yt-dlp/youtube_dl/extractor/infoq.py

56 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>[^/]+)$'
2013-06-27 20:27:08 +02:00
_TEST = {
"name": "InfoQ",
"url": "http://www.infoq.com/presentations/A-Few-of-My-Favorite-Python-Things",
"file": "12-jan-pythonthings.mp4",
"info_dict": {
"description": "Mike Pirnat presents some tips and tricks, standard libraries and third party packages that make programming in Python a richer experience.",
"title": "A Few of My Favorite [Python] Things",
},
"params": {
"skip_download": True,
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
# Extract video URL
encoded_id = self._search_regex(r"jsclassref ?= ?'([^']*)'", webpage, 'encoded id')
real_id = compat_urllib_parse.unquote(base64.b64decode(encoded_id.encode('ascii')).decode('utf-8'))
2013-06-23 21:14:19 +02:00
video_url = 'rtmpe://video.infoq.com/cfx/st/' + real_id
# Extract title
video_title = self._search_regex(r'contentTitle = "(.*?)";',
webpage, 'title')
2013-06-23 21:14:19 +02:00
# Extract description
video_description = self._html_search_regex(r'<meta name="description" content="(.*)"(?:\s*/)?>',
webpage, 'description', fatal=False)
2013-06-23 21:14:19 +02:00
video_filename = video_url.split('/')[-1]
video_id, extension = video_filename.split('.')
return {
2013-06-23 21:14:19 +02:00
'id': video_id,
'url': video_url,
'title': video_title,
2014-02-02 12:03:36 +01:00
'ext': extension, # Extension is always(?) mp4, but seems to be flv
2013-06-23 21:14:19 +02:00
'description': video_description,
}