[bilibili] Simplify

This commit is contained in:
Yen Chi Hsuan 2016-09-05 13:53:58 +08:00
parent 95843da529
commit 1f85029d82
No known key found for this signature in database
GPG key ID: 3FDDD575826C5C30

View file

@ -10,15 +10,12 @@
int_or_none, int_or_none,
float_or_none, float_or_none,
unified_timestamp, unified_timestamp,
urlencode_postdata,
) )
HEADERS = {
'Content-Type': 'application/x-www-form-urlencoded; charset=UTF-8',
}
class BiliBiliIE(InfoExtractor): class BiliBiliIE(InfoExtractor):
_VALID_URL = r'https?://(www.|bangumi.|)bilibili\.(?:tv|com)/(video/av|anime/v/)(?P<id>\d+)' _VALID_URL = r'https?://(?:www\.|bangumi\.|)bilibili\.(?:tv|com)/(?:video/av|anime/v/)(?P<id>\d+)'
_TESTS = [{ _TESTS = [{
'url': 'http://www.bilibili.tv/video/av1074402/', 'url': 'http://www.bilibili.tv/video/av1074402/',
@ -89,7 +86,7 @@ class BiliBiliIE(InfoExtractor):
'ext': 'mp4', 'ext': 'mp4',
'duration': 1402.357, 'duration': 1402.357,
'title': '混沌武士 : 第7集 四面楚歌 A Risky Racket', 'title': '混沌武士 : 第7集 四面楚歌 A Risky Racket',
'description': "故事发生在日本的江户时代。风是一个小酒馆的打工女。一日酒馆里来了一群恶霸虽然他们的举动令风十分不满但是毕竟风只是一届女流无法对他们采取什么行动只能在心里嘟哝。这时酒家里又进来了个“不良份子”无幻说以50个丸子帮她搞定这群人风觉得他莫名其妙也就没多搭理他。而在这时风因为一个意外而将茶水泼在了恶霸头领——龙次郎身上。愤怒的恶霸们欲将风的手指砍掉风在无奈中大喊道“丸子100个”……   另一方面,龙次郎的父亲也就是当地的代官,依仗自己有着雄厚的保镖实力,在当地欺压穷人,当看到一穷人无法交齐足够的钱过桥时,欲下令将其杀死,武士仁看不惯这一幕,于是走上前,与代官的保镖交手了……   酒馆内因为风答应给无幻100个团子无幻将恶霸们打败了就在这时仁进来了。好战的无幻立刻向仁发了战书最后两败俱伤被代官抓入牢房预计第二天斩首……   得知该状况的风,为报救命之恩,来到了刑场,利用烟花救出了无幻和仁。而风则以救命恩人的身份,命令二人和她一起去寻找带着向日葵香味的武士……(by百科)", 'description': 'md5:6a9622b911565794c11f25f81d6a97d2',
'thumbnail': 're:^http?://.+\.jpg', 'thumbnail': 're:^http?://.+\.jpg',
}, },
}] }]
@ -101,17 +98,16 @@ def _real_extract(self, url):
video_id = self._match_id(url) video_id = self._match_id(url)
webpage = self._download_webpage(url, video_id) webpage = self._download_webpage(url, video_id)
_is_episode = 'anime/v' in url if 'anime/v' not in url:
if not _is_episode:
cid = compat_parse_qs(self._search_regex( cid = compat_parse_qs(self._search_regex(
[r'EmbedPlayer\([^)]+,\s*"([^"]+)"\)', [r'EmbedPlayer\([^)]+,\s*"([^"]+)"\)',
r'<iframe[^>]+src="https://secure\.bilibili\.com/secure,([^"]+)"'], r'<iframe[^>]+src="https://secure\.bilibili\.com/secure,([^"]+)"'],
webpage, 'player parameters'))['cid'][0] webpage, 'player parameters'))['cid'][0]
else: else:
url_t = 'http://bangumi.bilibili.com/web_api/get_source' js = self._download_json(
js = self._download_json(url_t, video_id, 'http://bangumi.bilibili.com/web_api/get_source', video_id,
data='episode_id=%s' % video_id, data=urlencode_postdata({'episode_id': video_id}),
headers=HEADERS) headers={'Content-Type': 'application/x-www-form-urlencoded; charset=UTF-8'})
cid = js['result']['cid'] cid = js['result']['cid']
payload = 'appkey=%s&cid=%s&otype=json&quality=2&type=mp4' % (self._APP_KEY, cid) payload = 'appkey=%s&cid=%s&otype=json&quality=2&type=mp4' % (self._APP_KEY, cid)
@ -147,10 +143,7 @@ def _real_extract(self, url):
description = self._html_search_meta('description', webpage) description = self._html_search_meta('description', webpage)
timestamp = unified_timestamp(self._html_search_regex( timestamp = unified_timestamp(self._html_search_regex(
r'<time[^>]+datetime="([^"]+)"', webpage, 'upload time', fatal=False)) r'<time[^>]+datetime="([^"]+)"', webpage, 'upload time', fatal=False))
if _is_episode: thumbnail = self._html_search_meta(['og:image', 'thumbnailUrl'], webpage)
thumbnail = self._html_search_meta('og:image', webpage)
else:
thumbnail = self._html_search_meta('thumbnailUrl', webpage)
# TODO 'view_count' requires deobfuscating Javascript # TODO 'view_count' requires deobfuscating Javascript
info = { info = {