aboutsummaryrefslogtreecommitdiff
path: root/youtube_dl/extractor
diff options
context:
space:
mode:
authorSergey M․ <dstftw@gmail.com>2018-01-17 22:03:56 +0700
committerSergey M․ <dstftw@gmail.com>2018-01-17 22:03:56 +0700
commit4471affc348af40409188f133786780edd969623 (patch)
tree1e69768778f3e3402187731d58b4c366c128039e /youtube_dl/extractor
parent1370dba59f140dc5313ed1da1dad7bcba09cfb1c (diff)
[spiegel] Add support for nexx videos (closes #15285)
Diffstat (limited to 'youtube_dl/extractor')
-rw-r--r--youtube_dl/extractor/spiegel.py17
1 files changed, 16 insertions, 1 deletions
diff --git a/youtube_dl/extractor/spiegel.py b/youtube_dl/extractor/spiegel.py
index 84298fee4..fc995e8c1 100644
--- a/youtube_dl/extractor/spiegel.py
+++ b/youtube_dl/extractor/spiegel.py
@@ -4,7 +4,10 @@ from __future__ import unicode_literals
import re
from .common import InfoExtractor
-from .nexx import NexxEmbedIE
+from .nexx import (
+ NexxIE,
+ NexxEmbedIE,
+)
from .spiegeltv import SpiegeltvIE
from ..compat import compat_urlparse
from ..utils import (
@@ -51,6 +54,10 @@ class SpiegelIE(InfoExtractor):
}, {
'url': 'http://www.spiegel.de/video/astronaut-alexander-gerst-von-der-iss-station-beantwortet-fragen-video-1519126-iframe.html',
'only_matching': True,
+ }, {
+ # nexx video
+ 'url': 'http://www.spiegel.de/video/spiegel-tv-magazin-ueber-guellekrise-in-schleswig-holstein-video-99012776.html',
+ 'only_matching': True,
}]
def _real_extract(self, url):
@@ -61,6 +68,14 @@ class SpiegelIE(InfoExtractor):
if SpiegeltvIE.suitable(handle.geturl()):
return self.url_result(handle.geturl(), 'Spiegeltv')
+ nexx_id = self._search_regex(
+ r'nexxOmniaId\s*:\s*(\d+)', webpage, 'nexx id', default=None)
+ if nexx_id:
+ domain_id = NexxIE._extract_domain_id(webpage) or '748'
+ return self.url_result(
+ 'nexx:%s:%s' % (domain_id, nexx_id), ie=NexxIE.ie_key(),
+ video_id=nexx_id)
+
video_data = extract_attributes(self._search_regex(r'(<div[^>]+id="spVideoElements"[^>]+>)', webpage, 'video element', default=''))
title = video_data.get('data-video-title') or get_element_by_attribute('class', 'module-title', webpage)