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

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3 import re
   4 import datetime
   5
   6 from .common import InfoExtractor
   7
   8
   9 class TeleTaskIE(InfoExtractor):
  10     _VALID_URL = r'http?://(?:www\.)?tele-task\.de/archive/video/html5/(?P<id>[0-9]+)/'
  11     _TEST = {
  12         'url': 'http://www.tele-task.de/archive/video/html5/26168/',
  13         'info_dict': {
  14             'title': 'Duplicate Detection',
  15         },
  16         'playlist': [{
  17             'md5': '290ef69fb2792e481169c3958dbfbd57',
  18             'info_dict': {
  19                 'title': 'Duplicate Detection',
  20                 'upload_date': '20141218',
  21                 'id': 'speaker_26168',
  22                 'ext': 'mp4',
  23             }
  24         },
  25             {
  26             'md5': 'e1e7218c5f0e4790015a437fcf6c71b4',
  27             'info_dict': {
  28                 'title': 'Duplicate Detection',
  29                 'upload_date': '20141218',
  30                 'id': 'slides_26168',
  31                 'ext': 'mp4',
  32             }
  33         }]
  34     }
  35
  36     def _real_extract(self, url):
  37         lecture_id = self._match_id(url)
  38         webpage = self._download_webpage(url, lecture_id)
  39
  40         title = self._html_search_regex(
  41             r'itemprop="name">([^"]+)</a>', webpage, 'title')
  42         url_speaker = self._html_search_regex(
  43             r'class="speaker".*?src="([^"]+)"', webpage, 'video_url_speaker', flags=re.DOTALL)
  44         url_slides = self._html_search_regex(
  45             r'class="slides".*?src="([^"]+)"', webpage, 'video_url_slides', flags=re.DOTALL)
  46         date = self._html_search_regex(
  47             r'<td class="label">Date:</td><td>([^"]+)</td>', webpage, 'date')
  48         date = datetime.datetime.strptime(date, '%d.%m.%Y').strftime('%Y%m%d')
  49
  50         entries = [{
  51             'title': title,
  52             'upload_date': date,
  53             'id': "speaker_"+lecture_id,
  54             'url': url_speaker,
  55         },
  56             {
  57             'title': title,
  58             'upload_date': date,
  59             'id': "slides_"+lecture_id,
  60             'url': url_slides}]
  61
  62         return {
  63             '_type': "playlist",
  64             'id': lecture_id,
  65             'title': title,
  66             'entries': entries,
  67         }