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

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import base64
   5
   6 from .common import InfoExtractor
   7 from ..compat import compat_urllib_parse_unquote
   8 from ..utils import (
   9     int_or_none,
  10 )
  11
  12
  13 class MangomoloBaseIE(InfoExtractor):
  14     def _get_real_id(self, page_id):
  15         return page_id
  16
  17     def _real_extract(self, url):
  18         page_id = self._get_real_id(self._match_id(url))
  19         webpage = self._download_webpage(url, page_id)
  20         hidden_inputs = self._hidden_inputs(webpage)
  21         m3u8_entry_protocol = 'm3u8' if self._IS_LIVE else 'm3u8_native'
  22
  23         format_url = self._html_search_regex(
  24             [
  25                 r'file\s*:\s*"(https?://[^"]+?/playlist.m3u8)',
  26                 r'<a[^>]+href="(rtsp://[^"]+)"'
  27             ], webpage, 'format url')
  28         formats = self._extract_wowza_formats(
  29             format_url, page_id, m3u8_entry_protocol, ['smil'])
  30         self._sort_formats(formats)
  31
  32         return {
  33             'id': page_id,
  34             'title': self._live_title(page_id) if self._IS_LIVE else page_id,
  35             'uploader_id': hidden_inputs.get('userid'),
  36             'duration': int_or_none(hidden_inputs.get('duration')),
  37             'is_live': self._IS_LIVE,
  38             'formats': formats,
  39         }
  40
  41
  42 class MangomoloVideoIE(MangomoloBaseIE):
  43     IE_NAME = 'mangomolo:video'
  44     _VALID_URL = r'https?://admin\.mangomolo\.com/analytics/index\.php/customers/embed/video\?.*?\bid=(?P<id>\d+)'
  45     _IS_LIVE = False
  46
  47
  48 class MangomoloLiveIE(MangomoloBaseIE):
  49     IE_NAME = 'mangomolo:live'
  50     _VALID_URL = r'https?://admin\.mangomolo\.com/analytics/index\.php/customers/embed/index\?.*?\bchannelid=(?P<id>(?:[A-Za-z0-9+/=]|%2B|%2F|%3D)+)'
  51     _IS_LIVE = True
  52
  53     def _get_real_id(self, page_id):
  54         return base64.b64decode(compat_urllib_parse_unquote(page_id).encode()).decode()