Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/extractor/videomega.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import re
   5
   6 from .common import InfoExtractor
   7 from ..compat import compat_urllib_request
   8
   9
  10 class VideoMegaIE(InfoExtractor):
  11     _VALID_URL = r'(?:videomega:|https?://(?:www\.)?videomega\.tv/(?:(?:view|iframe|cdn)\.php)?\?ref=)(?P<id>[A-Za-z0-9]+)'
  12     _TESTS = [{
  13         'url': 'http://videomega.tv/cdn.php?ref=AOSQBJYKIDDIKYJBQSOA',
  14         'md5': 'cc1920a58add3f05c6a93285b84fb3aa',
  15         'info_dict': {
  16             'id': 'AOSQBJYKIDDIKYJBQSOA',
  17             'ext': 'mp4',
  18             'title': '1254207',
  19             'thumbnail': 're:^https?://.*\.jpg$',
  20         }
  21     }, {
  22         'url': 'http://videomega.tv/cdn.php?ref=AOSQBJYKIDDIKYJBQSOA&width=1070&height=600',
  23         'only_matching': True,
  24     }, {
  25         'url': 'http://videomega.tv/view.php?ref=090051111052065112106089103052052103089106112065052111051090',
  26         'only_matching': True,
  27     }]
  28
  29     def _real_extract(self, url):
  30         video_id = self._match_id(url)
  31
  32         iframe_url = 'http://videomega.tv/cdn.php?ref=%s' % video_id
  33         req = compat_urllib_request.Request(iframe_url)
  34         req.add_header('Referer', url)
  35         req.add_header('Cookie', 'noadvtday=0')
  36         webpage = self._download_webpage(req, video_id)
  37
  38         title = self._html_search_regex(
  39             r'<title>(.+?)</title>', webpage, 'title')
  40         title = re.sub(
  41             r'(?:^[Vv]ideo[Mm]ega\.tv\s-\s*|\s*-\svideomega\.tv$)', '', title)
  42         thumbnail = self._search_regex(
  43             r'<video[^>]+?poster="([^"]+)"', webpage, 'thumbnail', fatal=False)
  44         video_url = self._search_regex(
  45             r'<source[^>]+?src="([^"]+)"', webpage, 'video URL')
  46
  47         return {
  48             'id': video_id,
  49             'title': title,
  50             'url': video_url,
  51             'thumbnail': thumbnail,
  52             'http_headers': {
  53                 'Referer': iframe_url,
  54             },
  55         }