yt-dlc/youtube_dl/extractor/infoq.py

55 lines
1.8 KiB
Python
Raw Normal View History

from __future__ import unicode_literals
2013-06-23 21:14:19 +02:00
import base64
import re
from .common import InfoExtractor
from ..utils import (
compat_urllib_parse,
)
class InfoQIE(InfoExtractor):
2014-02-02 12:03:36 +01:00
_VALID_URL = r'https?://(?:www\.)?infoq\.com/[^/]+/(?P<id>[^/]+)$'
2014-04-20 01:01:37 +02:00
2013-06-27 20:27:08 +02:00
_TEST = {
2014-04-20 01:01:37 +02:00
u'name': u'InfoQ',
u'url': u'http://www.infoq.com/presentations/A-Few-of-My-Favorite-Python-Things',
u'md5': u'fcaa3d995e04080dcb9465d86b5eef62',
u'info_dict': {
u'id': u'12-jan-pythonthings',
u'ext': u'mp4',
u'description': u'Mike Pirnat presents some tips and tricks, standard libraries and third party packages that make programming in Python a richer experience.',
u'title': u'A Few of My Favorite [Python] Things',
2013-06-27 20:27:08 +02:00
},
}
2013-06-23 21:14:19 +02:00
def _real_extract(self, url):
mobj = re.match(self._VALID_URL, url)
2014-02-02 12:03:36 +01:00
video_id = mobj.group('id')
2013-06-23 21:14:19 +02:00
2014-02-02 12:03:36 +01:00
webpage = self._download_webpage(url, video_id)
2013-06-23 21:14:19 +02:00
2014-04-20 01:01:37 +02:00
self.report_extraction(video_id)
2013-06-23 21:14:19 +02:00
2014-04-20 01:01:37 +02:00
video_title = self._html_search_regex(r'<title>(.*?)</title>', webpage, 'title')
video_description = self._html_search_meta('description', webpage, 'description')
2013-06-23 21:14:19 +02:00
2014-04-20 01:01:37 +02:00
video_url = 'rtmpe://video.infoq.com/cfx/st/'
base64playpath = self._search_regex(r"jsclassref = '([^']*)'", webpage, 'jsclassref')
playpath = 'mp4:' + base64.b64decode(base64playpath).decode('utf-8')
2013-06-23 21:14:19 +02:00
2014-04-20 01:01:37 +02:00
video_filename = playpath.split('/')[-1]
2013-06-23 21:14:19 +02:00
video_id, extension = video_filename.split('.')
2014-04-20 01:01:37 +02:00
return [{
2013-06-23 21:14:19 +02:00
'id': video_id,
'title': video_title,
'description': video_description,
2014-04-20 01:01:37 +02:00
'formats': [{
'url': video_url,
'ext': extension,
'play_path': playpath,
}],
}]