[teletask] Add new extractor
This commit is contained in:
parent
3f7421b71b
commit
33b53b6021
1 changed files with 43 additions and 19 deletions
|
@ -1,6 +1,5 @@
|
|||
# coding: utf-8
|
||||
from __future__ import unicode_literals
|
||||
|
||||
import re
|
||||
import datetime
|
||||
|
||||
|
@ -11,33 +10,58 @@ class TeleTaskIE(InfoExtractor):
|
|||
_VALID_URL = r'http?://(?:www\.)?tele-task\.de/archive/video/html5/(?P<id>[0-9]+)/'
|
||||
_TEST = {
|
||||
'url': 'http://www.tele-task.de/archive/video/html5/26168/',
|
||||
'md5': '290ef69fb2792e481169c3958dbfbd57',
|
||||
'info_dict': {
|
||||
'id': '26168',
|
||||
'ext': 'mp4',
|
||||
'title': 'Duplicate Detection',
|
||||
'upload_date': '20141218',
|
||||
}
|
||||
},
|
||||
'playlist': [{
|
||||
'md5': '290ef69fb2792e481169c3958dbfbd57',
|
||||
'info_dict': {
|
||||
'title': 'Duplicate Detection',
|
||||
'upload_date': '20141218',
|
||||
'id': 'speaker_26168',
|
||||
'ext': 'mp4',
|
||||
}
|
||||
},
|
||||
{
|
||||
'md5': 'e1e7218c5f0e4790015a437fcf6c71b4',
|
||||
'info_dict': {
|
||||
'title': 'Duplicate Detection',
|
||||
'upload_date': '20141218',
|
||||
'id': 'slides_26168',
|
||||
'ext': 'mp4',
|
||||
}
|
||||
}]
|
||||
}
|
||||
|
||||
def _real_extract(self, url):
|
||||
video_id = self._match_id(url)
|
||||
webpage = self._download_webpage(url, video_id)
|
||||
lecture_id = self._match_id(url)
|
||||
webpage = self._download_webpage(url, lecture_id)
|
||||
|
||||
title = self._html_search_regex(
|
||||
r'itemprop="name">([^"]+)</a>', webpage, 'title')
|
||||
url = self._html_search_regex(
|
||||
r'class="speaker".*?src="([^"]+)"', webpage, 'video_url', flags=re.DOTALL)
|
||||
|
||||
url_speaker = self._html_search_regex(
|
||||
r'class="speaker".*?src="([^"]+)"', webpage, 'video_url_speaker', flags=re.DOTALL)
|
||||
url_slides = self._html_search_regex(
|
||||
r'class="slides".*?src="([^"]+)"', webpage, 'video_url_slides', flags=re.DOTALL)
|
||||
date = self._html_search_regex(
|
||||
r'<td class="label">Date:</td><td>([^"]+)</td>', webpage, 'date')
|
||||
date = datetime.datetime.strptime(date, '%d.%m.%Y').strftime('%Y%m%d')
|
||||
|
||||
|
||||
entries = [{
|
||||
'title': title,
|
||||
'upload_date': date,
|
||||
'id': "speaker_"+lecture_id,
|
||||
'url': url_speaker,
|
||||
},
|
||||
{
|
||||
'title': title,
|
||||
'upload_date': date,
|
||||
'id': "slides_"+lecture_id,
|
||||
'url': url_slides}]
|
||||
|
||||
return {
|
||||
'id': video_id,
|
||||
'_type': "playlist",
|
||||
'id': lecture_id,
|
||||
'title': title,
|
||||
'url': url,
|
||||
'upload_date': date,
|
||||
'entries': entries,
|
||||
}
|
Loading…
Reference in a new issue