yt-dlp/yt_dlp/extractor/bpb.py

57 lines
2.0 KiB
Python
Raw Normal View History

2016-01-31 22:00:09 +01:00
import re
2014-11-24 20:02:00 +01:00
from .common import InfoExtractor
2016-01-31 22:00:09 +01:00
from ..utils import (
js_to_json,
determine_ext,
)
2014-11-24 20:02:00 +01:00
2014-11-24 22:47:23 +01:00
2014-11-24 20:02:00 +01:00
class BpbIE(InfoExtractor):
2014-11-24 22:47:23 +01:00
IE_DESC = 'Bundeszentrale für politische Bildung'
2016-09-08 13:29:05 +02:00
_VALID_URL = r'https?://(?:www\.)?bpb\.de/mediathek/(?P<id>[0-9]+)/'
2014-11-24 22:47:23 +01:00
_TEST = {
'url': 'http://www.bpb.de/mediathek/297/joachim-gauck-zu-1989-und-die-erinnerung-an-die-ddr',
2016-01-31 22:00:09 +01:00
'md5': 'c4f84c8a8044ca9ff68bb8441d300b3f',
2014-11-24 22:47:23 +01:00
'info_dict': {
'id': '297',
'ext': 'mp4',
'title': 'Joachim Gauck zu 1989 und die Erinnerung an die DDR',
'description': 'Joachim Gauck, erster Beauftragter für die Stasi-Unterlagen, spricht auf dem Geschichtsforum über die friedliche Revolution 1989 und eine "gewisse Traurigkeit" im Umgang mit der DDR-Vergangenheit.'
}
}
def _real_extract(self, url):
video_id = self._match_id(url)
webpage = self._download_webpage(url, video_id)
title = self._html_search_regex(
r'<h2 class="white">(.*?)</h2>', webpage, 'title')
2016-01-31 22:00:09 +01:00
video_info_dicts = re.findall(
2017-09-03 11:38:43 +02:00
r"({\s*src\s*:\s*'https?://film\.bpb\.de/[^}]+})", webpage)
2016-01-31 22:00:09 +01:00
formats = []
for video_info in video_info_dicts:
2017-09-03 11:38:43 +02:00
video_info = self._parse_json(
video_info, video_id, transform_source=js_to_json, fatal=False)
if not video_info:
continue
video_url = video_info.get('src')
if not video_url:
continue
quality = 'high' if '_high' in video_url else 'low'
2016-01-31 22:00:09 +01:00
formats.append({
'url': video_url,
'quality': 10 if quality == 'high' else 0,
2016-01-31 22:00:09 +01:00
'format_note': quality,
'format_id': '%s-%s' % (quality, determine_ext(video_url)),
})
2014-11-24 22:47:23 +01:00
return {
'id': video_id,
2016-01-31 22:00:09 +01:00
'formats': formats,
2014-11-24 22:47:23 +01:00
'title': title,
'description': self._og_search_description(webpage),
}