"file": "0732f586d7.mp4",
"md5": "f647e9e90064b53b6e046e75d0241fbd"
},
- {
- "name": "TweetReel",
- "url": "http://tweetreel.com/?77smq",
- "file": "77smq.mov",
- "md5": "56b4d9ca9de467920f3f99a6d91255d6",
- "info_dict": {
- "uploader": "itszero",
- "uploader_id": "itszero",
- "upload_date": "20091225",
- "description": "Installing Gentoo Linux on Powerbook G4, it turns out the sleep indicator becomes HDD activity indicator :D"
- }
- },
{
"name": "Steam",
"url": "http://store.steampowered.com/video/105600/",
}
return [info]
-class TweetReelIE(InfoExtractor):
- _VALID_URL = r'^(?:https?://)?(?:www\.)?tweetreel\.com/[?](?P<id>[0-9a-z]+)$'
-
- def _real_extract(self, url):
- mobj = re.match(self._VALID_URL, url)
- if mobj is None:
- self._downloader.trouble(u'ERROR: invalid URL: %s' % url)
- return
-
- video_id = mobj.group('id')
- webpage = self._download_webpage(url, video_id)
-
- m = re.search(r'<div id="left" status_id="([0-9]+)">', webpage)
- if not m:
- self._downloader.trouble(u'ERROR: Cannot find status ID')
- status_id = m.group(1)
-
- m = re.search(r'<div class="tweet_text">(.*?)</div>', webpage, flags=re.DOTALL)
- if not m:
- self._downloader.trouble(u'WARNING: Cannot find description')
- desc = unescapeHTML(re.sub('<a.*?</a>', '', m.group(1))).strip()
-
- m = re.search(r'<div class="tweet_info">.*?from <a target="_blank" href="https?://twitter.com/(?P<uploader_id>.+?)">(?P<uploader>.+?)</a>', webpage, flags=re.DOTALL)
- if not m:
- self._downloader.trouble(u'ERROR: Cannot find uploader')
- uploader = unescapeHTML(m.group('uploader'))
- uploader_id = unescapeHTML(m.group('uploader_id'))
-
- m = re.search(r'<span unixtime="([0-9]+)"', webpage)
- if not m:
- self._downloader.trouble(u'ERROR: Cannot find upload date')
- upload_date = datetime.datetime.fromtimestamp(int(m.group(1))).strftime('%Y%m%d')
-
- title = desc
- video_url = 'http://files.tweetreel.com/video/' + status_id + '.mov'
-
- info = {
- 'id': video_id,
- 'url': video_url,
- 'ext': 'mov',
- 'title': title,
- 'description': desc,
- 'uploader': uploader,
- 'uploader_id': uploader_id,
- 'internal_id': status_id,
- 'upload_date': upload_date
- }
- return [info]
-
class SteamIE(InfoExtractor):
_VALID_URL = r"""http://store.steampowered.com/
(?P<urltype>video|app)/ #If the page is only for videos or for a game
NBAIE(),
JustinTVIE(),
FunnyOrDieIE(),
- TweetReelIE(),
SteamIE(),
UstreamIE(),
RBMARadioIE(),