]>
Commit | Line | Data |
---|---|---|
8adec2b9 | 1 | # coding: utf-8 |
2 | from __future__ import unicode_literals | |
3 | ||
4 | import re | |
5 | ||
6 | from .common import InfoExtractor | |
8adec2b9 | 7 | from ..utils import ( |
9732d77e | 8 | parse_duration, |
19f35402 YCH |
9 | parse_filesize, |
10 | str_to_int, | |
8adec2b9 | 11 | ) |
12 | ||
9732d77e | 13 | |
8adec2b9 | 14 | class SnotrIE(InfoExtractor): |
15 | _VALID_URL = r'http?://(?:www\.)?snotr\.com/video/(?P<id>\d+)/([\w]+)' | |
9732d77e | 16 | _TESTS = [{ |
8adec2b9 | 17 | 'url': 'http://www.snotr.com/video/13708/Drone_flying_through_fireworks', |
18 | 'info_dict': { | |
19 | 'id': '13708', | |
19f35402 | 20 | 'ext': 'mp4', |
8adec2b9 | 21 | 'title': 'Drone flying through fireworks!', |
19f35402 YCH |
22 | 'duration': 248, |
23 | 'filesize_approx': 40700000, | |
da8fb858 | 24 | 'description': 'A drone flying through Fourth of July Fireworks', |
ec85ded8 | 25 | 'thumbnail': r're:^https?://.*\.jpg$', |
19f35402 YCH |
26 | }, |
27 | 'expected_warnings': ['description'], | |
9732d77e | 28 | }, { |
8adec2b9 | 29 | 'url': 'http://www.snotr.com/video/530/David_Letteman_-_George_W_Bush_Top_10', |
30 | 'info_dict': { | |
31 | 'id': '530', | |
19f35402 | 32 | 'ext': 'mp4', |
8adec2b9 | 33 | 'title': 'David Letteman - George W. Bush Top 10', |
34 | 'duration': 126, | |
19f35402 | 35 | 'filesize_approx': 8500000, |
da8fb858 | 36 | 'description': 'The top 10 George W. Bush moments, brought to you by David Letterman!', |
ec85ded8 | 37 | 'thumbnail': r're:^https?://.*\.jpg$', |
9732d77e PH |
38 | } |
39 | }] | |
8adec2b9 | 40 | |
41 | def _real_extract(self, url): | |
42 | mobj = re.match(self._VALID_URL, url) | |
43 | video_id = mobj.group('id') | |
44 | ||
8adec2b9 | 45 | webpage = self._download_webpage(url, video_id) |
46 | title = self._og_search_title(webpage) | |
47 | ||
48 | description = self._og_search_description(webpage) | |
ad120ae1 YCH |
49 | info_dict = self._parse_html5_media_entries( |
50 | url, webpage, video_id, m3u8_entry_protocol='m3u8_native')[0] | |
8adec2b9 | 51 | |
9732d77e | 52 | view_count = str_to_int(self._html_search_regex( |
19f35402 | 53 | r'<p[^>]*>\s*<strong[^>]*>Views:</strong>\s*<span[^>]*>([\d,\.]+)', |
9732d77e | 54 | webpage, 'view count', fatal=False)) |
8adec2b9 | 55 | |
9732d77e | 56 | duration = parse_duration(self._html_search_regex( |
19f35402 | 57 | r'<p[^>]*>\s*<strong[^>]*>Length:</strong>\s*<span[^>]*>([\d:]+)', |
9732d77e | 58 | webpage, 'duration', fatal=False)) |
8adec2b9 | 59 | |
19f35402 YCH |
60 | filesize_approx = parse_filesize(self._html_search_regex( |
61 | r'<p[^>]*>\s*<strong[^>]*>Filesize:</strong>\s*<span[^>]*>([^<]+)', | |
62 | webpage, 'filesize', fatal=False)) | |
8adec2b9 | 63 | |
19f35402 | 64 | info_dict.update({ |
8adec2b9 | 65 | 'id': video_id, |
da8fb858 | 66 | 'description': description, |
8adec2b9 | 67 | 'title': title, |
9732d77e PH |
68 | 'view_count': view_count, |
69 | 'duration': duration, | |
70 | 'filesize_approx': filesize_approx, | |
19f35402 YCH |
71 | }) |
72 | ||
73 | return info_dict |