[nbcnews] Simplify

This commit is contained in:
Sergey M․ 2015-02-14 17:42:12 +06:00
parent d9aa2b784d
commit a4f3d779db

View File

@ -52,9 +52,9 @@ def _real_extract(self, url):
class NBCNewsIE(InfoExtractor): class NBCNewsIE(InfoExtractor):
_VALID_URL = r'''(?x)https?://www\.nbcnews\.com/ _VALID_URL = r'''(?x)https?://(?:www\.)?nbcnews\.com/
((video/.+?/(?P<id>\d+))| (?:video/.+?/(?P<id>\d+)|
((?P<program>feature|nightly-news)/[^/]+/(?P<title>.+))) (?:feature|nightly-news)/[^/]+/(?P<title>.+))
''' '''
_TESTS = [ _TESTS = [
@ -120,17 +120,10 @@ def _real_extract(self, url):
# "feature" and "nightly-news" pages use theplatform.com # "feature" and "nightly-news" pages use theplatform.com
title = mobj.group('title') title = mobj.group('title')
webpage = self._download_webpage(url, title) webpage = self._download_webpage(url, title)
program = mobj.group('program') bootstrap_json = self._search_regex(
if program == 'feature': r'var\s+(?:bootstrapJson|playlistData)\s*=\s*({.+});?\s*$',
bootstrap_json = self._search_regex( webpage, 'bootstrap json', flags=re.MULTILINE)
r'var bootstrapJson = ({.+})\s*$', webpage, 'bootstrap json', bootstrap = self._parse_json(bootstrap_json, video_id)
flags=re.MULTILINE)
else:
# nightly-news
bootstrap_json = self._search_regex(
r'var playlistData = ({.+});\s*$', webpage, 'playlist data',
flags=re.MULTILINE)
bootstrap = json.loads(bootstrap_json)
info = bootstrap['results'][0]['video'] info = bootstrap['results'][0]['video']
mpxid = info['mpxId'] mpxid = info['mpxId']