yt-dlc/youtube_dl/extractor/normalboots.py

57 lines
2.2 KiB
Python
Raw Normal View History

2016-10-02 13:39:18 +02:00
# coding: utf-8
2014-02-23 18:28:22 +01:00
from __future__ import unicode_literals
2014-01-30 18:26:50 +01:00
from .common import InfoExtractor
from .screenwavemedia import ScreenwaveMediaIE
2014-01-30 18:26:50 +01:00
from ..utils import (
unified_strdate,
)
2014-02-23 18:28:22 +01:00
2014-01-30 18:26:50 +01:00
class NormalbootsIE(InfoExtractor):
_VALID_URL = r'https?://(?:www\.)?normalboots\.com/video/(?P<id>[0-9a-z-]*)/?$'
2014-01-30 21:01:35 +01:00
_TEST = {
2014-02-23 18:28:22 +01:00
'url': 'http://normalboots.com/video/home-alone-games-jontron/',
'info_dict': {
'id': 'home-alone-games-jontron',
'ext': 'mp4',
'title': 'Home Alone Games - JonTron - NormalBoots',
'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/',
'uploader': 'JonTron',
'upload_date': '20140125',
},
'params': {
# m3u8 download
'skip_download': True,
},
'add_ie': ['ScreenwaveMedia'],
2014-01-30 21:01:35 +01:00
}
2014-02-23 18:28:22 +01:00
2014-01-30 18:26:50 +01:00
def _real_extract(self, url):
2015-02-02 23:04:39 +01:00
video_id = self._match_id(url)
2014-01-30 18:26:50 +01:00
webpage = self._download_webpage(url, video_id)
2014-02-23 18:28:22 +01:00
2015-02-02 23:04:39 +01:00
video_uploader = self._html_search_regex(
r'Posted\sby\s<a\shref="[A-Za-z0-9/]*">(?P<uploader>[A-Za-z]*)\s</a>',
webpage, 'uploader', fatal=False)
video_upload_date = unified_strdate(self._html_search_regex(
r'<span style="text-transform:uppercase; font-size:inherit;">[A-Za-z]+, (?P<date>.*)</span>',
webpage, 'date', fatal=False))
screenwavemedia_url = self._html_search_regex(
ScreenwaveMediaIE.EMBED_PATTERN, webpage, 'screenwave URL',
group='url')
2014-02-23 18:28:22 +01:00
return {
'_type': 'url_transparent',
2014-02-23 18:28:22 +01:00
'id': video_id,
'url': screenwavemedia_url,
'ie_key': ScreenwaveMediaIE.ie_key(),
2014-02-23 18:28:22 +01:00
'title': self._og_search_title(webpage),
'description': self._og_search_description(webpage),
'thumbnail': self._og_search_thumbnail(webpage),
'uploader': video_uploader,
'upload_date': video_upload_date,
}