[gazeta] Add new extractor (Closes #4222)

This commit is contained in:
Naglis Jonaitis 2015-03-08 13:54:01 +02:00
parent d34e79492d
commit 1132eae56d
2 changed files with 36 additions and 0 deletions

View file

@ -175,6 +175,7 @@ from .gameone import (
from .gamespot import GameSpotIE from .gamespot import GameSpotIE
from .gamestar import GameStarIE from .gamestar import GameStarIE
from .gametrailers import GametrailersIE from .gametrailers import GametrailersIE
from .gazeta import GazetaIE
from .gdcvault import GDCVaultIE from .gdcvault import GDCVaultIE
from .generic import GenericIE from .generic import GenericIE
from .giantbomb import GiantBombIE from .giantbomb import GiantBombIE

View file

@ -0,0 +1,35 @@
# coding: utf-8
from __future__ import unicode_literals
import re
from .common import InfoExtractor
class GazetaIE(InfoExtractor):
_VALID_URL = r'(?P<url>https?://(?:www\.)?gazeta\.ru/(?:(?P<category>[^/]*)/)?video/(?:main/)?(?P<id>[A-Za-z0-9-_]+)\.s?html)'
_TEST = {
'url': 'http://www.gazeta.ru/video/main/zadaite_vopros_vladislavu_yurevichu.shtml',
'md5': 'd49c9bdc6e5a7888f27475dc215ee789',
'info_dict': {
'id': '205566',
'ext': 'mp4',
'title': '«7080 процентов гражданских в Донецке на грани голода»',
'description': 'md5:38617526050bd17b234728e7f9620a71',
'thumbnail': 're:^https?://.*\.jpg',
},
}
def _real_extract(self, url):
mobj = re.match(self._VALID_URL, url)
display_id = mobj.group('id')
embed_url = '%s?p=embed' % mobj.group('url')
embed_page = self._download_webpage(
embed_url, display_id, 'Downloading embed page')
video_id = self._search_regex(
r'<div[^>]*?class="eagleplayer"[^>]*?data-id="([^"]+)"', embed_page, 'video id')
return self.url_result(
'eagleplatform:gazeta.media.eagleplatform.com:%s' % video_id, 'EaglePlatform')