[acast] fix channel extraction(closes #9117)
This commit is contained in:
parent
9a32e80477
commit
a1ff3cd5f9
|
@ -2,10 +2,14 @@
|
|||
from __future__ import unicode_literals
|
||||
|
||||
import re
|
||||
import functools
|
||||
|
||||
from .common import InfoExtractor
|
||||
from ..compat import compat_str
|
||||
from ..utils import int_or_none
|
||||
from ..utils import (
|
||||
int_or_none,
|
||||
OnDemandPagedList,
|
||||
)
|
||||
|
||||
|
||||
class ACastIE(InfoExtractor):
|
||||
|
@ -26,7 +30,8 @@ class ACastIE(InfoExtractor):
|
|||
|
||||
def _real_extract(self, url):
|
||||
channel, display_id = re.match(self._VALID_URL, url).groups()
|
||||
cast_data = self._download_json('https://embed.acast.com/api/acasts/%s/%s' % (channel, display_id), display_id)
|
||||
cast_data = self._download_json(
|
||||
'https://embed.acast.com/api/acasts/%s/%s' % (channel, display_id), display_id)
|
||||
return {
|
||||
'id': compat_str(cast_data['id']),
|
||||
'display_id': display_id,
|
||||
|
@ -52,15 +57,26 @@ class ACastChannelIE(InfoExtractor):
|
|||
'playlist_mincount': 20,
|
||||
}
|
||||
_API_BASE_URL = 'https://www.acast.com/api/'
|
||||
_PAGE_SIZE = 10
|
||||
|
||||
@classmethod
|
||||
def suitable(cls, url):
|
||||
return False if ACastIE.suitable(url) else super(ACastChannelIE, cls).suitable(url)
|
||||
|
||||
def _real_extract(self, url):
|
||||
display_id = self._match_id(url)
|
||||
channel_data = self._download_json(self._API_BASE_URL + 'channels/%s' % display_id, display_id)
|
||||
casts = self._download_json(self._API_BASE_URL + 'channels/%s/acasts' % display_id, display_id)
|
||||
entries = [self.url_result('https://www.acast.com/%s/%s' % (display_id, cast['url']), 'ACast') for cast in casts]
|
||||
def _fetch_page(self, channel_slug, page):
|
||||
casts = self._download_json(
|
||||
self._API_BASE_URL + 'channels/%s/acasts?page=%s' % (channel_slug, page),
|
||||
channel_slug, note='Download page %d of channel data' % page)
|
||||
for cast in casts:
|
||||
yield self.url_result(
|
||||
'https://www.acast.com/%s/%s' % (channel_slug, cast['url']),
|
||||
'ACast', cast['id'])
|
||||
|
||||
return self.playlist_result(entries, compat_str(channel_data['id']), channel_data['name'], channel_data.get('description'))
|
||||
def _real_extract(self, url):
|
||||
channel_slug = self._match_id(url)
|
||||
channel_data = self._download_json(
|
||||
self._API_BASE_URL + 'channels/%s' % channel_slug, channel_slug)
|
||||
entries = OnDemandPagedList(functools.partial(
|
||||
self._fetch_page, channel_slug), self._PAGE_SIZE)
|
||||
return self.playlist_result(entries, compat_str(
|
||||
channel_data['id']), channel_data['name'], channel_data.get('description'))
|
||||
|
|
Loading…
Reference in New Issue