mirror of
https://codeberg.org/polarisfm/youtube-dl
synced 2024-11-24 01:14:32 +01:00
47 lines
1.5 KiB
Python
47 lines
1.5 KiB
Python
from __future__ import unicode_literals
|
|
|
|
import re
|
|
import random
|
|
|
|
from .common import InfoExtractor
|
|
|
|
|
|
class VideoPremiumIE(InfoExtractor):
|
|
_VALID_URL = r'https?://(?:www\.)?videopremium\.(?:tv|me)/(?P<id>\w+)(?:/.*)?'
|
|
_TEST = {
|
|
'url': 'http://videopremium.tv/4w7oadjsf156',
|
|
'info_dict': {
|
|
'id': '4w7oadjsf156',
|
|
'ext': 'f4v',
|
|
'title': 'youtube-dl_test_video____a_________-BaW_jenozKc.mp4.mp4'
|
|
},
|
|
'params': {
|
|
'skip_download': True,
|
|
},
|
|
'skip': 'Test file has been deleted.',
|
|
}
|
|
|
|
def _real_extract(self, url):
|
|
video_id = self._match_id(url)
|
|
webpage_url = 'http://videopremium.tv/' + video_id
|
|
webpage = self._download_webpage(webpage_url, video_id)
|
|
|
|
if re.match(r"^<html><head><script[^>]*>window.location\s*=", webpage):
|
|
# Download again, we need a cookie
|
|
webpage = self._download_webpage(
|
|
webpage_url, video_id,
|
|
note='Downloading webpage again (with cookie)')
|
|
|
|
video_title = self._html_search_regex(
|
|
r'<h2(?:.*?)>\s*(.+?)\s*<', webpage, 'video title')
|
|
|
|
return {
|
|
'id': video_id,
|
|
'url': "rtmp://e%d.md.iplay.md/play" % random.randint(1, 16),
|
|
'play_path': "mp4:%s.f4v" % video_id,
|
|
'page_url': "http://videopremium.tv/" + video_id,
|
|
'player_url': "http://videopremium.tv/uplayer/uppod.swf",
|
|
'ext': 'f4v',
|
|
'title': video_title,
|
|
}
|