2014-02-10 20:24:12 +01:00
|
|
|
from __future__ import unicode_literals
|
|
|
|
|
2013-07-01 21:08:54 +02:00
|
|
|
import re
|
|
|
|
|
|
|
|
from .common import InfoExtractor
|
|
|
|
|
2014-02-10 20:24:12 +01:00
|
|
|
|
2013-07-01 21:08:54 +02:00
|
|
|
class InstagramIE(InfoExtractor):
|
2014-02-10 20:24:12 +01:00
|
|
|
_VALID_URL = r'http://instagram\.com/p/(?P<id>.*?)/'
|
2013-07-01 21:08:54 +02:00
|
|
|
_TEST = {
|
2014-02-10 20:24:12 +01:00
|
|
|
'url': 'http://instagram.com/p/aye83DjauH/?foo=bar#abc',
|
|
|
|
'md5': '0d2da106a9d2631273e192b372806516',
|
|
|
|
'info_dict': {
|
|
|
|
'id': 'aye83DjauH',
|
|
|
|
'ext': 'mp4',
|
|
|
|
'uploader_id': 'naomipq',
|
|
|
|
'title': 'Video by naomipq',
|
|
|
|
'description': 'md5:1f17f0ab29bd6fe2bfad705f58de3cb8',
|
2013-07-01 21:08:54 +02:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
def _real_extract(self, url):
|
|
|
|
mobj = re.match(self._VALID_URL, url)
|
2014-02-10 20:24:12 +01:00
|
|
|
video_id = mobj.group('id')
|
2013-07-01 21:08:54 +02:00
|
|
|
webpage = self._download_webpage(url, video_id)
|
2013-07-18 13:12:27 +02:00
|
|
|
uploader_id = self._search_regex(r'"owner":{"username":"(.+?)"',
|
2014-02-10 20:24:12 +01:00
|
|
|
webpage, 'uploader id', fatal=False)
|
|
|
|
desc = self._search_regex(r'"caption":"(.*?)"', webpage, 'description',
|
2013-07-18 13:12:27 +02:00
|
|
|
fatal=False)
|
2013-07-01 21:08:54 +02:00
|
|
|
|
2014-02-10 20:24:12 +01:00
|
|
|
return {
|
|
|
|
'id': video_id,
|
|
|
|
'url': self._og_search_video_url(webpage, secure=False),
|
|
|
|
'ext': 'mp4',
|
|
|
|
'title': 'Video by %s' % uploader_id,
|
2013-07-12 19:00:19 +02:00
|
|
|
'thumbnail': self._og_search_thumbnail(webpage),
|
2014-02-10 20:24:12 +01:00
|
|
|
'uploader_id': uploader_id,
|
2013-07-18 13:12:27 +02:00
|
|
|
'description': desc,
|
2014-02-10 20:24:12 +01:00
|
|
|
}
|