1 from __future__ import unicode_literals
5 import xml.etree.ElementTree
7 from .common import InfoExtractor
8 from ..utils import unified_strdate
11 class GDCVaultIE(InfoExtractor):
12 _VALID_URL = r'https?://(?:www\.)?gdcvault\.com/play/(?P<id>\d+)/(?P<name>(\w|-)+)'
15 u'url': u'http://www.gdcvault.com/play/1019721/Doki-Doki-Universe-Sweet-Simple',
16 u'md5': u'7ce8388f544c88b7ac11c7ab1b593704',
20 u"title": u"Doki-Doki Universe: Sweet, Simple and Genuine (GDC Next 10)"
24 def _real_extract(self, url):
25 mobj = re.match(self._VALID_URL, url)
27 video_id = mobj.group('id')
28 webpage_url = 'http://www.gdcvault.com/play/' + video_id
30 start_page = self._download_webpage(webpage_url, video_id)
32 self.report_extraction(video_id)
34 xml_root = self._html_search_regex(r'<iframe src="(?P<xml_root>.*?)player.html.*?".*?</iframe>', start_page, 'xml root')
35 xml_name = self._html_search_regex(r'<iframe src=".*?\?xml=(?P<xml_file>.+?\.xml).*?".*?</iframe>', start_page, 'xml filename')
36 xml_decription_url = xml_root + 'xml/' + xml_name
38 xml_description = self._download_xml(xml_decription_url, video_id)
40 video_title = xml_description.find('./metadata/title').text
42 mp4_video = xml_description.find('./metadata/mp4video').text
43 mobj = re.match(r'(?P<root>https?://.*?/).*', mp4_video)
44 video_root = mobj.group('root')
46 formats = xml_description.findall('./metadata/MBRVideos/MBRVideo')
48 for format in formats:
49 mobj = re.match(r'mp4\:(?P<path>.*)', format.find('streamName').text)
50 url = video_root + mobj.group('path')
51 vbr = format.find('bitrate').text
52 video_formats.append({
59 'formats': video_formats,