_ Git - youtube-dl/blob - youtube_dl/extractor/infoq.py

   1 from __future__ import unicode_literals
   2
   3 import base64
   4 import re
   5
   6 from .common import InfoExtractor
   7 from ..utils import (
   8     compat_urllib_parse,
   9 )
  10
  11
  12 class InfoQIE(InfoExtractor):
  13     _VALID_URL = r'https?://(?:www\.)?infoq\.com/[^/]+/(?P<id>[^/]+)$'
  14
  15     _TEST = {
  16         'name': 'InfoQ',
  17         'url': 'http://www.infoq.com/presentations/A-Few-of-My-Favorite-Python-Things',
  18         'md5': 'fcaa3d995e04080dcb9465d86b5eef62',
  19         'info_dict': {
  20             'id': '12-jan-pythonthings',
  21             'ext': 'mp4',
  22             'description': 'Mike Pirnat presents some tips and tricks, standard libraries and third party packages that make programming in Python a richer experience.',
  23             'title': 'A Few of My Favorite [Python] Things',
  24         },
  25     }
  26
  27     def _real_extract(self, url):
  28         mobj = re.match(self._VALID_URL, url)
  29         video_id = mobj.group('id')
  30
  31         webpage = self._download_webpage(url, video_id)
  32
  33         video_title = self._html_search_regex(r'<title>(.*?)</title>', webpage, 'title')
  34         video_description = self._html_search_meta('description', webpage, 'description')
  35
  36         # The server URL is hardcoded
  37         video_url = 'rtmpe://video.infoq.com/cfx/st/'
  38
  39         # Extract video URL
  40         encoded_id = self._search_regex(r"jsclassref ?= ?'([^']*)'", webpage, 'encoded id')
  41         real_id = compat_urllib_parse.unquote(base64.b64decode(encoded_id.encode('ascii')).decode('utf-8'))
  42         playpath = 'mp4:' + real_id
  43
  44         video_filename = playpath.split('/')[-1]
  45         video_id, extension = video_filename.split('.')
  46
  47         return {
  48             'id': video_id,
  49             'title': video_title,
  50             'description': video_description,
  51             'formats': [{
  52                 'url': video_url,
  53                 'ext': extension,
  54                 'play_path': playpath,
  55             }],
  56         }