mirror of
https://github.com/ytdl-org/youtube-dl.git
synced 2024-11-25 08:10:31 +01:00
Merge remote-tracking branch 'origin/master'
This commit is contained in:
commit
204da0d3e3
@ -50,6 +50,7 @@ from .keek import KeekIE
|
||||
from .liveleak import LiveLeakIE
|
||||
from .livestream import LivestreamIE
|
||||
from .metacafe import MetacafeIE
|
||||
from .mit import TechTVMITIE, MITIE
|
||||
from .mixcloud import MixcloudIE
|
||||
from .mtv import MTVIE
|
||||
from .muzu import MuzuTVIE
|
||||
|
76
youtube_dl/extractor/mit.py
Normal file
76
youtube_dl/extractor/mit.py
Normal file
@ -0,0 +1,76 @@
|
||||
import re
|
||||
import json
|
||||
|
||||
from .common import InfoExtractor
|
||||
from ..utils import (
|
||||
clean_html,
|
||||
get_element_by_id,
|
||||
)
|
||||
|
||||
|
||||
class TechTVMITIE(InfoExtractor):
|
||||
IE_NAME = u'techtv.mit.edu'
|
||||
_VALID_URL = r'https?://techtv\.mit\.edu/(videos|embeds)/(?P<id>\d+)'
|
||||
|
||||
_TEST = {
|
||||
u'url': u'http://techtv.mit.edu/videos/25418-mit-dna-learning-center-set',
|
||||
u'file': u'25418.mp4',
|
||||
u'md5': u'1f8cb3e170d41fd74add04d3c9330e5f',
|
||||
u'info_dict': {
|
||||
u'title': u'MIT DNA Learning Center Set',
|
||||
u'description': u'md5:82313335e8a8a3f243351ba55bc1b474',
|
||||
},
|
||||
}
|
||||
|
||||
def _real_extract(self, url):
|
||||
mobj = re.match(self._VALID_URL, url)
|
||||
video_id = mobj.group('id')
|
||||
webpage = self._download_webpage(
|
||||
'http://techtv.mit.edu/videos/%s' % video_id, video_id)
|
||||
embed_page = self._download_webpage(
|
||||
'http://techtv.mit.edu/embeds/%s/' % video_id, video_id,
|
||||
note=u'Downloading embed page')
|
||||
|
||||
base_url = self._search_regex(r'ipadUrl: \'(.+?cloudfront.net/)',
|
||||
embed_page, u'base url')
|
||||
formats_json = self._search_regex(r'bitrates: (\[.+?\])', embed_page,
|
||||
u'video formats')
|
||||
formats = json.loads(formats_json)
|
||||
formats = sorted(formats, key=lambda f: f['bitrate'])
|
||||
|
||||
title = get_element_by_id('edit-title', webpage)
|
||||
description = clean_html(get_element_by_id('edit-description', webpage))
|
||||
thumbnail = self._search_regex(r'playlist:.*?url: \'(.+?)\'',
|
||||
embed_page, u'thumbnail', flags=re.DOTALL)
|
||||
|
||||
return {'id': video_id,
|
||||
'title': title,
|
||||
'url': base_url + formats[-1]['url'].replace('mp4:', ''),
|
||||
'ext': 'mp4',
|
||||
'description': description,
|
||||
'thumbnail': thumbnail,
|
||||
}
|
||||
|
||||
|
||||
class MITIE(TechTVMITIE):
|
||||
IE_NAME = u'video.mit.edu'
|
||||
_VALID_URL = r'https?://video\.mit\.edu/watch/(?P<title>[^/]+)'
|
||||
|
||||
_TEST = {
|
||||
u'url': u'http://video.mit.edu/watch/the-government-is-profiling-you-13222/',
|
||||
u'file': u'21783.mp4',
|
||||
u'md5': u'7db01d5ccc1895fc5010e9c9e13648da',
|
||||
u'info_dict': {
|
||||
u'title': u'The Government is Profiling You',
|
||||
u'description': u'md5:ad5795fe1e1623b73620dbfd47df9afd',
|
||||
},
|
||||
}
|
||||
|
||||
def _real_extract(self, url):
|
||||
mobj = re.match(self._VALID_URL, url)
|
||||
page_title = mobj.group('title')
|
||||
webpage = self._download_webpage(url, page_title)
|
||||
self.to_screen('%s: Extracting %s url' % (page_title, TechTVMITIE.IE_NAME))
|
||||
embed_url = self._search_regex(r'<iframe .*?src="(.+?)"', webpage,
|
||||
u'embed url')
|
||||
return self.url_result(embed_url, ie='TechTVMIT')
|
Loading…
Reference in New Issue
Block a user