mirror of
https://github.com/ytdl-org/youtube-dl.git
synced 2024-01-07 17:16:08 +00:00
[9c9media] Add new extractor
This commit is contained in:
parent
66a42309fa
commit
05a0068a76
|
@ -526,6 +526,7 @@ from .nick import (
|
|||
NickDeIE,
|
||||
)
|
||||
from .niconico import NiconicoIE, NiconicoPlaylistIE
|
||||
from .ninecninemedia import NineCNineMediaIE
|
||||
from .ninegag import NineGagIE
|
||||
from .noco import NocoIE
|
||||
from .normalboots import NormalbootsIE
|
||||
|
|
55
youtube_dl/extractor/ninecninemedia.py
Normal file
55
youtube_dl/extractor/ninecninemedia.py
Normal file
|
@ -0,0 +1,55 @@
|
|||
# coding: utf-8
|
||||
from __future__ import unicode_literals
|
||||
|
||||
import re
|
||||
|
||||
from .common import InfoExtractor
|
||||
from ..utils import (
|
||||
parse_iso8601,
|
||||
parse_duration,
|
||||
ExtractorError
|
||||
)
|
||||
|
||||
|
||||
class NineCNineMediaIE(InfoExtractor):
|
||||
_VALID_URL = r'9c9media:(?P<destination_code>[^:]+):(?P<id>\d+)'
|
||||
|
||||
def _real_extract(self, url):
|
||||
destination_code, video_id = re.match(self._VALID_URL, url).groups()
|
||||
api_base_url = 'http://capi.9c9media.com/destinations/%s/platforms/desktop/contents/%s/' % (destination_code, video_id)
|
||||
content = self._download_json(api_base_url, video_id, query={
|
||||
'$include': '[contentpackages]',
|
||||
})
|
||||
title = content['Name']
|
||||
if len(content['ContentPackages']) > 1:
|
||||
raise ExtractorError('multiple content packages')
|
||||
content_package = content['ContentPackages'][0]
|
||||
stacks_base_url = api_base_url + 'contentpackages/%s/stacks/' % content_package['Id']
|
||||
stacks = self._download_json(stacks_base_url, video_id)['Items']
|
||||
if len(stacks) > 1:
|
||||
raise ExtractorError('multiple stacks')
|
||||
stack = stacks[0]
|
||||
stack_base_url = '%s%s/manifest.' % (stacks_base_url, stack['Id'])
|
||||
formats = []
|
||||
formats.extend(self._extract_m3u8_formats(
|
||||
stack_base_url + 'm3u8', video_id, 'mp4',
|
||||
'm3u8_native', m3u8_id='hls', fatal=False))
|
||||
formats.extend(self._extract_f4m_formats(
|
||||
stack_base_url + 'f4m', video_id,
|
||||
f4m_id='hds', fatal=False))
|
||||
mp4_url = self._download_webpage(stack_base_url + 'pd', video_id, fatal=False)
|
||||
if mp4_url:
|
||||
formats.append({
|
||||
'url': mp4_url,
|
||||
'format_id': 'mp4',
|
||||
})
|
||||
self._sort_formats(formats)
|
||||
|
||||
return {
|
||||
'id': video_id,
|
||||
'title': title,
|
||||
'description': content.get('Desc') or content.get('ShortDesc'),
|
||||
'timestamp': parse_iso8601(content.get('BroadcastDateTime')),
|
||||
'duration': parse_duration(content.get('BroadcastTime')),
|
||||
'formats': formats,
|
||||
}
|
Loading…
Reference in a new issue