[ondemandkorea] Fix extraction (closes #10772)

This commit is contained in:
Sergey M․ 2016-12-17 18:47:49 +07:00
parent 732d116aa7
commit 47c914f995
No known key found for this signature in database
GPG key ID: 2C393E0F18A9236D

View file

@ -1,14 +1,14 @@
# coding: utf-8 # coding: utf-8
from __future__ import unicode_literals from __future__ import unicode_literals
import json from .jwplatform import JWPlatformBaseIE
import re from ..utils import (
ExtractorError,
from .common import InfoExtractor js_to_json,
from ..utils import ExtractorError )
class OnDemandKoreaIE(InfoExtractor): class OnDemandKoreaIE(JWPlatformBaseIE):
_VALID_URL = r'https?://(?:www\.)?ondemandkorea\.com/(?P<id>[^/]+)\.html' _VALID_URL = r'https?://(?:www\.)?ondemandkorea\.com/(?P<id>[^/]+)\.html'
_TEST = { _TEST = {
'url': 'http://www.ondemandkorea.com/ask-us-anything-e43.html', 'url': 'http://www.ondemandkorea.com/ask-us-anything-e43.html',
@ -29,30 +29,32 @@ def _real_extract(self, url):
if not webpage: if not webpage:
# Page sometimes returns captcha page with HTTP 403 # Page sometimes returns captcha page with HTTP 403
raise ExtractorError('Unable to access page. You may have been blocked.', expected=True) raise ExtractorError(
'Unable to access page. You may have been blocked.',
expected=True)
if 'msg_block_01.png' in webpage: if 'msg_block_01.png' in webpage:
raise ExtractorError('This content is not available in your region.', expected=True) self.raise_geo_restricted(
'This content is not available in your region')
if 'This video is only available to ODK PLUS members.' in webpage: if 'This video is only available to ODK PLUS members.' in webpage:
raise ExtractorError('This video is only available to ODK PLUS members.', expected=True) raise ExtractorError(
'This video is only available to ODK PLUS members.',
expected=True)
title = self._og_search_title(webpage) title = self._og_search_title(webpage)
thumbnail = self._og_search_thumbnail(webpage)
manifest_url = self._search_regex(r'file:\s"(https?://[\S].+?/manifest\.m3u8)', webpage, 'manifest') jw_config = self._parse_json(
formats = self._extract_m3u8_formats(manifest_url, video_id, 'mp4', m3u8_id='hls') self._search_regex(
self._sort_formats(formats) r'(?s)jwplayer\(([\'"])(?:(?!\1).)+\1\)\.setup\s*\((?P<options>.+?)\);',
webpage, 'jw config', group='options'),
video_id, transform_source=js_to_json)
info = self._parse_jwplayer_data(
jw_config, video_id, require_title=False, m3u8_id='hls',
base_url=url)
subs = re.findall(r'file:\s\'(?P<file>[^\']+\.vtt)\',\s+label:\s+\'(?P<lang>[^\']+)\'', webpage) info.update({
subtitles = {}
for sub in subs:
subtitles[sub[1]] = [{'url': 'http://www.ondemandkorea.com' + sub[0], 'ext': sub[0][-3:]}]
return {
'id': video_id,
'title': title, 'title': title,
'thumbnail': thumbnail, 'thumbnail': self._og_search_thumbnail(webpage),
'formats': formats, })
'subtitles': subtitles, return info
}