yt_dlp/extractor/normalboots.py

   1 from .common import InfoExtractor
   2 from .jwplatform import JWPlatformIE
   3
   4 from ..utils import (
   5     unified_strdate,
   6 )
   7
   8
   9 class NormalbootsIE(InfoExtractor):
  10     _VALID_URL = r'https?://(?:www\.)?normalboots\.com/video/(?P<id>[0-9a-z-]*)/?$'
  11     _TEST = {
  12         'url': 'http://normalboots.com/video/home-alone-games-jontron/',
  13         'info_dict': {
  14             'id': 'home-alone-games-jontron',
  15             'ext': 'mp4',
  16             'title': 'Home Alone Games - JonTron - NormalBoots',
  17             'description': 'Jon is late for Christmas. Typical. Thanks to: Paul Ritchey for Co-Writing/Filming: http://www.youtube.com/user/ContinueShow Michael Azzi for Christmas Intro Animation: http://michafrar.tumblr.com/ Jerrod Waters for Christmas Intro Music: http://www.youtube.com/user/xXJerryTerryXx Casey Ormond for ‘Tense Battle Theme’:\xa0http://www.youtube.com/Kiamet/',
  18             'uploader': 'JonTron',
  19             'upload_date': '20140125',
  20         },
  21         'params': {
  22             # m3u8 download
  23             'skip_download': True,
  24         },
  25         'add_ie': ['JWPlatform'],
  26     }
  27
  28     def _real_extract(self, url):
  29         video_id = self._match_id(url)
  30         webpage = self._download_webpage(url, video_id)
  31
  32         video_uploader = self._html_search_regex(
  33             r'Posted\sby\s<a\shref="[A-Za-z0-9/]*">(?P<uploader>[A-Za-z]*)\s</a>',
  34             webpage, 'uploader', fatal=False)
  35         video_upload_date = unified_strdate(self._html_search_regex(
  36             r'<span style="text-transform:uppercase; font-size:inherit;">[A-Za-z]+, (?P<date>.*)</span>',
  37             webpage, 'date', fatal=False))
  38
  39         jwplatform_url = JWPlatformIE._extract_url(webpage)
  40
  41         return {
  42             '_type': 'url_transparent',
  43             'id': video_id,
  44             'url': jwplatform_url,
  45             'ie_key': JWPlatformIE.ie_key(),
  46             'title': self._og_search_title(webpage),
  47             'description': self._og_search_description(webpage),
  48             'thumbnail': self._og_search_thumbnail(webpage),
  49             'uploader': video_uploader,
  50             'upload_date': video_upload_date,
  51         }