X-Git-Url: https://git.rapsys.eu/youtubedl/blobdiff_plain/9c8b63077a48f758bf0c9a7351669557071bbd74..5dafebb045625934d6c679b05132c61c5f9793c5:/youtube_dl/extractor/cinemassacre.py diff --git a/youtube_dl/extractor/cinemassacre.py b/youtube_dl/extractor/cinemassacre.py index 6925b96..6d9cd8a 100644 --- a/youtube_dl/extractor/cinemassacre.py +++ b/youtube_dl/extractor/cinemassacre.py @@ -1,91 +1,107 @@ # encoding: utf-8 +from __future__ import unicode_literals + import re from .common import InfoExtractor -from ..utils import ( - ExtractorError, -) +from ..utils import ExtractorError +from .screenwavemedia import ScreenwaveMediaIE class CinemassacreIE(InfoExtractor): - _VALID_URL = r'(?:http://)?(?:www\.)?(?Pcinemassacre\.com/(?P[0-9]{4})/(?P[0-9]{2})/(?P[0-9]{2})/.+?)(?:[/?].*)?' - _TESTS = [{ - u'url': u'http://cinemassacre.com/2012/11/10/avgn-the-movie-trailer/', - u'file': u'19911.flv', - u'info_dict': { - u'upload_date': u'20121110', - u'title': u'“Angry Video Game Nerd: The Movie” – Trailer', - u'description': u'md5:fb87405fcb42a331742a0dce2708560b', + _VALID_URL = 'https?://(?:www\.)?cinemassacre\.com/(?P[0-9]{4})/(?P[0-9]{2})/(?P[0-9]{2})/(?P[^?#/]+)' + _TESTS = [ + { + 'url': 'http://cinemassacre.com/2012/11/10/avgn-the-movie-trailer/', + 'md5': 'fde81fbafaee331785f58cd6c0d46190', + 'info_dict': { + 'id': 'Cinemassacre-19911', + 'ext': 'mp4', + 'upload_date': '20121110', + 'title': '“Angry Video Game Nerd: The Movie” – Trailer', + 'description': 'md5:fb87405fcb42a331742a0dce2708560b', + }, }, - u'params': { - # rtmp download - u'skip_download': True, + { + 'url': 'http://cinemassacre.com/2013/10/02/the-mummys-hand-1940', + 'md5': 'd72f10cd39eac4215048f62ab477a511', + 'info_dict': { + 'id': 'Cinemassacre-521be8ef82b16', + 'ext': 'mp4', + 'upload_date': '20131002', + 'title': 'The Mummy’s Hand (1940)', + }, }, - }, - { - u'url': u'http://cinemassacre.com/2013/10/02/the-mummys-hand-1940', - u'file': u'521be8ef82b16.flv', - u'info_dict': { - u'upload_date': u'20131002', - u'title': u'The Mummy’s Hand (1940)', + { + # Youtube embedded video + 'url': 'http://cinemassacre.com/2006/12/07/chronologically-confused-about-bad-movie-and-video-game-sequel-titles/', + 'md5': 'df4cf8a1dcedaec79a73d96d83b99023', + 'info_dict': { + 'id': 'OEVzPCY2T-g', + 'ext': 'mp4', + 'title': 'AVGN: Chronologically Confused about Bad Movie and Video Game Sequel Titles', + 'upload_date': '20061207', + 'uploader': 'Cinemassacre', + 'uploader_id': 'JamesNintendoNerd', + 'description': 'md5:784734696c2b8b7f4b8625cc799e07f6', + } }, - u'params': { - # rtmp download - u'skip_download': True, + { + # Youtube embedded video + 'url': 'http://cinemassacre.com/2006/09/01/mckids/', + 'md5': '6eb30961fa795fedc750eac4881ad2e1', + 'info_dict': { + 'id': 'FnxsNhuikpo', + 'ext': 'mp4', + 'upload_date': '20060901', + 'uploader': 'Cinemassacre Extras', + 'description': 'md5:de9b751efa9e45fbaafd9c8a1123ed53', + 'uploader_id': 'Cinemassacre', + 'title': 'AVGN: McKids', + } }, - }] + { + 'url': 'http://cinemassacre.com/2015/05/25/mario-kart-64-nintendo-64-james-mike-mondays/', + 'md5': '1376908e49572389e7b06251a53cdd08', + 