X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Fscreenwavemedia.py;h=3bc84989e2b4d4c842714a1d815070eeb05e45c9;hb=11bed5827dace09b5483b159476ce9f8c29d6078;hp=6c9fdb7c1aceb35efc166c9207fd503603040b9b;hpb=cf372f0778e82cdc181a6173909589e640ac29fb;p=youtube-dl diff --git a/youtube_dl/extractor/screenwavemedia.py b/youtube_dl/extractor/screenwavemedia.py index 6c9fdb7c1..3bc84989e 100644 --- a/youtube_dl/extractor/screenwavemedia.py +++ b/youtube_dl/extractor/screenwavemedia.py @@ -1,17 +1,16 @@ # encoding: utf-8 from __future__ import unicode_literals -import re - from .common import InfoExtractor from ..utils import ( int_or_none, unified_strdate, + js_to_json, ) class ScreenwaveMediaIE(InfoExtractor): - _VALID_URL = r'http://player\.screenwavemedia\.com/play/[a-zA-Z]+\.php\?[^"]*\bid=(?P.+)' + _VALID_URL = r'http://player\d?\.screenwavemedia\.com/(?:play/)?[a-zA-Z]+\.php\?[^"]*\bid=(?P.+)' _TESTS = [{ 'url': 'http://player.screenwavemedia.com/play/play.php?playerdiv=videoarea&companiondiv=squareAd&id=Cinemassacre-19911', @@ -20,58 +19,50 @@ class ScreenwaveMediaIE(InfoExtractor): def _real_extract(self, url): video_id = self._match_id(url) - playerdata = self._download_webpage(url, video_id, 'Downloading player webpage') + + playerdata = self._download_webpage( + 'http://player.screenwavemedia.com/player.php?id=%s' % video_id, + video_id, 'Downloading player webpage') vidtitle = self._search_regex( r'\'vidtitle\'\s*:\s*"([^"]+)"', playerdata, 'vidtitle').replace('\\/', '/') - vidurl = self._search_regex( - r'\'vidurl\'\s*:\s*"([^"]+)"', playerdata, 'vidurl').replace('\\/', '/') - - videolist_url = None - - mobj = re.search(r"'videoserver'\s*:\s*'(?P[^']+)'", playerdata) - if mobj: - videoserver = mobj.group('videoserver') - mobj = re.search(r'\'vidid\'\s*:\s*"(?P[^\']+)"', playerdata) - vidid = mobj.group('vidid') if mobj else video_id - videolist_url = 'http://%s/vod/smil:%s.smil/jwplayer.smil' % (videoserver, vidid) - else: - mobj = re.search(r"file\s*:\s*'(?Phttp.+?/jwplayer\.smil)'", playerdata) - if mobj: - videolist_url = mobj.group('smil') - - if videolist_url: - videolist = self._download_xml(videolist_url, video_id, 'Downloading videolist XML') - formats = [] - baseurl = vidurl[:vidurl.rfind('/') + 1] - for video in videolist.findall('.//video'): - src = video.get('src') - if not src: - continue - file_ = src.partition(':')[-1] - width = int_or_none(video.get('width')) - height = int_or_none(video.get('height')) - bitrate = int_or_none(video.get('system-bitrate'), scale=1000) - format = { - 'url': baseurl + file_, - 'format_id': src.rpartition('.')[0].rpartition('_')[-1], - } - if width or height: - format.update({ - 'tbr': bitrate, - 'width': width, - 'height': height, - }) - else: - format.update({ - 'abr': bitrate, - 'vcodec': 'none', - }) - formats.append(format) - else: - formats = [{ - 'url': vidurl, - }] + + playerconfig = self._download_webpage( + 'http://player.screenwavemedia.com/player.js', + video_id, 'Downloading playerconfig webpage') + + videoserver = self._search_regex(r"\[ipaddress\]\s*=>\s*([\d\.]+)", playerdata, 'videoserver') + + sources = self._parse_json( + js_to_json( + self._search_regex( + r"sources\s*:\s*(\[[^\]]+?\])", playerconfig, + 'sources', + ).replace( + "' + thisObj.options.videoserver + '", + videoserver + ).replace( + "' + playerVidId + '", + video_id + ) + ), + video_id + ) + + formats = [] + for source in sources: + if source['type'] == 'hls': + formats.extend(self._extract_m3u8_formats(source['file'], video_id)) + else: + format_label = source.get('label') + height = int_or_none(self._search_regex( + r'^(\d+)[pP]', format_label, 'height', default=None)) + formats.append({ + 'url': source['file'], + 'format': format_label, + 'ext': source.get('type'), + 'height': height, + }) self._sort_formats(formats) return { @@ -81,60 +72,6 @@ class ScreenwaveMediaIE(InfoExtractor): } -class CinemassacreIE(InfoExtractor): - _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', - }, - }, - { - '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)', - }, - } - ] - - 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 = self._download_webpage(url, display_id) - - playerdata_url = self._search_regex( - r'src="(http://player\.screenwavemedia\.com/play/[a-zA-Z]+\.php\?[^"]*\bid=.+?)"', - webpage, 'player data URL') - video_title = self._html_search_regex( - r'(?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) - - return { - '_type': 'url_transparent', - 'display_id': display_id, - 'title': video_title, - 'description': video_description, - 'upload_date': video_date, - 'thumbnail': video_thumbnail, - 'url': playerdata_url, - } - - class TeamFourIE(InfoExtractor): _VALID_URL = r'https?://(?:www\.)?teamfourstar\.com/video/(?P<id>[a-z0-9\-]+)/?' _TEST = { @@ -153,7 +90,7 @@ class TeamFourIE(InfoExtractor): webpage = self._download_webpage(url, display_id) playerdata_url = self._search_regex( - r'src="(http://player\.screenwavemedia\.com/play/[a-zA-Z]+\.php\?[^"]*\bid=.+?)"', + r'src="(http://player\d?\.screenwavemedia\.com/(?:play/)?[a-zA-Z]+\.php\?[^"]*\bid=.+?)"', webpage, 'player data URL') video_title = self._html_search_regex(