[ceskateleveize:porady] Add extractor (closes #7411, closes #12645)

This commit is contained in:
Sergey M․ 2017-04-08 19:42:09 +07:00
parent e18f1da97a
commit 28b674ca23
No known key found for this signature in database
GPG key ID: 2C393E0F18A9236D
2 changed files with 49 additions and 1 deletions

View file

@ -12,6 +12,7 @@ from ..utils import (
ExtractorError, ExtractorError,
float_or_none, float_or_none,
sanitized_Request, sanitized_Request,
unescapeHTML,
urlencode_postdata, urlencode_postdata,
USER_AGENTS, USER_AGENTS,
) )
@ -232,3 +233,47 @@ class CeskaTelevizeIE(InfoExtractor):
yield line yield line
return '\r\n'.join(_fix_subtitle(subtitles)) return '\r\n'.join(_fix_subtitle(subtitles))
class CeskaTelevizePoradyIE(InfoExtractor):
_VALID_URL = r'https?://(?:www\.)?ceskatelevize\.cz/porady/(?:[^/?#&]+/)*(?P<id>[^/#?]+)'
_TESTS = [{
# video with 18+ caution trailer
'url': 'http://www.ceskatelevize.cz/porady/10520528904-queer/215562210900007-bogotart/',
'info_dict': {
'id': '215562210900007-bogotart',
'title': 'Queer: Bogotart',
'description': 'Alternativní průvodce současným queer světem',
},
'playlist': [{
'info_dict': {
'id': '61924494876844842',
'ext': 'mp4',
'title': 'Queer: Bogotart (Varování 18+)',
'duration': 10.2,
},
}, {
'info_dict': {
'id': '61924494877068022',
'ext': 'mp4',
'title': 'Queer: Bogotart (Queer)',
'thumbnail': r're:^https?://.*\.jpg',
'duration': 1558.3,
},
}],
'params': {
# m3u8 download
'skip_download': True,
},
}]
def _real_extract(self, url):
video_id = self._match_id(url)
webpage = self._download_webpage(url, video_id)
data_url = unescapeHTML(self._search_regex(
r'<span[^>]*\bdata-url=(["\'])(?P<url>(?:(?!\1).)+)\1',
webpage, 'iframe player url', group='url'))
return self.url_result(data_url, ie=CeskaTelevizeIE.ie_key())

View file

@ -165,7 +165,10 @@ from .ccc import CCCIE
from .ccma import CCMAIE from .ccma import CCMAIE
from .cctv import CCTVIE from .cctv import CCTVIE
from .cda import CDAIE from .cda import CDAIE
from .ceskatelevize import CeskaTelevizeIE from .ceskatelevize import (
CeskaTelevizeIE,
CeskaTelevizePoradyIE,
)
from .channel9 import Channel9IE from .channel9 import Channel9IE
from .charlierose import CharlieRoseIE from .charlierose import CharlieRoseIE
from .chaturbate import ChaturbateIE from .chaturbate import ChaturbateIE