projects / yt-dlp.git / blame
| Commit | Line | Data |
|---|---|---|
| fc6861b1 S |
1 | # coding: utf-8 |
| 2 | from __future__ import unicode_literals | |
| 3 | ||
| 4 | import re | |
| 5 | ||
| 6 | from .common import InfoExtractor | |
| 7 | from ..utils import ( | |
| 8 | parse_duration, | |
| 9 | parse_iso8601, | |
| 10 | int_or_none, | |
| 11 | ) | |
| 12 | ||
| 13 | ||
| 14 | class SportBoxIE(InfoExtractor): | |
| 15 | _VALID_URL = r'https?://news\.sportbox\.ru/Vidy_sporta/(?:[^/]+/)+spbvideo_NI\d+_(?P<display_id>.+)' | |
| 16 | _TESTS = [ | |
| 17 | { | |
| 18 | 'url': 'http://news.sportbox.ru/Vidy_sporta/Avtosport/Rossijskij/spbvideo_NI483529_Gonka-2-zaezd-Obyedinenniy-2000-klassi-Turing-i-S', | |
| 19 | 'md5': 'ff56a598c2cf411a9a38a69709e97079', | |
| 20 | 'info_dict': { | |
| 21 | 'id': '80822', | |
| 22 | 'ext': 'mp4', | |
| 23 | 'title': 'Гонка 2 заезд ««Объединенный 2000»: классы Туринг и Супер-продакшн', | |
| 24 | 'description': 'md5:81715fa9c4ea3d9e7915dc8180c778ed', | |
| 25 | 'thumbnail': 're:^https?://.*\.jpg$', | |
| 26 | 'timestamp': 1411896237, | |
| 27 | 'upload_date': '20140928', | |
| 28 | 'duration': 4846, | |
| 29 | 'view_count': int, | |
| 30 | }, | |
| 31 | 'params': { | |
| 32 | # m3u8 download | |
| 33 | 'skip_download': True, | |
| 34 | }, | |
| 35 | }, { | |
| 36 | 'url': 'http://news.sportbox.ru/Vidy_sporta/billiard/spbvideo_NI486287_CHempionat-mira-po-dinamichnoy-piramide-4', | |
| 37 | 'only_matching': True, | |
| 38 | } | |
| 39 | ] | |
| 40 | ||
| 41 | def _real_extract(self, url): | |
| 42 | mobj = re.match(self._VALID_URL, url) | |
| 43 | display_id = mobj.group('display_id') | |
| 44 | ||
| 45 | webpage = self._download_webpage(url, display_id) | |
| 46 | ||
| 47 | video_id = self._search_regex( | |
| 48 | r'src="/vdl/player/media/(\d+)"', webpage, 'video id') | |
| 49 | ||
| 50 | player = self._download_webpage( | |
| 51 | 'http://news.sportbox.ru/vdl/player/media/%s' % video_id, | |
| 52 | display_id, 'Downloading player webpage') | |
| 53 | ||
| 54 | hls = self._search_regex( | |
| 55 | r"var\s+original_hls_file\s*=\s*'([^']+)'", player, 'hls file') | |
| 56 | ||
| 57 | formats = self._extract_m3u8_formats(hls, display_id, 'mp4') | |
| 58 | ||
| 59 | title = self._html_search_regex( | |
| 60 | r'<h1 itemprop="name">([^<]+)</h1>', webpage, 'title') | |
| 61 | description = self._html_search_regex( | |
| 62 | r'(?s)<div itemprop="description">(.+?)</div>', webpage, 'description', fatal=False) | |
| 63 | thumbnail = self._og_search_thumbnail(webpage) | |
| 64 | timestamp = parse_iso8601(self._search_regex( | |
| 65 | r'<span itemprop="uploadDate">([^<]+)</span>', webpage, 'timestamp', fatal=False)) | |
| 66 | duration = parse_duration(self._html_search_regex( | |
| 67 | r'<meta itemprop="duration" content="PT([^"]+)">', webpage, 'duration', fatal=False)) | |
| 68 | view_count = int_or_none(self._html_search_regex( | |
| 69 | r'<span>Просмотров: (\d+)</span>', player, 'view count', fatal=False)) | |
| 70 | ||
| 71 | return { | |
| 72 | 'id': video_id, | |
| 73 | 'display_id': display_id, | |
| 74 | 'title': title, | |
| 75 | 'description': description, | |
| 76 | 'thumbnail': thumbnail, | |
| 77 | 'timestamp': timestamp, | |
| 78 | 'duration': duration, | |
| 79 | 'view_count': view_count, | |
| 80 | 'formats': formats, | |
| 81 | } |