mirror of
https://github.com/yt-dlp/yt-dlp.git
synced 2024-11-27 02:42:30 +00:00
[dbtv] Add new extractor
This commit is contained in:
parent
ee90ddab94
commit
f063a04f07
@ -62,6 +62,7 @@
|
||||
DailymotionUserIE,
|
||||
)
|
||||
from .daum import DaumIE
|
||||
from .dbtv import DBTVIE
|
||||
from .dotsub import DotsubIE
|
||||
from .dreisat import DreiSatIE
|
||||
from .drtv import DRTVIE
|
||||
|
76
youtube_dl/extractor/dbtv.py
Normal file
76
youtube_dl/extractor/dbtv.py
Normal file
@ -0,0 +1,76 @@
|
||||
# coding: utf-8
|
||||
from __future__ import unicode_literals
|
||||
|
||||
import re
|
||||
|
||||
from .common import InfoExtractor
|
||||
|
||||
from ..utils import (
|
||||
ExtractorError
|
||||
)
|
||||
|
||||
class DBTVIE(InfoExtractor):
|
||||
_VALID_URL = r'http://dbtv.no/(?P<id>[0-9]+)/?(?P<slug>.*)$'
|
||||
_TEST = {
|
||||
'url': 'http://dbtv.no/3649835190001#Skulle_teste_ut_fornøyelsespark,_men_kollegaen_var_bare_opptatt_av_bikinikroppen',
|
||||
'md5': 'b89953ed25dacb6edb3ef6c6f430f8bc',
|
||||
'info_dict': {
|
||||
'id': '3649835190001',
|
||||
'ext': 'mp4',
|
||||
'title': 'Skulle teste ut fornøyelsespark, men kollegaen var bare opptatt av bikinikroppen',
|
||||
'description': 'md5:d681bf2bb7dd3503892cedb9c2d0e6f2',
|
||||
'thumbnail': 'http://gfx.dbtv.no/thumbs/still/33100.jpg',
|
||||
'timestamp': 1404039863,
|
||||
'upload_date': '20140629',
|
||||
'duration': 69544,
|
||||
}
|
||||
}
|
||||
|
||||
def _real_extract(self, url):
|
||||
mobj = re.match(self._VALID_URL, url)
|
||||
video_id = mobj.group('id')
|
||||
|
||||
# Download JSON file containing video info.
|
||||
data = self._download_json('http://api.dbtv.no/discovery/%s' % video_id, video_id, 'Downloading media JSON')
|
||||
# We only want the first video in the JSON API file.
|
||||
video = data['playlist'][0]
|
||||
|
||||
# Check for full HD video, else use the standard video URL
|
||||
for i in range(0, len(video['renditions'])):
|
||||
if int(video['renditions'][i]['width']) == 1280:
|
||||
video_url = video['renditions'][i]['URL']
|
||||
break
|
||||
else:
|
||||
video_url = video['URL']
|
||||
|
||||
# Add access token to image or it will fail.
|
||||
thumbnail = video['splash']
|
||||
|
||||
# Duration int.
|
||||
duration = int(video['length'])
|
||||
|
||||
# Timestamp is given in milliseconds.
|
||||
timestamp = float(str(video['publishedAt'])[0:-3])
|
||||
|
||||
formats = []
|
||||
|
||||
# Video URL.
|
||||
if video['URL'] is not None:
|
||||
formats.append({
|
||||
'url': video_url,
|
||||
'format_id': 'mp4',
|
||||
'ext': 'mp4'
|
||||
})
|
||||
else:
|
||||
raise ExtractorError('No download URL found for video: %s.' % video_id, expected=True)
|
||||
|
||||
return {
|
||||
'id': video_id,
|
||||
'title': video['title'],
|
||||
'description': video['desc'],
|
||||
'thumbnail': thumbnail,
|
||||
'timestamp': timestamp,
|
||||
'duration': duration,
|
||||
'view_count': video['views'],
|
||||
'formats': formats,
|
||||
}
|
Loading…
Reference in New Issue
Block a user