yt-dlp/youtube_dl/extractor/pyvideo.py

60 lines
1.9 KiB
Python
Raw Normal View History

2014-03-31 08:31:48 -04:00
from __future__ import unicode_literals
2013-12-07 00:11:01 -05:00
import re
import os
2013-12-07 00:11:01 -05:00
from .common import InfoExtractor
class PyvideoIE(InfoExtractor):
_VALID_URL = r'https?://(?:www\.)?pyvideo\.org/video/(?P<id>\d+)/(.*)'
2014-03-31 08:31:48 -04:00
_TESTS = [
{
'url': 'http://pyvideo.org/video/1737/become-a-logging-expert-in-30-minutes',
'md5': '520915673e53a5c5d487c36e0c4d85b5',
2014-03-31 08:31:48 -04:00
'info_dict': {
'id': '24_4WWkSmNo',
'ext': 'webm',
2014-03-31 08:31:48 -04:00
'title': 'Become a logging expert in 30 minutes',
'description': 'md5:9665350d466c67fb5b1598de379021f7',
'upload_date': '20130320',
'uploader': 'Next Day Video',
2014-03-31 08:31:48 -04:00
'uploader_id': 'NextDayVideo',
},
'add_ie': ['Youtube'],
2013-12-07 04:59:18 -05:00
},
2014-03-31 08:31:48 -04:00
{
'url': 'http://pyvideo.org/video/2542/gloriajw-spotifywitherikbernhardsson182m4v',
'md5': '5fe1c7e0a8aa5570330784c847ff6d12',
'info_dict': {
'id': '2542',
'ext': 'm4v',
'title': 'Gloriajw-SpotifyWithErikBernhardsson182',
},
},
]
2013-12-07 00:11:01 -05:00
def _real_extract(self, url):
mobj = re.match(self._VALID_URL, url)
video_id = mobj.group('id')
2014-03-31 08:31:48 -04:00
2013-12-07 00:11:01 -05:00
webpage = self._download_webpage(url, video_id)
2014-03-31 08:31:48 -04:00
m_youtube = re.search(r'(https?://www\.youtube\.com/watch\?v=.*)', webpage)
2013-12-07 00:11:01 -05:00
if m_youtube is not None:
return self.url_result(m_youtube.group(1), 'Youtube')
2014-03-31 08:31:48 -04:00
title = self._html_search_regex(
2014-07-13 13:38:10 -04:00
r'<div class="section">\s*<h3(?:\s+class="[^"]*"[^>]*)?>([^>]+?)</h3>',
2014-04-10 20:20:50 -04:00
webpage, 'title', flags=re.DOTALL)
2014-03-31 08:31:48 -04:00
video_url = self._search_regex(
[r'<source src="(.*?)"', r'<dt>Download</dt>.*?<a href="(.+?)"'],
webpage, 'video url', flags=re.DOTALL)
return {
'id': video_id,
'title': os.path.splitext(title)[0],
'url': video_url,
}