|
|
@ -1,21 +1,11 @@
|
|
|
|
# coding: utf-8
|
|
|
|
# coding: utf-8
|
|
|
|
from __future__ import unicode_literals
|
|
|
|
from __future__ import unicode_literals
|
|
|
|
|
|
|
|
|
|
|
|
import re
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
from .common import InfoExtractor
|
|
|
|
from .common import InfoExtractor
|
|
|
|
from ..utils import (
|
|
|
|
|
|
|
|
int_or_none,
|
|
|
|
|
|
|
|
parse_iso8601,
|
|
|
|
|
|
|
|
xpath_with_ns,
|
|
|
|
|
|
|
|
xpath_text,
|
|
|
|
|
|
|
|
find_xpath_attr,
|
|
|
|
|
|
|
|
)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
class AftenpostenIE(InfoExtractor):
|
|
|
|
class AftenpostenIE(InfoExtractor):
|
|
|
|
_VALID_URL = r'https?://(?:www\.)?aftenposten\.no/webtv/(?:#!/)?video/(?P<id>\d+)'
|
|
|
|
_VALID_URL = r'https?://(?:www\.)?aftenposten\.no/webtv/(?:#!/)?video/(?P<id>\d+)'
|
|
|
|
|
|
|
|
|
|
|
|
_TEST = {
|
|
|
|
_TEST = {
|
|
|
|
'url': 'http://www.aftenposten.no/webtv/#!/video/21039/trailer-sweatshop-i-can-t-take-any-more',
|
|
|
|
'url': 'http://www.aftenposten.no/webtv/#!/video/21039/trailer-sweatshop-i-can-t-take-any-more',
|
|
|
|
'md5': 'fd828cd29774a729bf4d4425fe192972',
|
|
|
|
'md5': 'fd828cd29774a729bf4d4425fe192972',
|
|
|
@ -30,69 +20,4 @@ class AftenpostenIE(InfoExtractor):
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
def _real_extract(self, url):
|
|
|
|
def _real_extract(self, url):
|
|
|
|
video_id = self._match_id(url)
|
|
|
|
return self.url_result('xstream:ap:%s' % self._match_id(url), 'Xstream')
|
|
|
|
|
|
|
|
|
|
|
|
data = self._download_xml(
|
|
|
|
|
|
|
|
'http://frontend.xstream.dk/ap/feed/video/?platform=web&id=%s' % video_id, video_id)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
NS_MAP = {
|
|
|
|
|
|
|
|
'atom': 'http://www.w3.org/2005/Atom',
|
|
|
|
|
|
|
|
'xt': 'http://xstream.dk/',
|
|
|
|
|
|
|
|
'media': 'http://search.yahoo.com/mrss/',
|
|
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
entry = data.find(xpath_with_ns('./atom:entry', NS_MAP))
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
title = xpath_text(
|
|
|
|
|
|
|
|
entry, xpath_with_ns('./atom:title', NS_MAP), 'title')
|
|
|
|
|
|
|
|
description = xpath_text(
|
|
|
|
|
|
|
|
entry, xpath_with_ns('./atom:summary', NS_MAP), 'description')
|
|
|
|
|
|
|
|
timestamp = parse_iso8601(xpath_text(
|
|
|
|
|
|
|
|
entry, xpath_with_ns('./atom:published', NS_MAP), 'upload date'))
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
formats = []
|
|
|
|
|
|
|
|
media_group = entry.find(xpath_with_ns('./media:group', NS_MAP))
|
|
|
|
|
|
|
|
for media_content in media_group.findall(xpath_with_ns('./media:content', NS_MAP)):
|
|
|
|
|
|
|
|
media_url = media_content.get('url')
|
|
|
|
|
|
|
|
if not media_url:
|
|
|
|
|
|
|
|
continue
|
|
|
|
|
|
|
|
tbr = int_or_none(media_content.get('bitrate'))
|
|
|
|
|
|
|
|
mobj = re.search(r'^(?P<url>rtmp://[^/]+/(?P<app>[^/]+))/(?P<playpath>.+)$', media_url)
|
|
|
|
|
|
|
|
if mobj:
|
|
|
|
|
|
|
|
formats.append({
|
|
|
|
|
|
|
|
'url': mobj.group('url'),
|
|
|
|
|
|
|
|
'play_path': 'mp4:%s' % mobj.group('playpath'),
|
|
|
|
|
|
|
|
'app': mobj.group('app'),
|
|
|
|
|
|
|
|
'ext': 'flv',
|
|
|
|
|
|
|
|
'tbr': tbr,
|
|
|
|
|
|
|
|
'format_id': 'rtmp-%d' % tbr,
|
|
|
|
|
|
|
|
})
|
|
|
|
|
|
|
|
else:
|
|
|
|
|
|
|
|
formats.append({
|
|
|
|
|
|
|
|
'url': media_url,
|
|
|
|
|
|
|
|
'tbr': tbr,
|
|
|
|
|
|
|
|
})
|
|
|
|
|
|
|
|
self._sort_formats(formats)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
link = find_xpath_attr(
|
|
|
|
|
|
|
|
entry, xpath_with_ns('./atom:link', NS_MAP), 'rel', 'original')
|
|
|
|
|
|
|
|
if link is not None:
|
|
|
|
|
|
|
|
formats.append({
|
|
|
|
|
|
|
|
'url': link.get('href'),
|
|
|
|
|
|
|
|
'format_id': link.get('rel'),
|
|
|
|
|
|
|
|
})
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
thumbnails = [{
|
|
|
|
|
|
|
|
'url': splash.get('url'),
|
|
|
|
|
|
|
|
'width': int_or_none(splash.get('width')),
|
|
|
|
|
|
|
|
'height': int_or_none(splash.get('height')),
|
|
|
|
|
|
|
|
} for splash in media_group.findall(xpath_with_ns('./xt:splash', NS_MAP))]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
return {
|
|
|
|
|
|
|
|
'id': video_id,
|
|
|
|
|
|
|
|
'title': title,
|
|
|
|
|
|
|
|
'description': description,
|
|
|
|
|
|
|
|
'timestamp': timestamp,
|
|
|
|
|
|
|
|
'formats': formats,
|
|
|
|
|
|
|
|
'thumbnails': thumbnails,
|
|
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|