'info_dict': { + 'id': 'Cinemassacre-555779690c440', + 'ext': 'mp4', + 'description': 'Let’s Play Mario Kart 64 !! Mario Kart 64 is a classic go-kart racing game released for the Nintendo 64 (N64). Today James & Mike do 4 player Battle Mode with Kyle and Bootsy!', + 'title': 'Mario Kart 64 (Nintendo 64) James & Mike Mondays', + 'upload_date': '20150525', + } + } + ] def _real_extract(self, url): mobj = re.match(self._VALID_URL, url) + display_id = mobj.group('display_id') + video_date = mobj.group('date_y') + mobj.group('date_m') + mobj.group('date_d') - webpage_url = u'http://' + mobj.group('url') - webpage = self._download_webpage(webpage_url, None) # Don't know video id yet - video_date = mobj.group('date_Y') + mobj.group('date_m') + mobj.group('date_d') - mobj = re.search(r'src="(?Phttp://player\.screenwavemedia\.com/play/(?:embed|player)\.php\?id=(?:Cinemassacre-)?(?P.+?))"', webpage) - if not mobj: - raise ExtractorError(u'Can\'t extract embed url and video id') - playerdata_url = mobj.group(u'embed_url') - video_id = mobj.group(u'video_id') + webpage = self._download_webpage(url, display_id) - video_title = self._html_search_regex(r'(?P<title>.+?)\|', - webpage, u'title') - video_description = self._html_search_regex(r'<div class="entry-content">(?P<description>.+?)</div>', - webpage, u'description', flags=re.DOTALL, fatal=False) - if len(video_description) == 0: - video_description = None + playerdata_url = self._search_regex( + [ + ScreenwaveMediaIE.EMBED_PATTERN, + r'<iframe[^>]+src="(?P<url>(?:https?:)?//(?:[^.]+\.)?youtube\.com/.+?)"', + ], + webpage, 'player data URL', default=None, group='url') + if not playerdata_url: + raise ExtractorError('Unable to find player data') - playerdata = self._download_webpage(playerdata_url, video_id) - base_url = self._html_search_regex(r'\'streamer\': \'(?P<base_url>rtmp://.*?)/(?:vod|Cinemassacre)\'', - playerdata, u'base_url') - base_url += '/Cinemassacre/' - # Important: The file names in playerdata are not used by the player and even wrong for some videos - sd_file = 'Cinemassacre-%s_high.mp4' % video_id - hd_file = 'Cinemassacre-%s.mp4' % video_id - video_thumbnail = 'http://image.screenwavemedia.com/Cinemassacre/Cinemassacre-%s_thumb_640x360.jpg' % video_id - - formats = [ - { - 'url': base_url + sd_file, - 'ext': 'flv', - 'format': 'sd', - 'format_id': 'sd', - }, - { - 'url': base_url + hd_file, - 'ext': 'flv', - 'format': 'hd', - 'format_id': 'hd', - }, - ] + video_title = self._html_search_regex( + r'<title>(?P<title>.+?)\|', webpage, 'title') + video_description = self._html_search_regex( + r'<div class="entry-content">(?P<description>.+?)</div>', + webpage, 'description', flags=re.DOTALL, fatal=False) + video_thumbnail = self._og_search_thumbnail(webpage) - info = { - 'id': video_id, + return { + '_type': 'url_transparent', + 'display_id': display_id, 'title': video_title, - 'formats': formats, 'description': video_description, 'upload_date': video_date, 'thumbnail': video_thumbnail, + 'url': playerdata_url, } - # TODO: Remove when #980 has been merged - info.update(formats[-1]) - return info