[wimp] Detect youtube videos (fixes #2686)

This commit is contained in:
Jaime Marquínez Ferrándiz 2014-04-03 20:44:51 +02:00
parent f34e6a2cd6
commit 983af2600f

View file

@ -3,11 +3,12 @@ from __future__ import unicode_literals
import re import re
from .common import InfoExtractor from .common import InfoExtractor
from .youtube import YoutubeIE
class WimpIE(InfoExtractor): class WimpIE(InfoExtractor):
_VALID_URL = r'http://(?:www\.)?wimp\.com/([^/]+)/' _VALID_URL = r'http://(?:www\.)?wimp\.com/([^/]+)/'
_TEST = { _TESTS = [{
'url': 'http://www.wimp.com/maruexhausted/', 'url': 'http://www.wimp.com/maruexhausted/',
'md5': 'f1acced123ecb28d9bb79f2479f2b6a1', 'md5': 'f1acced123ecb28d9bb79f2479f2b6a1',
'info_dict': { 'info_dict': {
@ -16,7 +17,20 @@ class WimpIE(InfoExtractor):
'title': 'Maru is exhausted.', 'title': 'Maru is exhausted.',
'description': 'md5:57e099e857c0a4ea312542b684a869b8', 'description': 'md5:57e099e857c0a4ea312542b684a869b8',
} }
} }, {
# youtube video
'url': 'http://www.wimp.com/clowncar/',
'info_dict': {
'id': 'cG4CEr2aiSg',
'ext': 'mp4',
'title': 'Basset hound clown car...incredible!',
'description': 'md5:8d228485e0719898c017203f900b3a35',
'uploader': 'Gretchen Hoey',
'uploader_id': 'gretchenandjeff1',
'upload_date': '20140303',
},
'add_ie': ['Youtube'],
}]
def _real_extract(self, url): def _real_extract(self, url):
mobj = re.match(self._VALID_URL, url) mobj = re.match(self._VALID_URL, url)
@ -24,6 +38,13 @@ class WimpIE(InfoExtractor):
webpage = self._download_webpage(url, video_id) webpage = self._download_webpage(url, video_id)
video_url = self._search_regex( video_url = self._search_regex(
r's1\.addVariable\("file",\s*"([^"]+)"\);', webpage, 'video URL') r's1\.addVariable\("file",\s*"([^"]+)"\);', webpage, 'video URL')
if YoutubeIE.suitable(video_url):
self.to_screen('Found YouTube video')
return {
'_type': 'url',
'url': video_url,
'ie_key': YoutubeIE.ie_key(),
}
return { return {
'id': video_id, 'id': video_id,
@ -31,4 +52,4 @@ class WimpIE(InfoExtractor):
'title': self._og_search_title(webpage), 'title': self._og_search_title(webpage),
'thumbnail': self._og_search_thumbnail(webpage), 'thumbnail': self._og_search_thumbnail(webpage),
'description': self._og_search_description(webpage), 'description': self._og_search_description(webpage),
} }