import re import xml.etree.ElementTree from .common import InfoExtractor from ..utils import ( compat_urlparse, xpath_with_ns, determine_ext, ) class InternetVideoArchiveIE(InfoExtractor): _VALID_URL = r'https?://video\.internetvideoarchive\.net/flash/players/.*?\?.*?publishedid.*?' _TEST = { u'url': u'http://video.internetvideoarchive.net/flash/players/flashconfiguration.aspx?customerid=69249&publishedid=452693&playerid=247', u'file': u'452693.mp4', u'info_dict': { u'title': u'SKYFALL', u'description': u'In SKYFALL, Bond\'s loyalty to M is tested as her past comes back to haunt her. As MI6 comes under attack, 007 must track down and destroy the threat, no matter how personal the cost.', u'duration': 156, }, } @staticmethod def _build_url(query): return 'http://video.internetvideoarchive.net/flash/players/flashconfiguration.aspx?' + query def _real_extract(self, url): query = compat_urlparse.urlparse(url).query query_dic = compat_urlparse.parse_qs(query) video_id = query_dic['publishedid'][0] url = self._build_url(query) flashconfiguration_xml = self._download_webpage(url, video_id, u'Downloading flash configuration') flashconfiguration = xml.etree.ElementTree.fromstring(flashconfiguration_xml.encode('utf-8')) file_url = flashconfiguration.find('file').text file_url = file_url.replace('/playlist.aspx', '/mrssplaylist.aspx') info_xml = self._download_webpage(file_url, video_id, u'Downloading video info') info = xml.etree.ElementTree.fromstring(info_xml.encode('utf-8')) item = info.find('channel/item') def _bp(p): return xpath_with_ns(p, {'media': 'http://search.yahoo.com/mrss/', 'jwplayer': 'http://developer.longtailvideo.com/trac/wiki/FlashFormats'}) formats = [] for content in item.findall(_bp('media:group/media:content')): attr = content.attrib f_url = attr['url'] formats.append({ 'url': f_url, 'ext': determine_ext(f_url), 'width': int(attr['width']), 'bitrate': int(attr['bitrate']), }) formats = sorted(formats, key=lambda f: f['bitrate']) info = { 'id': video_id, 'title': item.find('title').text, 'formats': formats, 'thumbnail': item.find(_bp('media:thumbnail')).attrib['url'], 'description': item.find('description').text, 'duration': int(attr['duration']), } # TODO: Remove when #980 has been merged info.update(formats[-1]) return info