[twitter] Add support for user_id-less URLs (closes #14270)

This commit is contained in:
Giuseppe Fabiano 2017-09-23 01:38:09 +02:00 committed by Sergey M
parent 4bb58fa118
commit 5c1452e8f1

View File

@ -242,8 +242,9 @@ def _real_extract(self, url):
class TwitterIE(InfoExtractor): class TwitterIE(InfoExtractor):
IE_NAME = 'twitter' IE_NAME = 'twitter'
_VALID_URL = r'https?://(?:www\.|m\.|mobile\.)?twitter\.com/(?P<user_id>[^/]+)/status/(?P<id>\d+)' _VALID_URL = r'https?://(?:www\.|m\.|mobile\.)?twitter\.com/(?:i/web|(?P<user_id>[^/]+))/status/(?P<id>\d+)'
_TEMPLATE_URL = 'https://twitter.com/%s/status/%s' _TEMPLATE_URL = 'https://twitter.com/%s/status/%s'
_TEMPLATE_STATUSES_URL = 'https://twitter.com/statuses/%s'
_TESTS = [{ _TESTS = [{
'url': 'https://twitter.com/freethenipple/status/643211948184596480', 'url': 'https://twitter.com/freethenipple/status/643211948184596480',
@ -322,9 +323,9 @@ class TwitterIE(InfoExtractor):
'info_dict': { 'info_dict': {
'id': 'MIOxnrUteUd', 'id': 'MIOxnrUteUd',
'ext': 'mp4', 'ext': 'mp4',
'title': 'FilmDrunk - Vine of the day', 'title': 'Vince Mancini - Vine of the day',
'description': 'FilmDrunk on Twitter: "Vine of the day https://t.co/xmTvRdqxWf"', 'description': 'Vince Mancini on Twitter: "Vine of the day https://t.co/xmTvRdqxWf"',
'uploader': 'FilmDrunk', 'uploader': 'Vince Mancini',
'uploader_id': 'Filmdrunk', 'uploader_id': 'Filmdrunk',
'timestamp': 1402826626, 'timestamp': 1402826626,
'upload_date': '20140615', 'upload_date': '20140615',
@ -372,6 +373,21 @@ class TwitterIE(InfoExtractor):
'params': { 'params': {
'format': 'best[format_id^=http-]', 'format': 'best[format_id^=http-]',
}, },
}, {
'url': 'https://twitter.com/i/web/status/910031516746514432',
'info_dict': {
'id': '910031516746514432',
'ext': 'mp4',
'title': 'Préfet de Guadeloupe - [Direct] #Maria Le centre se trouve actuellement au sud de Basse-Terre. Restez confinés. Réfugiez-vous dans la pièce la + sûre.',
'thumbnail': r're:^https?://.*\.jpg',
'description': 'Préfet de Guadeloupe on Twitter: "[Direct] #Maria Le centre se trouve actuellement au sud de Basse-Terre. Restez confinés. Réfugiez-vous dans la pièce la + sûre. https://t.co/mwx01Rs4lo"',
'uploader': 'Préfet de Guadeloupe',
'uploader_id': 'Prefet971',
'duration': 47.48,
},
'params': {
'skip_download': True, # requires ffmpeg
},
}] }]
def _real_extract(self, url): def _real_extract(self, url):
@ -380,11 +396,15 @@ def _real_extract(self, url):
twid = mobj.group('id') twid = mobj.group('id')
webpage, urlh = self._download_webpage_handle( webpage, urlh = self._download_webpage_handle(
self._TEMPLATE_URL % (user_id, twid), twid) self._TEMPLATE_STATUSES_URL % twid, twid)
if 'twitter.com/account/suspended' in urlh.geturl(): if 'twitter.com/account/suspended' in urlh.geturl():
raise ExtractorError('Account suspended by Twitter.', expected=True) raise ExtractorError('Account suspended by Twitter.', expected=True)
if user_id is None:
mobj = re.match(self._VALID_URL, urlh.geturl())
user_id = mobj.group('user_id')
username = remove_end(self._og_search_title(webpage), ' on Twitter') username = remove_end(self._og_search_title(webpage), ' on Twitter')
title = description = self._og_search_description(webpage).strip('').replace('\n', ' ').strip('“”') title = description = self._og_search_description(webpage).strip('').replace('\n', ' ').strip('“”')