1
0
Fork 0
mirror of https://github.com/ytdl-org/youtube-dl.git synced 2024-11-16 16:05:53 +00:00
youtube-dl/youtube_dl/extractor/dispeak.py

119 lines
4.3 KiB
Python
Raw Normal View History

from __future__ import unicode_literals
import re
from .common import InfoExtractor
from ..utils import (
int_or_none,
parse_duration,
remove_end,
xpath_element,
xpath_text,
)
class DigitallySpeakingIE(InfoExtractor):
_VALID_URL = r'https?://(?:s?evt\.dispeak|events\.digitallyspeaking)\.com/(?:[^/]+/)+xml/(?P<id>[^.]+)\.xml'
_TESTS = [{
2016-04-21 15:58:36 +00:00
# From http://gdcvault.com/play/1023460/Tenacious-Design-and-The-Interface
'url': 'http://evt.dispeak.com/ubm/gdc/sf16/xml/840376_BQRC.xml',
'md5': 'a8efb6c31ed06ca8739294960b2dbabd',
'info_dict': {
'id': '840376_BQRC',
'ext': 'mp4',
'title': 'Tenacious Design and The Interface of \'Destiny\'',
},
}, {
# From http://www.gdcvault.com/play/1014631/Classic-Game-Postmortem-PAC
'url': 'http://events.digitallyspeaking.com/gdc/sf11/xml/12396_1299111843500GMPX.xml',
'only_matching': True,
}, {
# From http://www.gdcvault.com/play/1013700/Advanced-Material
'url': 'http://sevt.dispeak.com/ubm/gdc/eur10/xml/11256_1282118587281VNIT.xml',
'only_matching': True,
}]
def _parse_mp4(self, metadata):
video_formats = []
video_root = None
mp4_video = xpath_text(metadata, './mp4video', default=None)
if mp4_video is not None:
mobj = re.match(r'(?P<root>https?://.*?/).*', mp4_video)
video_root = mobj.group('root')
if video_root is None:
http_host = xpath_text(metadata, 'httpHost', default=None)
if http_host:
video_root = 'http://%s/' % http_host
if video_root is None:
# Hard-coded in http://evt.dispeak.com/ubm/gdc/sf16/custom/player2.js
# Works for GPUTechConf, too
video_root = 'http://s3-2u.digitallyspeaking.com/'
formats = metadata.findall('./MBRVideos/MBRVideo')
if not formats:
return None
for a_format in formats:
stream_name = xpath_text(a_format, 'streamName', fatal=True)
video_path = re.match(r'mp4\:(?P<path>.*)', stream_name).group('path')
url = video_root + video_path
vbr = xpath_text(a_format, 'bitrate')
video_formats.append({
'url': url,
'vbr': int_or_none(vbr),
})
return video_formats
def _parse_flv(self, metadata):
formats = []
akamai_url = xpath_text(metadata, './akamaiHost', fatal=True)
audios = metadata.findall('./audios/audio')
for audio in audios:
formats.append({
'url': 'rtmp://%s/ondemand?ovpfv=1.1' % akamai_url,
'play_path': remove_end(audio.get('url'), '.flv'),
'ext': 'flv',
'vcodec': 'none',
'format_id': audio.get('code'),
})
slide_video_path = xpath_text(metadata, './slideVideo', fatal=True)
formats.append({
'url': 'rtmp://%s/ondemand?ovpfv=1.1' % akamai_url,
'play_path': remove_end(slide_video_path, '.flv'),
'ext': 'flv',
'format_note': 'slide deck video',
'quality': -2,
'preference': -2,
'format_id': 'slides',
})
speaker_video_path = xpath_text(metadata, './speakerVideo', fatal=True)
formats.append({
'url': 'rtmp://%s/ondemand?ovpfv=1.1' % akamai_url,
'play_path': remove_end(speaker_video_path, '.flv'),
'ext': 'flv',
'format_note': 'speaker video',
'quality': -1,
'preference': -1,
'format_id': 'speaker',
})
return formats
def _real_extract(self, url):
video_id = self._match_id(url)
xml_description = self._download_xml(url, video_id)
metadata = xpath_element(xml_description, 'metadata')
video_formats = self._parse_mp4(metadata)
if video_formats is None:
video_formats = self._parse_flv(metadata)
return {
'id': video_id,
'formats': video_formats,
'title': xpath_text(metadata, 'title', fatal=True),
'duration': parse_duration(xpath_text(metadata, 'endTime')),
'creator': xpath_text(metadata, 'speaker'),
}