youtube-dl/youtube_dl/extractor/teletask.py

54 lines
1.7 KiB
Python
Raw Normal View History

2014-12-21 03:01:28 -07:00
from __future__ import unicode_literals
2014-12-25 10:26:57 -07:00
2014-12-21 04:26:47 -07:00
import re
2014-12-21 03:01:28 -07:00
from .common import InfoExtractor
2014-12-25 10:26:57 -07:00
from ..utils import unified_strdate
2014-12-21 03:01:28 -07:00
class TeleTaskIE(InfoExtractor):
2014-12-25 10:26:57 -07:00
_VALID_URL = r'https?://(?:www\.)?tele-task\.de/archive/video/html5/(?P<id>[0-9]+)'
2014-12-21 03:01:28 -07:00
_TEST = {
2014-12-21 04:26:47 -07:00
'url': 'http://www.tele-task.de/archive/video/html5/26168/',
2014-12-21 03:01:28 -07:00
'info_dict': {
'title': 'Duplicate Detection',
2014-12-21 04:26:47 -07:00
},
'playlist': [{
'md5': '290ef69fb2792e481169c3958dbfbd57',
'info_dict': {
2014-12-25 10:26:57 -07:00
'id': '26168-speaker',
'ext': 'mp4',
2014-12-21 04:26:47 -07:00
'title': 'Duplicate Detection',
'upload_date': '20141218',
}
2014-12-25 10:26:57 -07:00
}, {
2014-12-21 04:26:47 -07:00
'md5': 'e1e7218c5f0e4790015a437fcf6c71b4',
'info_dict': {
2014-12-25 10:26:57 -07:00
'id': '26168-slides',
'ext': 'mp4',
2014-12-21 04:26:47 -07:00
'title': 'Duplicate Detection',
'upload_date': '20141218',
}
}]
2014-12-21 03:01:28 -07:00
}
def _real_extract(self, url):
2014-12-21 04:26:47 -07:00
lecture_id = self._match_id(url)
2014-12-25 10:26:57 -07:00
2014-12-21 04:26:47 -07:00
webpage = self._download_webpage(url, lecture_id)
2014-12-21 03:01:28 -07:00
title = self._html_search_regex(
2014-12-25 10:26:57 -07:00
r'itemprop="name">([^<]+)</a>', webpage, 'title')
upload_date = unified_strdate(self._html_search_regex(
r'Date:</td><td>([^<]+)</td>', webpage, 'date', fatal=False))
2014-12-21 03:01:28 -07:00
2014-12-21 04:26:47 -07:00
entries = [{
2014-12-25 10:26:57 -07:00
'id': '%s-%s' % (lecture_id, format_id),
'url': video_url,
2014-12-21 04:26:47 -07:00
'title': title,
2014-12-25 10:26:57 -07:00
'upload_date': upload_date,
} for format_id, video_url in re.findall(
r'<video class="([^"]+)"[^>]*>\s*<source src="([^"]+)"', webpage)]
2014-12-21 03:01:28 -07:00
2014-12-25 10:26:57 -07:00
return self.playlist_result(entries, lecture_id, title)