url)
84 request.add_header('User-Agent', user_agent)
85 webpage = self._download_webpage(request, video_id)
87 iframe_url = self._html_search_regex(
88 r'<iframe[^>]+src="((?:https?:)?//(?:www.youtube.com/embed/[^"]+|(?:www\.)?vine\.co/v/\w+/card))"',
89 webpage, 'video iframe', default=None)
91 return self.url_result(iframe_url)
93 config = self._parse_json(self._html_search_regex(
94 r'data-player-config="([^"]+)"', webpage, 'data player config'),
96 if 'playlist' not in config:
97 if 'vmapUrl' in config:
98 vmap_data = self._download_xml(config['vmapUrl'], video_id)
99 video_url = xpath_text(vmap_data, './/MediaFile').strip()
103 break # same video regardless of UA
106 video_url = config['playlist'][0]['source']
112 m = re.search(r'/(?P<width>\d+)x(?P<height>\d+)/', video_url)
115 'width': int(m.group('width')),
116 'height': int(m.group('height')),
119 self._sort_formats(formats)
121 thumbnail = config.get('posterImageUrl')
122 duration = float_or_none(config.get('duration'))
126 'title': 'TwitterCard',
127 'thumbnail': thumbnail,
128 'duration': duration,
133 class TwitterIE(InfoExtractor):
135 _VALID_URL = r'https?://(?:www\.|m\.|mobile\.)?twitter\.com/(?P<user_id>[^/]+)/status/(?P<id>\d+)'
136 _TEMPLATE_URL = 'https://twitter.com/%s/status/%s'
139 'url': 'https://twitter.com/freethenipple/status/643211948184596480',
140 'md5': 'db6612ec5d03355953c3ca9250c97e5e',
142 'id': '643211948184596480',
144 'title': 'FREE THE NIPPLE - FTN supporters on Hollywood Blvd today!',
145 'thumbnail': 're:^https?://.*\.jpg',
147 'description': 'FREE THE NIPPLE on Twitter: "FTN supporters on Hollywood Blvd today! http://t.co/c7jHH749xJ"',
148 'uploader': 'FREE THE NIPPLE',
149 'uploader_id': 'freethenipple',
152 'url': 'https://twitter.com/giphz/status/657991469417025536/photo/1',
153 'md5': 'f36dcd5fb92bf7057f155e7d927eeb42',
155 'id': '657991469417025536',
157 'title': 'Gifs - tu vai cai tu vai cai tu nao eh capaz disso tu vai cai',
158 'description': 'Gifs on Twitter: "tu vai cai tu vai cai tu nao eh capaz disso tu vai cai https://t.co/tM46VHFlO5"',
159 'thumbnail': 're:^https?://.*\.png',
161 'uploader_id': 'giphz',
164 'url': 'https://twitter.com/starwars/status/665052190608723968',
165 'md5': '39b7199856dee6cd4432e72c74bc69d4',
167 'id': '665052190608723968',
169 'title': 'Star Wars - A new beginning is coming December 18. Watch the official 60 second #TV spot for #StarWars: #TheForceAwakens.',
170 'description': 'Star Wars on Twitter: "A new beginning is coming December 18. Watch the official 60 second #TV spot for #StarWars: #TheForceAwakens."',
171 'uploader_id': 'starwars',
172 'uploader': 'Star Wars',
176 def _real_extract(self, url):
177 mobj = re.match(self._VALID_URL, url)
178 user_id = mobj.group('user_id')
179 twid = mobj.group('id')
181 webpage = self._download_webpage(self._TEMPLATE_URL % (user_id, twid), twid)
183 username = remove_end(self._og_search_title(webpage), ' on Twitter')
185 title = description = self._og_search_description(webpage).strip('').replace('\n', ' ').strip('“”')
187 # strip 'https -_t.co_BJYgOjSeGA' junk from filenames
188 title = re.sub(r'\s+(https?://[^ ]+)', '', title)
191 'uploader_id': user_id,
192 'uploader': username,
194 'description': '%s on Twitter: "%s"' % (username, description),
195 'title': username + ' - ' + title,
198 card_id = self._search_regex(
199 r'["\']/i/cards/tfw/v1/(\d+)', webpage, 'twitter card url', default=None)
201 card_url = 'https://twitter.com/i/cards/tfw/v1/' + card_id
203 '_type': 'url_transparent',
204 'ie_key': 'TwitterCard',
209 mobj = re.search(r'''(?x)
210 <video[^>]+class="animated-gif"[^>]+
211 (?:data-height="(?P<height>\d+)")?[^>]+
212 (?:data-width="(?P<width>\d+)")?[^>]+
213 (?:poster="(?P<poster>[^"]+)")?[^>]*>\s*
214 <source[^>]+video-src="(?P<url>[^"]+)"
220 'url': mobj.group('url'),
221 'height': int_or_none(mobj.group('height')),
222 'width': int_or_none(mobj.group('width')),
223 'thumbnail': mobj.group('poster'),
227 raise ExtractorError('There\'s not video in this tweet.')