projects / yt-dlp.git / blame
| Commit | Line | Data |
|---|---|---|
| e94443de YCH |
1 | # encoding: utf-8 |
| 2 | from __future__ import unicode_literals | |
| 3 | ||
| 4 | import re | |
| 5 | ||
| 6 | from .common import InfoExtractor | |
| 7 | from ..utils import ExtractorError | |
| 8 | from .bliptv import BlipTVIE | |
| 9 | ||
| 10 | ||
| 11 | class CinemassacreIE(InfoExtractor): | |
| 12 | _VALID_URL = 'https?://(?:www\.)?cinemassacre\.com/(?P<date_y>[0-9]{4})/(?P<date_m>[0-9]{2})/(?P<date_d>[0-9]{2})/(?P<display_id>[^?#/]+)' | |
| 13 | _TESTS = [ | |
| 14 | { | |
| 15 | 'url': 'http://cinemassacre.com/2012/11/10/avgn-the-movie-trailer/', | |
| 16 | 'md5': 'fde81fbafaee331785f58cd6c0d46190', | |
| 17 | 'info_dict': { | |
| 18 | 'id': 'Cinemassacre-19911', | |
| 19 | 'ext': 'mp4', | |
| 20 | 'upload_date': '20121110', | |
| 21 | 'title': '“Angry Video Game Nerd: The Movie” – Trailer', | |
| 22 | 'description': 'md5:fb87405fcb42a331742a0dce2708560b', | |
| 23 | }, | |
| 24 | }, | |
| 25 | { | |
| 26 | 'url': 'http://cinemassacre.com/2013/10/02/the-mummys-hand-1940', | |
| 27 | 'md5': 'd72f10cd39eac4215048f62ab477a511', | |
| 28 | 'info_dict': { | |
| 29 | 'id': 'Cinemassacre-521be8ef82b16', | |
| 30 | 'ext': 'mp4', | |
| 31 | 'upload_date': '20131002', | |
| 32 | 'title': 'The Mummy’s Hand (1940)', | |
| 33 | }, | |
| 34 | }, | |
| 35 | { | |
| 36 | 'url': 'http://cinemassacre.com/2006/12/07/chronologically-confused-about-bad-movie-and-video-game-sequel-titles/', | |
| 37 | 'md5': 'ca9b3c8dd5a66f9375daeb5135f5a3de', | |
| 38 | 'info_dict': { | |
| 39 | 'id': '4065369', | |
| 40 | 'ext': 'flv', | |
| 41 | 'title': 'AVGN: Chronologically Confused about Bad Movie and Video Game Sequel Titles', | |
| 42 | 'upload_date': '20061207', | |
| 43 | 'uploader': 'cinemassacre', | |
| 44 | 'uploader_id': '250778', | |
| 45 | 'timestamp': 1283233867, | |
| 46 | 'description': 'md5:0a108c78d130676b207d0f6d029ecffd', | |
| 47 | } | |
| 48 | } | |
| 49 | ] | |
| 50 | ||
| 51 | def _real_extract(self, url): | |
| 52 | mobj = re.match(self._VALID_URL, url) | |
| 53 | display_id = mobj.group('display_id') | |
| 54 | video_date = mobj.group('date_y') + mobj.group('date_m') + mobj.group('date_d') | |
| 55 | ||
| 56 | webpage = self._download_webpage(url, display_id) | |
| 57 | ||
| 58 | playerdata_url = self._search_regex( | |
| 59 | r'src="(http://player\.screenwavemedia\.com/play/[a-zA-Z]+\.php\?[^"]*\bid=.+?)"', | |
| 60 | webpage, 'player data URL', default=None) | |
| 61 | if not playerdata_url: | |
| 62 | playerdata_url = BlipTVIE._extract_url(webpage) | |
| 63 | if not playerdata_url: | |
| 64 | raise ExtractorError('Unable to find player data') | |
| 65 | ||
| 66 | video_title = self._html_search_regex( | |
| 67 | r'<title>(?P<title>.+?)\|', webpage, 'title') | |
| 68 | video_description = self._html_search_regex( | |
| 69 | r'<div class="entry-content">(?P<description>.+?)</div>', | |
| 70 | webpage, 'description', flags=re.DOTALL, fatal=False) | |
| 71 | video_thumbnail = self._og_search_thumbnail(webpage) | |
| 72 | ||
| 73 | return { | |
| 74 | '_type': 'url_transparent', | |
| 75 | 'display_id': display_id, | |
| 76 | 'title': video_title, | |
| 77 | 'description': video_description, | |
| 78 | 'upload_date': video_date, | |
| 79 | 'thumbnail': video_thumbnail, | |
| 80 | 'url': playerdata_url, | |
| 81 | } |