1
mirror of https://github.com/yt-dlp/yt-dlp synced 2024-12-11 22:23:56 +01:00
yt-dlp/yt_dlp/extractor/techtalks.py
Pccode66 7a5c1cfe93
Completely change project name to yt-dlp (#85)
* All modules and binary names are changed
* All documentation references changed
* yt-dlp no longer loads youtube-dlc config files
* All URLs changed to point to organization account

Co-authored-by: Pccode66
Co-authored-by: pukkandan
2021-02-25 00:15:56 +05:30

83 lines
2.5 KiB
Python

from __future__ import unicode_literals
import re
from .common import InfoExtractor
from ..utils import (
get_element_by_attribute,
clean_html,
)
class TechTalksIE(InfoExtractor):
_VALID_URL = r'https?://techtalks\.tv/talks/(?:[^/]+/)?(?P<id>\d+)'
_TESTS = [{
'url': 'http://techtalks.tv/talks/learning-topic-models-going-beyond-svd/57758/',
'info_dict': {
'id': '57758',
'title': 'Learning Topic Models --- Going beyond SVD',
},
'playlist': [
{
'info_dict': {
'id': '57758',
'ext': 'flv',
'title': 'Learning Topic Models --- Going beyond SVD',
},
},
{
'info_dict': {
'id': '57758-slides',
'ext': 'flv',
'title': 'Learning Topic Models --- Going beyond SVD',
},
},
],
'params': {
# rtmp download
'skip_download': True,
},
}, {
'url': 'http://techtalks.tv/talks/57758',
'only_matching': True,
}]
def _real_extract(self, url):
mobj = re.match(self._VALID_URL, url)
talk_id = mobj.group('id')
webpage = self._download_webpage(url, talk_id)
rtmp_url = self._search_regex(
r'netConnectionUrl: \'(.*?)\'', webpage, 'rtmp url')
play_path = self._search_regex(
r'href=\'(.*?)\' [^>]*id="flowplayer_presenter"',
webpage, 'presenter play path')
title = clean_html(get_element_by_attribute('class', 'title', webpage))
video_info = {
'id': talk_id,
'title': title,
'url': rtmp_url,
'play_path': play_path,
'ext': 'flv',
}
m_slides = re.search(r'<a class="slides" href=\'(.*?)\'', webpage)
if m_slides is None:
return video_info
else:
return {
'_type': 'playlist',
'id': talk_id,
'title': title,
'entries': [
video_info,
# The slides video
{
'id': talk_id + '-slides',
'title': title,
'url': rtmp_url,
'play_path': m_slides.group(1),
'ext': 'flv',
},
],
}