af44c3bb5714bc0079e3d1307782a8ff1fe5ba84
[youtube-dl] / youtube_dl / extractor / normalboots.py
1 # encoding: utf-8
2 from __future__ import unicode_literals
3
4 from .common import InfoExtractor
5 from .screenwavemedia import ScreenwaveMediaIE
6
7 from ..utils import (
8     unified_strdate,
9 )
10
11
12 class NormalbootsIE(InfoExtractor):
13     _VALID_URL = r'https?://(?:www\.)?normalboots\.com/video/(?P<id>[0-9a-z-]*)/?$'
14     _TEST = {
15         'url': 'http://normalboots.com/video/home-alone-games-jontron/',
16         'info_dict': {
17             'id': 'home-alone-games-jontron',
18             'ext': 'mp4',
19             'title': 'Home Alone Games - JonTron - NormalBoots',
20             '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/',
21             'uploader': 'JonTron',
22             'upload_date': '20140125',
23         },
24         'params': {
25             # m3u8 download
26             'skip_download': True,
27         },
28         'add_ie': ['ScreenwaveMedia'],
29     }
30
31     def _real_extract(self, url):
32         video_id = self._match_id(url)
33         webpage = self._download_webpage(url, video_id)
34
35         video_uploader = self._html_search_regex(
36             r'Posted\sby\s<a\shref="[A-Za-z0-9/]*">(?P<uploader>[A-Za-z]*)\s</a>',
37             webpage, 'uploader', fatal=False)
38         video_upload_date = unified_strdate(self._html_search_regex(
39             r'<span style="text-transform:uppercase; font-size:inherit;">[A-Za-z]+, (?P<date>.*)</span>',
40             webpage, 'date', fatal=False))
41
42         screenwavemedia_url = self._html_search_regex(
43             ScreenwaveMediaIE.EMBED_PATTERN, webpage, 'screenwave URL',
44             group='url')
45
46         return {
47             '_type': 'url_transparent',
48             'id': video_id,
49             'url': screenwavemedia_url,
50             'ie_key': ScreenwaveMediaIE.ie_key(),
51             'title': self._og_search_title(webpage),
52             'description': self._og_search_description(webpage),
53             'thumbnail': self._og_search_thumbnail(webpage),
54             'uploader': video_uploader,
55             'upload_date': video_upload_date,
56         }