yt_dlp/extractor/screencastomatic.py

   1 from .common import InfoExtractor
   2 from ..utils import (
   3     ExtractorError,
   4     get_element_by_class,
   5     int_or_none,
   6     remove_start,
   7     strip_or_none,
   8     unified_strdate,
   9     urlencode_postdata,
  10 )
  11
  12
  13 class ScreencastOMaticIE(InfoExtractor):
  14     _VALID_URL = r'https?://screencast-o-matic\.com/(?:(?:watch|player)/|embed\?.*?\bsc=)(?P<id>[0-9a-zA-Z]+)'
  15     _TESTS = [{
  16         'url': 'http://screencast-o-matic.com/watch/c2lD3BeOPl',
  17         'md5': '483583cb80d92588f15ccbedd90f0c18',
  18         'info_dict': {
  19             'id': 'c2lD3BeOPl',
  20             'ext': 'mp4',
  21             'title': 'Welcome to 3-4 Philosophy @ DECV!',
  22             'thumbnail': r're:^https?://.*\.jpg$',
  23             'description': 'as the title says! also: some general info re 1) VCE philosophy and 2) distance learning.',
  24             'duration': 369,
  25             'upload_date': '20141216',
  26         },
  27     }, {
  28         'url': 'http://screencast-o-matic.com/player/c2lD3BeOPl',
  29         'only_matching': True,
  30     }, {
  31         'url': 'http://screencast-o-matic.com/embed?ff=true&sc=cbV2r4Q5TL&fromPH=true&a=1',
  32         'only_matching': True,
  33     }]
  34
  35     def _real_extract(self, url):
  36         video_id = self._match_id(url)
  37         webpage = self._download_webpage(
  38             'https://screencast-o-matic.com/player/' + video_id, video_id)
  39
  40         if (self._html_extract_title(webpage) == 'Protected Content'
  41                 or 'This video is private and requires a password' in webpage):
  42             password = self.get_param('videopassword')
  43
  44             if not password:
  45                 raise ExtractorError('Password protected video, use --video-password <password>', expected=True)
  46
  47             form = self._search_regex(
  48                 r'(?is)<form[^>]*>(?P<form>.+?)</form>', webpage, 'login form', group='form')
  49             form_data = self._hidden_inputs(form)
  50             form_data.update({
  51                 'scPassword': password,
  52             })
  53
  54             webpage = self._download_webpage(
  55                 'https://screencast-o-matic.com/player/password', video_id, 'Logging in',
  56                 data=urlencode_postdata(form_data))
  57
  58             if '<small class="text-danger">Invalid password</small>' in webpage:
  59                 raise ExtractorError('Unable to login: Invalid password', expected=True)
  60
  61         info = self._parse_html5_media_entries(url, webpage, video_id)[0]
  62         info.update({
  63             'id': video_id,
  64             'title': get_element_by_class('overlayTitle', webpage),
  65             'description': strip_or_none(get_element_by_class('overlayDescription', webpage)) or None,
  66             'duration': int_or_none(self._search_regex(
  67                 r'player\.duration\s*=\s*function\(\)\s*{\s*return\s+(\d+);\s*};',
  68                 webpage, 'duration', default=None)),
  69             'upload_date': unified_strdate(remove_start(
  70                 get_element_by_class('overlayPublished', webpage), 'Published: ')),
  71         })
  72         return info