aboutsummaryrefslogtreecommitdiff
diff options
context:
space:
mode:
-rw-r--r--youtube_dl/extractor/reddit.py9
1 files changed, 7 insertions, 2 deletions
diff --git a/youtube_dl/extractor/reddit.py b/youtube_dl/extractor/reddit.py
index 4d44b9d74..f36bc648c 100644
--- a/youtube_dl/extractor/reddit.py
+++ b/youtube_dl/extractor/reddit.py
@@ -1,5 +1,7 @@
from __future__ import unicode_literals
+import re
+
from .common import InfoExtractor
from ..utils import (
ExtractorError,
@@ -45,7 +47,7 @@ class RedditIE(InfoExtractor):
class RedditRIE(InfoExtractor):
- _VALID_URL = r'https?://(?:www\.)?reddit\.com/r/[^/]+/comments/(?P<id>[^/]+)'
+ _VALID_URL = r'(?P<url>https?://(?:www\.)?reddit\.com/r/[^/]+/comments/(?P<id>[^/?#&]+))'
_TESTS = [{
'url': 'https://www.reddit.com/r/videos/comments/6rrwyj/that_small_heart_attack/',
'info_dict': {
@@ -83,10 +85,13 @@ class RedditRIE(InfoExtractor):
}]
def _real_extract(self, url):
+ mobj = re.match(self._VALID_URL, url)
+ url, video_id = mobj.group('url', 'id')
+
video_id = self._match_id(url)
data = self._download_json(
- url + '.json', video_id)[0]['data']['children'][0]['data']
+ url + '/.json', video_id)[0]['data']['children'][0]['data']
video_url = data['url']